Datasets:

taln-ls2n

/

pubmed

like 2

J_Muscle_Res_Cell_Motil-3-1-2045119

Conduction velocities in amphibian skeletal muscle fibres exposed to hyperosmotic extracellular solutions

Early quantitative analyses of conduction velocities in unmyelinated nerve studied in a constantly iso-osmotic volume conductor were extended to an analysis of the effects of varying extracellular osmolarities on conduction velocities of surface membrane action potentials in Rana esculenta skeletal muscle fibres. Previous papers had reported that skeletal muscle fibres exposed to a wide range of extracellular sucrose concentrations resemble perfect osmometers with increased extracellular osmolarity proportionally decreasing fibre volume and therefore diminishing fibre radius, a. However, classical electrolyte theory (Robinson and Stokes 1959, Electrolyte solutions 2nd edn. Butterworth & Co. pp 41–42) would then predict that the consequent increases in intracellular ionic strength would correspondingly decrease sarcoplasmic resistivity, Ri. An extension of the original cable analysis then demonstrated that the latter would precisely offset its expected effect of alterations in a on the fibre axial resistance, ri, and leave action potential conduction velocity constant. In contrast, other reports (Hodgkin and Nakajima J Physiol 221:105–120, 1972) had suggested that Riincreased with extracellular osmolarity, owing to alterations in cytosolic viscosity. This led to a prediction of a decreased conduction velocity. These opposing hypotheses were then tested in muscle fibres subject to just-suprathreshold stimulation at a Vaseline seal at one end and measuring action potentials and their first order derivatives, dV/dt, using 5–20 MΩ, 3 M KCl glass microelectrodes at defined distances away from the stimulus sites. Exposures to hyperosmotic, sucrose-containing, Ringer solutions then reversibly reduced both conduction velocity and maximum values of dV/dt. This was compatible with an increase in Ri in the event that conduction depended upon a discharge of membrane capacitance by propagating local circuit currents through initially passive electrical elements. Conduction velocity then showed graded decreases with increasing extracellular osmolarity from 250–750 mOsm. Action potential waveforms through these osmolarity changes remained similar, including both early surface and the late after-depolarisation events reflecting transverse tubular activation. Quantitative comparisons of reduced-χ 2 values derived from a comparison of these results and the differing predictions from the two hypotheses strongly favoured the hypothesis in which Riincreased rather than decreased with hyperosmolarity. Introduction A classic paper by Hodgkin (1954) (see also: Adrian 1975; Noble 1979) performed a quantitative cable analysis of the local circuit currents thought responsible for action potential propagation in unmyelinated nerve fibres studied in large volumes of constantly iso-osmotic extracellular fluid and demonstrated that their conduction velocities should vary as the square root of fibre radius, a. Action potential conduction velocity is also physiologically important in striated muscle: it ensures a rapid, sarcomeric activation leading to near-synchronous muscle contraction. The present paper accordingly extends these early analyses to effects of varying extracellular osmolarity on conduction velocities of surface membrane action potentials in Rana esculenta skeletal muscle. Skeletal muscle membrane differs from unmyelinated nerve in including an excitable transverse (T-) system whose activation initiates muscle contraction (Adrian et al. 1969, 1970; Huang 1993; Nielsen et al. 2004; Stephenson 2006). Nevertheless, it is likely that the initial sarcolemmal membrane activation precedes the subsequent inward tubular excitation and that the latter may take place partially independently of the propagation of the surface excitation wave. Sheikh et al. (2001) suggested a partially separable T-tubule excitation initiated by Na+ channels selectively clustered around the mouths of the T-tubular lumina: detubulation produced by osmotic shock left surface membrane conduction velocities unchanged. Furthermore, increases in extracellular osmolarity did not alter the tubular diameters at the necks of the T-tubular system important in electrical connectivity between T-tubule network and remaining extracellular space (Launikonis and Stephenson 2002, 2004). However, variations in surface conduction velocities in skeletal muscle studied under varying osmolarities would be expected to differ from those of nerve fibres studied under constant, iso-osmotic conditions owing to the consequent changes in their cell volumes and fibre diameters. Nevertheless, skeletal muscle volume exhibits close to ideal osmotic behaviour, varying inversely with changes in extracellular osmolarity that would in turn predictably increase solute concentration. The Results section of this paper accordingly first develops the original treatment (Hodgkin 1954; Adrian 1975; Noble 1979) for situations in which fibre diameter is varied by alterations in extracellular osmolarity that would in turn alter intracellular ionic strength. This provided quantitative expectations for any resulting change in conduction velocity that corresponded to two contrasting situations. First, classical electrolyte theory (Robinson and Stokes 1959; Atkins 1998) would predict that increased intracellular ionic strength should proportionally decrease sarcoplasmic resistivity Ri. Our analysis then indicated that this would precisely offset any expected changes in fibre axial resistance, ri, produced by the osmotically induced alterations in fibre diameter and thereby leave conduction velocity unchanged. Secondly, Hodgkin and Nakajima (1972) reported that Ri increased with extracellular osmolarity, possibly due to increases in myoplasmic viscosity, although we certainly do not exclude possible factors arising from the more complex membrane structures found in muscle as opposed to nerve (see e.g. Sheikh et al. 2001). This led to predictions that conduction velocity would decrease with increasing extracellular osmolarity. These predictions were then investigated by experimental determinations of conduction velocity obtained from action potential records derived from microelectrode measurements made at known distances from defined stimulation sites in surface muscle fibres of Rana esculenta studied at a fixed (∼7°C) temperature at varied extracellular osmolarities. These findings demonstrated reversible changes in the conduction velocities of action potentials the nature of whose waveforms, including early surface and later tubular components, were otherwise unchanged, and distinguished between hypotheses through the observed dependence of conduction velocity on extracellular osmolarity. Materials and methods Cold-adapted frogs, Rana esculenta, were killed by concussion followed by pithing (Schedule I: Animal Procedures Act, Home Office, UK). The skin was removed and the tendon of insertion of the sartorius into the patella ligated, cut distally and dissected along its borders up to and including its origin at the pelvic girdle and acetabulum. This was performed at room temperature (21.7 ± 0.51°C; mean ± standard error of the mean; n = 15) in standard Ringer solution consisting of (mM): 115 NaCl, 2.5 KCl, 1.8 CaCl2, 3 Hepes, titrated to pH 7, osmolarity 250 mOsm. The muscle was then stretched to 1.4 times its in situ length and secured in a transilluminated methacrylate polymer (Perspex) chamber by pinning the cleaned acetabulum and the ligature to minimize contractile artefact, which became less evident in solutions of higher osmolarities, and entirely absent with the higher (>350−400 mM) sucrose concentrations. The isolated sartorius muscles had an in situ length of 30.6 ± 0.94 mm and a length of 41.6 ± 1.15 mm (n = 15) when stretched. The muscle was laid over a supporting ramp with its dorsal surface uppermost. A watertight Perspex partition, coated with a layer of Vaseline, with the muscle running through a notch at its lower surface electrically isolated the recording chamber into two compartments. The only path conducting the brief voltage stimuli applied from two platinum electrodes at opposite ends of the chamber was therefore through the muscle across the partition. The platinum electrode on the side of the chamber containing the acetabulum was the cathode. The temperature of the bath solution was lowered by circulating distilled water cooled in a magnetically stirred ice bath. The water was circulated through a glass coil placed in the chamber in close proximity to the muscle, using a Minipuls 3 peristaltic pump (Gilson, France). A digital thermometer incorporating a remote thermistor (P. Frost, Department of Physiology, Development and Neuroscience, Cambridge), previously linearised and calibrated against a platinum film resistor was placed in the bathing solution near the muscle to allow the temperature to be continuously monitored and adjusted to remain constant throughout the recordings. The temperature was kept constant at 7.0 ± 0.03°C (n = 15) (Sheikh et al. 2001). Action potentials are temperature-sensitive so this allowed conditions to be optimised. Cooling also prolonged fibre viability especially in the hyperosmotic conditions. The muscle was studied in the range of extracellular osmolarities from 250 to 750 mOsm. Solutions of different osmolarities were made by adding to the standard Ringer solution the membrane-impermeant solute sucrose at varying (150, 250, 350, 400, 450, 500 mM respectively) concentrations to yield solutions whose increased osmolarities (mOsm) were calculated from their total solute concentrations, and checked against measurements using a standard vapour pressure osmometer. Hepes was obtained from Sigma-Aldrich (Gillingham, Dorset, UK) and all other reagents from BDH Limited (UK). Solutions were changed, avoiding contact with the muscle, between tests using a vacuum pump (Edwards, UK) after which approximately 5–10 min elapsed before recording to permit time for both temperature (7°C) and osmotic equilibration; previous reports (Ferenczi et al. 2004) have indicated that 200 s sufficed to permit complete volume changes in response to hyperosmotic solutions. 3 M KCl-filled glass capillary microelectrodes (resistances 5–20 MΩ; tip potentials <5 mV: Adrian 1956) drawn from 1.2 mm (inner diameter) glass tubing were used to obtain membrane potentials. These were mounted via Ag/AgCl junctions to a high impedance voltage follower balanced by matching Ag/AgCl junctions earthing the bath. Only surface fibres showing clear-cut penetrations and stable resting membrane potentials were studied. Rates of rise and fall, dV/dt, in the voltage traces, were obtained by electrical differentiation of the output, with voltage and dV/dt channels filtered between cutoffs of 0 Hz/5 kHz and 0 Hz/20 kHz respectively. Action potentials were elicited by direct just-suprathreshold stimulation which varied between 1.5 and 5 V, to minimise electrotonic spread of the stimulus voltage and the number of fibres in which electrical activity was initiated, across the Perspex partition, through two platinum stimulating electrodes. Conduction velocity was calculated by dividing the distance from the site of action potential initiation to the position of the recording microelectrode tip, ascertained by a Vernier scale, which gave measurements in cm to two decimal places, by the latency given by the time from a clear stimulus artefact generated to the peak of the dV/dt input, which will be referred to as the maximum dV/dt. The corresponding resting membrane potential, action potential overshoot and the value of maximum dV/dt were also noted. These relatively early events in the timecourse of the regenerative response often could be measured even in records showing small contractile artefact, which often only became evident at later times in the recorded traces. Results The experiments utilized an extension of early quantitative analyses (Hodgkin 1954; Adrian 1975) of conduction velocity in unmyelinated nerve studied in a constantly iso-osmotic volume conductor to changes that would result from varying the extracellular osmolarities in Rana esculenta skeletal muscle fibres. The initial analysis used the cable equation for the current density through any patch of flat membrane, expressed as current per unit membrane area, This describes conduction in a cable whose own finite internal volume is small compared to that of the extracellular conducting fluid in which it is studied to give large ratios between longitudinal intracellular, ri, and extracellular, ro, resistances to current flow per unit length, i.e., ro << ri, permitting extracellular terms relating to ro to be dropped. Equation (1) is combined with the expression for conduction velocity, using the chain rule of differentiation, to give:where Vm is the internal potential across the membrane at distance x along the fibre. Conduction at constant velocity through local circuit spread of excitation then requires Im to be a single valued function of Vm. The term containing the intrinsic membrane parameters consisting of the area of membrane per unit length of fibre s, and the axial cytoplasmic resistance per unit length ri, is then constant, whence , and where the constant k only depends on passive local membrane electrical properties. Thus: For unmyelinated cylindrical fibres for which the axial cytoplasmic resistance per unit length, ri, is related to fibre radius a and sarcoplasmic resistivity Ri by Ri = riπ a2, the total fibre volume, vol, = π a2L where L is the length of the fibre and for which the total surface area of the fibre, A = sL: In the case of skeletal muscle fibres exposed to solutions of different osmolarities, osm, the volume, vol, behaves as a perfect osmometer following the relationship: and reductions in fibre volume due to increased extracellular osmolarity would correspondingly increase solute concentration. Classical electrolyte theory (Robinson and Stokes 1959) then predicts that each participating ion contributes a specific conductance Ksp determined by its solute concentration ηC and a constant Λ defined for any given ion in any specified solute and referred to the conductivity of the ion at a 1 M concentration: The overall conductivity of such an ideal solution is then the algebraic sum of the conductivities of individual component ions. Decreases in fibre volume resulting from increased extracellular osmolarities then would increase intracellular ionic strength and in turn increase Ksp and proportionally decrease Ri: Equation (7) then gives: Equations (5) and (8) then give,thereby predicting a constant conduction velocity in view of the fact that the terms k′, resulting from the additional proportionalities above, and A are both constants. In contrast, Hodgkin and Nakajima (1972) suggested that sarcoplasmic resistivity, Ri, increased with increasing extracellular osmolarity, possibly owing to an increased myoplasmic viscosity, with reductions in fibre volume in hyperosmotic solutions, reporting a linear increase in Ri with increasing extracellular osmolarity:where D and E are constants. From (5) and (10): Because k′ and A are constants, we have:where D′ and E′ are also constants. This forms a contrasting expectation in which θ is expected to decrease with increasing extracellular osmolarity in a relationship modelled by Eq. (12) as a result of an increase in Ri. The alternative hypotheses could then be quantitatively tested by measurements of θ under different conditions of extracellular osmolarity. Reversible effects of hyperosmotic extracellular solutions on action potential waveforms and latencies Figure 1 shows typical action potential and dV/dt waveforms obtained at a temperature of ∼7°C before, during and after exposures to solutions with increased extracellular osmolarity with column A showing the action potential waveforms and column B their corresponding dV/dt records. Panel a shows typical results from fibres in iso-osmotic Ringer at 7°C. The muscle was then exposed to a hyperosmotic (600 mOsm) Ringer solution and recordings resumed 5–10 min following this solution change (Panel b). Finally, the muscle was then returned to iso-osmotic Ringer and recordings again resumed 5–10 min after the solution was restored (Panel c). Latencies were measured from the clear stimulus artefact to the maximum point of the dV/dt trace as seen in column B: this approach provided more consistent measurements of action potential latency than using arbitrarily chosen points on the action potential traces (see also Sheikh et al. 2001).Fig. 1Typical action potential (A) and dV/dt traces (B) obtained from muscle fibres exposed successively to (a) iso-osmotic Ringer (b) hyperosmotic (600 mOsm) Ringer and (c) returned to iso-osmotic Ringer, in the same sartorius muscle The action potential traces showed prolonged positive after-depolarisation phases lasting well beyond 20 ms following the surface action potential deflections (Fig.1Aa), consistent with persistent excitation of an intact transverse (T-) tubular system (Adrian and Peachey 1973). These persisted both in the hyperosmotic solution (Fig. 1Ab) and the iso-osmotic solution to which the fibres were finally returned (Fig. 1Ac), confirming a persistence in transverse tubular excitation and its excitation following initiation of the surface component of the action potential through these manipulations. Use of 600 mOsm as opposed to standard Ringer increased the action potential latencies and consequently the calculated values of conduction velocity. Thus, conduction velocities in the initial iso-osmotic Ringer solution, then at 600 mOsm-Ringer solution and finally, the returned iso-osmotic Ringer were 0.98 ± 0.092 m s−1 (n = 11 fibres), 0.58 ± 0.051 m s−1 (n = 10) and 0.72 ± 0.101 m s−1 (n = 6) respectively. The effect of hyperosmolarity on conduction velocity was thus at least partly reversible. However, this decrease in conduction velocity with increased osmolarity (from 250 to 600 mOsm) was not accompanied by qualitative changes in action potential or dV/dt waveform. Likewise, the waveforms showed no qualitative changes following return to the iso-osmotic Ringer solution after hyperosmotic exposure. Grading of conduction velocity changes with extracellular osmolarity Figure 2 summarises typical traces of action potentials (column A) and dV/dt (column B) for experiments that systematically investigated the effects of graded changes in extracellular osmolarity reflecting different extracellular sucrose concentrations on conduction velocity, action potential waveform, and dV/dt obtained at 7°C. Action potential waveforms again included both early rapid surface action potential deflections following the stimulus artefact and prolonged after-depolarisation phases that lasted well beyond 20 ms, reflecting T-tubular activation, at all the extracellular osmolarities studied (cf. Adrian and Peachey 1973). Measurements of latencies between the stimulus artefacts to the maximum value of the dV/dt traces then suggested a noticeable overall decrease in the calculated conduction velocities with increasing osmolarity. All these changes occurred in the absence of any excessive depolarization in resting membrane potential over the range of explored osmolarities. Thus resting potential at an osmolarity of 250 mOsm was −85.27 ± 1.05 mV (n = 87); that at an osmolarity of 700 mOsm was −75.94 ± 2.20 mV (n = 33). However, resting potential at an osmolarity of 750 mOsm was −62.71 ± 2.82 mV (n = 17) (cf. Fraser et al. 2006). At the latter extracellular osmolarity a small proportion of action potentials appeared to display markedly reduced amplitudes resulting in peak deflections that failed to show overshoots, and a firing of two consecutive action potentials after a single stimulation in the case of one fibre. Accordingly, the investigations were not performed at higher sucrose concentrations.Fig. 2Typical action potential (A) and dV/dt (B) traces recorded in the following extracellular osmolarities: (a) 250, (b) 400, (c) 500, (d) 600, (e) 650 and (f) 750 (all values given in mOsm). There is a notable overall increase in the latencies measured from (B) Plots of conduction velocity and maximum dV/dt with increasing extracellular osmolarity Figure 3 summarizes the results of systematic studies of the dependence of conduction velocity on extracellular osmolarity in a statistically larger number of muscle fibres. Action potentials and their corresponding dV/dt were measured systematically in n = 87, 18, 27, 11, 10, 33 and 13 fibres at sucrose concentrations of 0, 150, 250, 350, 400, 450 and 500 mM corresponding in turn to osmolarities of 250, 400, 500, 600, 650, 700 and 750 mOsm respectively. Figure 3 demonstrates that both conduction velocity (a) and maximum dV/dt (means ± standard errors of the mean) (b) monotonically decreased with extracellular osmolarity. Figure 3a then compares these experimental values of conduction velocity against predictions from the two hypotheses for the dependence of Ri on osmolarity. In both cases, θ was calculated from Ri and the extracellular osmolarity, osm, using the relationship derived from Eqs. (4) and (5) above that:Fig. 3Experimentally measured conduction velocities (a) and maximum dV/dt (mean ± standard error of the mean) plotted against extracellular osmolarity (◆). Dashed lines in (a): velocities predicted for a situation in which Ri decreases proportionally as a result of increased intracellular ionic strength produced by changes in cell volume brought about by the osmolarity changes. Continuous line: predictions when Ri increases with increasing extracellular osmolarity suggested by Hodgkin and Nakajima (1972). The changes in (a) were accompanied by decreases in maximum dV/dt (b) with increasing osmolarity, as expected for propagation brought about by local circuit currents The dotted line denotes expectations from the first situation outlined above in which Ri decreases proportionally with increasing extracellular osmolarity: this predicts that conduction velocity does not alter with osmolarity from its control value in iso-osmotic solutions corresponding to a constant value of 0.85 m s−1 obtained in iso-osmotic solutions. The continuous line represents the second hypothesis in which Ri depends both on temperature and extracellular osmolarity as reported by Hodgkin and Nakajima (1972). The function displayed assumes the sarcoplasmic conductivity Gi(= 1/Ri) to have a Q10 = 1.37, and a value of Ri at 2°C, of 298.51 Ω cm in a muscle fibre within an extracellular solution of 250 mOsm osmolarity and of 390.63 Ω cm in a muscle fibre within an extracellular solution of 600 mOsm osmolarity, as suggested by Hodgkin and Nakajima (1970, 1972). Then, let θ1, Ri1 and osm1 correspond to conditions in iso-osmotic Ringer and θ2, Ri2 and osm2 correspond to conditions in any hyperosmotic solution. Thus:Since θ1 = 0.85 ± 0.029 m s−1 in iso-osmotic solution,where The values generated by the above equations predicted a decline in conduction velocity with increasing extracellular osmolarity. To test the statistical significance of the goodness-of-fit of the two contrasting predictions to the experimental data obtained, an Fx test that takes into account the difference of the two reduced-χ2 values in proportion to the first χ2 term was performed (Bevington 1969). This derived for each hypothesis a value of χ2 depicting the summed squared deviation of the original data, yi, to the predicted findings, y(xi), obtained at any extracellular osmolarity, xi, such that: Values for χ2 obtained in the two cases, χ12 where the predicted conduction velocity is a constant value and χ22 where there is an increase in Ri with extracellular osmolarity were 21.360 and 15.141 respectively. These values were then used to compute a F-statistic, given by: where n, the sample size = 200 and n−1 is the number of degrees of freedom. The reduced-χ2 tests for goodness-of-fit yielded a F-statistic of 81.73 consistent with a significantly better fit (P << 0.001) to predictions from a situation in which Ri increased as described by Hodgkin and Nakajima (1972) rather than a decrease in Ri with increasing extracellular osmolarity. Discussion This paper begins from classic analyses (Hodgkin 1954; Adrian 1975) of the dependence of action potential conduction velocities upon the diameter of nerve fibres studied in large volume conductors under iso-osmotic extracellular conditions. This had employed cable analysis that attributed these propagation processes to local circuit currents driven by Na+ currents, INa, generating the rising phase of the action potential that in turn discharged initially passive circuit components (Valdiosera et al. 1974) equivalent to a membrane capacitance per unit area, Cm and resistance Rm of unit membrane area in series with an axial cytoplasmic resistance per unit length, ri. The latter is related to fibre radius a and sarcoplasmic resistivity Ri (kΩ cm) by Ri = riπa2 giving the original result that conduction velocity would vary as the square root of a. The present paper extends this analysis to the effects of varying extracellular osmolarity on conduction velocities of surface membrane action potentials in Rana esculenta skeletal muscle, a situation that differed in a number of respects. First, skeletal muscle contains excitable transverse (T-) tubular membrane system responsible for initiating excitation-contraction coupling (Adrian et al. 1969, 1970; Huang 1993; Nielsen et al. 2004; Stephenson 2006). Nevertheless, the rapid initial sarcolemmal membrane activation that ensures rapid action potential propagation producing a synchronous sarcomeric activation likely largely precedes full tubular excitation (Adrian and Peachey 1973). Recent detubulation experiments left surface membrane conduction velocities unchanged suggesting a separation of surface and T-tubule excitation, the latter possibly initiated separately by Na+ channels localized around the T-tubular luminal mouths (Sheikh et al. 2001). Increasing extracellular osmolarity did not increase the diameter of the necks of the T-tubules important in electrical connectivity between T-tubule network and the fibre membrane (Launikonis and Stephenson 2002, 2004). These findings would permit muscle fibre conduction velocities to be analysed in terms of surface cylinders. Second, skeletal muscle volume alters inversely with extracellular osmolarity (Blinks 1965; Ferenczi et al. 2004) in turn correspondingly increasing solute concentration, in contrast to the situation represented by comparisons of nerve fibres of different diameters in similarly iso-osmotic extracellular solutions. This paper then derived quantitative consequences from two possible hypotheses emerging from the above conditions. On the one hand, classical electrolyte theory (Robinson and Stokes 1959; Atkins 1998) predicts a specific conductance Ksp attributable to each intracellular ion species increasing proportionally with solute concentration following decreases in cell volume in hyperosmotic solution. This led to a prediction of a decrease in Ri precisely correcting out effects of any osmotically induced diameter change together leaving conduction velocity constant. On the other hand, empirical observations suggesting increases in Ri with extracellular osmolarity (Hodgkin and Nakajima 1972) permitted construction of a quantitative formula for the resulting variations in Ri with extracellular osmolarity as well as temperature. This led to predictions of a conduction velocity that decreased with increasing extracellular osmolarity. The experiments described in this paper then sought to distinguish between the two hypotheses. It investigated the effects of extracellular osmolarity on conduction velocities of surface membrane action potentials in surface fibres, that would be maximally exposed to these solution changes, in frog skeletal muscle, following stimulation at a Vaseline seal at defined distances from the microelectrode recording site. Simultaneous records were made of the rate of change of membrane potential dV/dt that would be provided a consistent time point at which there would be a maximal action potential slope as well as rate of discharge of the membrane capacitance, Cm, by local currents driven by the Na+ current INa. These studies showed that conduction velocity declined with increasing extracellular osmolarity along with maximum dV/dt as expected for a process dependent upon a local circuit current flow, despite relatively constant resting membrane potentials. These changes in conduction velocity in hyperosmotic solution were at least partially reversible. Furthermore, there were no changes in the nature of the action potential waveforms including early surface deflections and late after-depolarisation phases observed, consistent with a minimal change in the capacity for T-tubular excitation, through a still intact tubular system, that nevertheless followed generation of the initial surface component of the action potential wave. Further systematic study in larger fibre numbers through a range of osmolarities (250–750 mOsm) demonstrated corresponding declines in conduction velocity and maximum dV/dt. At the highest osmolarity (750 mOsm), a small proportion of fibres variously showed low action potential overshoots which however did not correlate with the situations where there was a reduced conduction velocity, as well as multiple firing in a single muscle fibre. Nevertheless, further studies were not made at these and higher osmolarities; in any case, muscle fibres are thought only to act close to perfect osmometers at osmolarities up to around four times that of the iso-osmotic solution (Blinks 1965). These findings are thus consistent with the second hypothesis in which sarcoplasmic resistivity, Ri, increases with the fibre shrinkage observed in hyperosmotic solutions, as suggested by Hodgkin and Nakajima (1972), as opposed to the first possibility outlined above in which conduction velocity should be constant. This was borne out by objective statistical analysis of the goodness-of-fit of the predictions derived from the two contrasting hypotheses as expressed in their resulting reduced-χ2 values, with the experimental values of conduction velocity generally assuming slightly higher values than predicted. The latter might reflect minor departures from a purely continuous conduction as postulated for unmyelinated cylindrical fibres, either due to contributions from peripheral regions of tubular membrane less isolated than the remaining tubular system (Hodgkin and Nakajima 1972), or the clustering of sodium channels around the tubular necks as reported by Sheikh et al. (2001) that may contribute to a more saltatory-like conduction in the muscle that would be expected to generally speed up propagation of electrical activity. Electronic supplementary material Below is the link to the electronic supplementary material. 49 kb doc

Pediatr_Nephrol-3-1-1794138

Pathophysiology of focal segmental glomerulosclerosis

Focal segmental glomerulosclerosis (FSGS) is a major cause of idiopathic steroid-resistant nephrotic syndrome (SRNS) and end-stage kidney disease (ESKD). In recent years, animal models and studies of familial forms of nephrotic syndrome helped elucidate some mechanisms of podocyte injury and disease progression in FSGS. This article reviews some of the experimental and clinical data on the pathophysiology of FSGS. Learning objectives Discuss the experimental and clinical data on the pathophysiology of FSGSReview the alterations in glomerular structure and function associated with FSGSTo identify potential mechanisms responsible for disease progression in FSGSDistinguish some targets for the future therapy of FSGS Focal segmental glomerulosclerosis (FSGS) is a disease entity defined by findings on kidney biopsy [1, 2]. FSGS is the major cause of idiopathic steroid-resistant nephrotic syndrome (SRNS) in children and adults [3]. FSGS is the most common cause of acquired chronic renal insufficiency in children and frequently leads to progression to end-stage kidney disease (ESKD) [2]. FSGS may occur secondary to such disparate disease processes as HIV and obesity [1, 4]; this review focuses on the pathophysiology of primary FSGS (i.e., with no underlying illness). Alterations of normal glomerular structure and function have been found in FSGS [5]. Normal function requires that the three major components of the glomerular filter (the endothelial cells, podocytes, and glomerular basement membrane) are intact and are able to provide a permselective filtration barrier (Fig. 1). Specialized tight junctions between podocyte foot processes create the slit diaphragm which is integral to preventing the loss of protein into the urinary space [6]. While the clinical presentation of FSGS is often heterogeneous, a characteristic feature of the disease is proteinuria, which implies the loss of this barrier [2, 7]. Indeed, electron microscopy has shown distortion of the normal architecture (or effacement) of the foot processes of podocytes in FSGS [1]. The connection between these projections of the epithelial cell and the underlying basement membrane can be disrupted, leading to the loss of nonspecific plasma proteins into the tubular filtrate [6]. Fig. 1A Components of the normal glomerular filtration barrier: (1) glomerular basement membrane (GBM); (2) podocyte foot process; (3) endothelial cell; B Progressive changes seen in focal segmental glomerulosclerosis (FSGS) leading to sclerosis: (1) foot process effacement; (2) podocyte apoptosis/loss and exposed glomerular basement membrane; (3) filtration of non-specific plasma proteins; (4) capillary expansion; (5) formation of synechiae; (6) misdirected filtration at points of synechiae; (7) mesangial matrix proliferation. Adapted from Kwoh et al. [9] However, unlike other causes of proteinuria and nephrotic syndrome, such as minimal change disease (MCD), FSGS often progresses to ESKD. While foot process effacement is seen in MCD as well as FSGS, histologically, FSGS is characterized by increased extracellular matrix within the glomerular tuft with obliteration of the glomerular capillary lumen. These sclerotic lesions occur focally and in only some segments of glomeruli, and are typically not associated with immune complex deposition [1]. The location of sclerotic lesions by light microscopy defines the variants of FSGS: perihilar variant (with sclerosis of the vascular pole), cellular variant (associated with hypercellularity of the capillary space), tip variant (involving the part of the glomerulus near the origin of the proximal tubule), and collapsing variant (with one or more glomeruli with global or segmental collapse) [1]. Clinically, the variants of FSGS differ; for example, the collapsing variant tends to progress more rapidly to ESKD and commonly occurs in the setting of HIV [1]. It is possible that different mechanisms may play a role in the pathogenesis of each variant of FSGS [7, 8]. Insight into the pathogenesis of FSGS developed over the past decade from studies of genetic mutations in mice, models of progressive glomerulosclerosis (such as the rat remnant kidney model), and identification of gene mutations in some familial forms of nephrotic syndrome (including congenital nephrotic syndrome and familial and autosomal dominant FSGS) [7, 9, 10]. Key in the pathogenesis of FSGS is podocyte damage and loss [5, 6]. Injury to podocytes occurs by four major mechanisms: alteration of the components of the slit diaphragm or interference with its structure, dysregulation of the actin cytoskeleton, alteration of the glomerular basement membrane or its interactions with the podocyte, or alteration of the negative surface charge of the podocyte [6, 9]. Damage to podocytes triggers apoptosis and their detachment of podocytes from the glomerular basement membrane [6, 9]. The ensuing reduction in podocyte number is felt to play an important role in the pathogenesis of FSGS [7]. The podocyte is normally a terminally differentiated cell with limited proliferative capacity in response to injury [7]. The initial insult to the podocyte leads to further damage mediated by cytokine release, mechanical stress, and further loss of polarity, resulting in sclerosis and scarring of the glomerulus [7, 9]. Genetic mutations seen in congenital forms of nephrotic syndrome and FSGS enabled researchers to identify specific gene mutations involved in podocyte damage [10]. Mutations of the nephrin gene, a podocyte-specific transmembrane component of the slit diaphragm, are found in congenital Finnish-type nephrotic syndrome, and may lead to loss of normal caliber slit diaphragms [6, 9–11]. In mouse models, mutations of nephrin-like transmembrane genes (NEPH-1) which also localize to the slit diaphragm result in proteinuria and early death [6, 10]. It is unclear how alteration of the slit diaphragm results in podocyte loss. The slit diaphragm may be integral to maintaining cell polarity or its damage may alter the balance of cell signals, resulting in apoptosis. Mutations in a Fyn kinase (one of the src tyrosine kinases) that phosphorylates nephrin and may regulate cell cycle and apoptosis resulted in proteinuria and foot process effacement in a mouse model [9, 10]. Other proteins which are part of the slit diaphragm complex include: podocin, CD2-associated protein (CD2AP), FAT, ZO-1, P-cadherin, an LAP (leucine rich repeat and PDZ domain) protein, and MAGI-1 [6, 10]. Mutations in podocin (a transmembrane protein that interacts with nephrin, NEPH-1 and CD2AP) have been identified in familial FSGS [9, 10, 12]. Recently, mutations in CD2AP, an immunoglobulin-like protein that is involved in nephrin integration with the podocyte cytoskeleton, have also been linked to genetic forms of FSGS [10, 13, 14]. In mouse models, the loss of FAT1 and FAT2 (transmembrane proteins with cadherin-like repeats) results in the absence of slit diaphragms, proteinuria, and early death [10]. The role of the other components of the slit diaphragm in the pathophysiology of FSGS is not yet clear. Alpha-actinin-4, an important structural component of the podocyte cytoskeleton, is mutated in some autosomal dominant forms of FSGS [10, 15–17]. Other mutations have been identified in association with FSGS in addition to abnormal structural proteins. For example, TRPC6 is a cation-selective ion-channel protein that mediates calcium signals and has also been associated with FSGS [18]. Certain clinical variants of FSGS are suggestive of different mechanisms of injury to the podocyte. For example, a circulating factor which leads to glomerular basement membrane injury has been proposed in the pathogenesis of some types of FSGS [19, 20]. For example, there appears to be a role of a circulatory factor in the recurrence of FSGS in transplanted kidneys [20]. In some patients with recurrent FSGS, proteinuria remits in response to plasmapheresis and the removal of serum proteins. In addition, injections of serum from patients with recurrent FSGS were capable of inducing proteinuria in rats [20]. Another example of alternative mechanisms of injury is collapsing FSGS, which occurs in the setting of viruses such as HIV. In collapsing FSGS, dysregulation of the podocyte cell cycle appears to result in immature, proliferative podocytes [21, 22]. Finally, recent work has focused on the role of the parietal epithelial cell in the pathophysiology of FSGS [23]. Proliferation of parietal epithelial cells was identified in both a transgenic model of FSGS and a biopsy from a patient with collapsing FSGS [23]. Of great clinical importance is the mechanism by which the initial podocyte injury progresses to the final sclerotic lesion (Fig. 1). As podocyte numbers decline, there is a relative exposure of the glomerular basement membrane. Maladaptive interactions develop between the glomerular basement membrane and the parietal epithelial cells. Expansion of synechiae and/or the leak of protein into Bowman’s space results in the deposition of collagen. Ultimately, this results in the collapse of the capillary loop and the loss of endothelial cells [5]. Factors resulting in the progression of FSGS to ESKD have also been the focus of recent research (Fig. 2). Cytokines and vasoactive factors are believed to play a major role in the progression of FSGS. The overexpression of transforming growth factor β (TGFβ) or its effector proteins, the Smads, leads to glomerulosclerosis in animal models [8, 24]. Activation of the renin-angiotensin system upregulates TGFβ and is felt to further lead to the progression of disease [7, 24]. Other angiogenic factors, such as platelet-derived growth factor (PDGF) and vascular endothelial growth factor (VEGF) may also play a role in disease progression [24]. The evidence for this is primarily based on animal models of progressive glomerulosclerosis, such as the rat remnant kidney model. In this model, PDGF and VEGF are upregulated and the later loss of VEGF expression correlates with progression of the glomerulosclerosis [24, 25]. Fig. 2Factors involved in the progression of FSGS to end-stage kidney disease (ESKD): initial loss or injury to podocytes (related to defects in membrane proteins or cytoskeleton instability) leads to cytokine release, mechanical stress, hyperfiltration, and glomerular hypertrophy. These factors lead to upregulation of an inflammatory response mediated by monocytes, macrophages, and T-cells. The end result is collagen matrix deposition and fibrosis, and progression to ESKD Mechanical stress is also believed to play a role in the progression of FSGS [9, 26]. Increased filtration due the defects of the filtration barrier results in increased single-nephron glomerular filtration rate (SNGFR). This hyperfiltration results in hypertrophy of glomeruli. The hypertrophy exacerbates the mismatch between the glomerular basement membrane and the decreased numbers of podocytes, resulting in further injury [9]. Another factor in the progression of FSGS is tubulointerstitial injury. Clinically, tubulointerstitial injury is a predictor of the loss of renal function in FSGS [1, 27]. The nonspecific entry of proteins into the tubular lumen is one potential source of damage to the interstitium. Indeed, persistence of nephrotic-range proteinuria is a negative prognostic factor for the progression of FSGS to ESKD [28]. While it is unclear if proteinuria itself is toxic to the tubulointerstitium, decreases in proteinuria achieved by angiotensin-converting enzyme (ACE) inhibitors and by angiotensin receptor blockers (ARB) appear to slow disease progression in some adults with FSGS [9, 29]. The presence of plasma proteins in the tubular filtrate may directly injure the tubulointerstitium. Cytokines (such as TGFβ), when present in the tubules, will recruit monocytes, macrophage, and T-cells. This stimulates other cytokines, including interleukin-1, tumor necrosis factor alpha, and other chemokines [24]. The inflammatory infiltrate leads to mesangial matrix deposition, promoting the collapse of glomeruli. The cellular infiltrate and cytokines also damage tubular epithelial cells, and some tubular epithelial cells may undergo transformation to mesenchymal cells (an epithelial-mesenchymal transition or EMT) [24]. These mesenchymal cells, as well as recruited and stimulated fibroblasts, result in collagen matrix deposition and tubulointerstitial fibrosis [24]. The beneficial effects of blocking the renin-angiotensin system may not be limited to their antiproteinuric or antihypertensive effects. As noted earlier, angiotensin stimulates TGFβ, contributing to fibrosis. It can also induce oxidative stress and it is stimulated by mechanical stress, such as hyperfiltration [24]. In addition, angiotensin affects intracellular calcium concentrations and the podocyte cytoskeleton [24]. Inhibition of angiotensin may slow progression by these local mechanisms [9, 29]. With the increasing incidence of FSGS in children [30], these pathways of podocyte injury and disease progression provide important targets for future intervention. Trials have already been initiated to antagonize cytokines, such as TGFβ. Future therapeutic targets may include factors involved in podocyte protection or tubulointerstitial injury. Questions (Answers appear following the reference list) Which of the following statements is TRUE regarding the current understanding of the pathogenesis of focal segmental glomerulosclerosis (FSGS)? FSGS may result from immune-complex-mediated damage to endothelial cellsAlterations in components of the slit diaphragm may play a role in the pathogenesis of FSGSProliferation of podocytes leads to cytokine release and mechanical stress, resulting in scarring and sclerosis of the glomeruliMutations in a chloride channel have been associated with FSGS and may be pathogenicAll of the following are mutations of structural proteins that have been identified as pathogenic in FSGS EXCEPT: Sodium channel mutationAlpha-actinin-4NephrinPodocinProgression of FSGS to end-stage kidney disease (ESKD) results from: Downregulation of transforming growth factor β (TGFβ)Decreased glomerular filtrationTubulointerstitial injuryBlockade of the renin-angiotensin systemProteinuria in the setting of FSGS: Has no effect on clinical courseMay be decreased by treatment with angiotensin-converting enzyme (ACE) inhibitorsResults from an increased number of glomerular foot processesLeads to the loss of mesangial matrixWhich of the following is FALSE: A circulating factor may play a role in the pathogenesis of FSGSProliferation of parietal epithelial cells has been identified in collapsing FSGSPodocyte loss due to necrosis appears to play a role in the pathogenesis of FSGSCD2-associated protein, FAT, nephrin, and podocin are examples of slit diaphragm proteins

Neurosurg_Rev-2-2-1564192

Clinical and radiological features related to the growth potential of meningioma

Clinical and radiological features that help predict the growth potential of meningioma would be beneficial. The purpose of this study is to clarify the characteristics related to proliferating potential using the MIB-1 staining index. We analyzed the relationship of MIB-1 staining indices to characteristics of 342 consecutive patients with meningioma surgically removed between 1995 and 2004 by logistic regression analysis. One hundred and forty-nine of the patients with meningioma were ≥60 in age; 89 male; 48 recurrent; 203 symptomatic; 157 at the skull base; 124 over 20 cm3; 24 multiple; 136 with edema; 117 with calcification. The MIB-1 staining index in 56 of 296 grade I meningiomas in WHO classification was ≥ 3.0; in 27 of 28 grade II; and in 17 of 18 grade III, respectively. Logistic regression analysis demonstrated that male (odds ratio [OR], 2.374, p=0.003), recurrence (OR, 7.574, p=0.0001), skull base (OR, 0.540, p=0.035), calcification (OR, 0.498, p=0.019) were independent risk factors for a high MIB-1 staining index (≥3.0); age, symptomatic, volume, multiple, edema were not. Male, recurrence, non-skull base, absence of calcification are independent risk factors for a high proliferative potential. These should be taken into consideration when managing meningiomas. Introduction An estimated 2–3% of the population has an incidental asymptomatic meningioma in autopsy studies [14, 20]. With the wider use of CT and MRI, many meningiomas are discovered as incidental findings during investigation for unrelated symptoms [9, 13, 16, 17, 20, 21]. The growth potential of meningiomas varies. Some meningiomas remain unchanged in size for a long period, whereas others grow rapidly [15]. Sex, age, initial tumor size, and calcification were reported to be related to the tumor growth judging from follow-up scans [9, 13, 16, 17, 21]. The nuclear antigen Ki-67 expressed by proliferating cells has become available for routinely processed paraffin section. The MIB-1 antibody detects an epitope on the Ki-67 antigen, a nuclear protein present only during active phase of the cell cycle (G1, S, G2, and M) [2]. Several studies investigated how Ki-67 labelling indices could help to predict recurrence [1, 3, 4, 6–8, 11, 12, 18, 19]. An increased MIB-1 staining index was highly correlated with a shorter tumor volume doubling time [12]. In the cases that showed an MIB-1 of ≥ 3%, the tumor volume doubling time was <2 years. Nakaguchi et al. [12] found the formula which can calculate tumor doubling time (Td) from the MIB-1 staining index at surgery: log Td=31.4–0.14×MIB-1 SI (R2=0.556). The time interval to the next recurrence for recurrent meningiomas is associated with the MIB-1 staining index. Meningiomas with MIB-1 staining index of 3% or higher had a significantly higher tendency of recurrence [11]. Although these cell kinetics methods are valuable for growth potential, they can be applied only after the verification of pathology [15]. The purpose of this study was to clarify the clinical and radiological features related to meningioma proliferation using the MIB-1 staining index. Sex, age, calcification, edema, symptom, size, and shape of tumor were already reported to be related to the tumor growth and MIB-1 staining index [1, 3, 4–8, 11–13, 16–19, 21]. The results, however, were inconsistent because of a lack of sufficient case number and inadequate statistical analysis. Independent risk factors for high growth rate should be determined. We analyzed the relationship of MIB-1 staining indices to the characteristics of 342 consecutive patients with meningioma surgically removed between 1995 and 2004 by logistic regression analysis. Materials and methods Three hundred and forty-two patients with meningiomas were surgically treated in our department of neurosurgery between 1995 and 2004. Radiological features were analyzed by CT scans and/or MRI. Location of the tumor was classified as follows: convexity, falx, parasagittal, sylvian fissure, tentorial, ventricular, foramen magnum, olfactory groove, petroclival, petrous, sphenoid ridge, and tuberculum sellae. The latter six locations were considered as skull base. The tumor volume was calculated using the formula: length × depth × width × 0.5 [9]. When patients had multiple meningiomas, only the largest tumor was included. On the basis of conventional CT and bone window CT, patients were divided into two groups according to the low density area around the tumor and calcification in the tumor. A low density area due to surgical scar was not included in edema in patients with recurrent meningioma. The tumors were histologically classified according to the World Health Organization classification of tumors [10]. An avidin-biotin immunoperoxidase or simple stain MAX-peroxidase (Nichirei, Tokyo) technique was used to perform MIB-1 monoclonal antibody (DAKO, Denmark) assay in selected sections of each case. All tissue sections were examined at high-power magnification (×400). The number of cells stained positively with MIB-1 and the total number of tumor cells were counted in several representative fields containing more than 1,000 cells. Their ratio was indicated as the MIB-1 staining index (%). Statistical analysis All data were stored on a personal computer and analyzed using commercially available statistical software (SPSS version 12.0, SPSS Inc.). Chi-squired analysis was used to compare the MIB-1 staining index to characteristics of patients with meningioma. All variables were included in a logistic regression model to determine which variables were independently associated with a high MIB-1 staining index (≥3.0). Significance was judged at a value of p<0.05 for all analyses. Results Table 1 shows the characteristics and MIB-1 staining index of the 342 patients. One hundred and forty-nine of patients with meningioma were ≥60 in age; 89 male; 48 recurrent; 203 symptomatic; 157 at the skull base; 124≥20 cm3 in volume; 24 multiple; 136 with edema; 117 with calcification. We compare these characteristics to the MIB-1 staining index. We divided them into two groups: <3.0 and ≥3.0 [11]. The MIB-1 staining index in 100 of 342 meningiomas was > 3.0. Sex (p=0.0001), recurrence (p=0.0001), symptomatic (p=0.013), volume (p=0.014), edema (p=0.001), and calcification (p=0.0001) were correlated with the MIB-1 staining index by chi-square test; age, skull base, and multiple were not. Table 1Characteristics and MIB-1 staining index in 342 meningiomas MIB-1 staining index (%) Factor<3.0≥3.0P ValueAge (years)0.937-49652450-723260-793470-2610Sex (male/female)46/19643/570.0001Recurrence (yes/no)14/22834/660.0001Symptomatic (yes/no)133/10970/300.013Skull base (yes/no)119/12338/620.059Volume (cm3)0.014-9.91133710-531520-7648Multiple (yes/no)15/2279/910.244Edema (yes/no)82/16054/460.001Calcification (yes/no)97/14520/800.0001Total242100 Meningothelial, transitional, and fibrous meningiomas were the three major subtypes, and they accounted for about three fourth of the total. Two hundred and ninety-six meningiomas belonged to grade I; 28 grade II; and 18 grade III. MIB-1 staining index in 56 of 296 grade I meningiomas was ≥ 3.0; that in 27 of 28 grade II; and that in 17 of 18 grade III, respectively (Table 2).Table 2Histological subtypes and MIB-1 staining index of 342 meningiomasSubtypeMIB-1 staining index (%)Total<3.0≥3.0Grade IMeningothelial 10628134Fibrous671380Transitional431356Psammomatous606Angiomatous13215Microcystic101Secretory101Lymphoplasmacyte-rich101Metaplastic20224056296Grade IIAtypical12627Chordoid01112728Grade IIIRhabdoid011Papillary112Anaplastic0151511718Total242100342 Logistic regression analysis demonstrated that male (odds ratio [OR], 2.374, p=0.003), recurrence (OR, 7.574, p=0.0001), skull base (OR, 0.540, p=0.035), calcification (OR, 0.498, p=0.019) were independent risk factors for a high MIB-1 staining index (≥ 3.0); age, symptomatic, volume, multiple, and edema were not (Table 3). Table 3Logistic regression analysis for factors independently related to MIB-1 staining indexFactorOdds ratio95%CIP ValueAge1.1090.841–1.4610.464Sex2.3741.336–4.2190.003Recurrence7.5743.558–16.1240.0001Symptomatic1.4680.774–2.7840.240Skull base0.5400.305–0.9560.035Volume1.3320.944–1.8790.103Multiple1.0270.398–2.6510.957Edema1.5080.838–2.7110.170Calcification0.4980.278–0.8920.019 Discussion We analyzed the relationship of the MIB-1 staining indices to the characteristics of 342 consecutive patients with meningioma surgically removed between 1995 and 2004 by logistic regression analysis. Logistic regression analysis demonstrated that male (odds ratio [OR], 2.374, p=0.003), recurrence (OR, 7.574, p=0.0001), skull base (OR, 0.540, p=0.035), calcification (OR, 0.498, p=0.019) were independent risk factors for a high MIB-1 staining index (≥ 3.0); age, symptomatic, volume, multiple, and edema were not. The relationship between the growth rate or MIB-1 staining index and age has been controversial: a higher MIB-1 staining index and higher growth rate were observed for younger patients [11, 13, 21]; but not in other reports [1, 12, 15, 19]. Our series of 342 patients with meningioma showed no relation. It is well known that atypical and anaplastic meningiomas are predominant in males [10]. Matsuno et al. [11] reported that the mean MIB-1 staining index in 50 male patients was 5.5%, whereas that in 77 female patients was 2.7%. Our findings show that male (odds ratio [OR], 2.374, p=0.003) was an independent risk factor for a high MIB-1 staining index. We also found a higher MIB-1 staining index in males even in grade I meningioma (MIB-1 staining index in 32 of 226 females, and in 24 of 70 males was ≥ 3.0, p=0.0001, chi-squire test). Recurrence (OR, 7.574, p=0.0001) was the most significant independent risk factor for a high MIB-1 staining index (≥ 3.0). Therefore, we propose prompt management for recurrent meningiomas. In most of the recurrent meningiomas, the MIB-1 staining index was higher at the time of recurrence than at the time of initial surgery [1, 11, 19]. Changes in histological morphology and malignant transformation are also known in meningiomas. Although there is a significant difference in the MIB-1 staining index between symptomatic and non-symptomatic meningiomas by chi-square test, symptomatic is not an independent risk factor for a high MIB-1 staining index. Meningiomas commonly present with seizure disorders, and are associated with location, perilesional edema, and convexity location. Symptoms and signs of elevated intracranial pressure could be due to the large size of meningioma itself, or to the pronounced cerebral swelling resulting from reactive vasogenic edema. Focal neurological deficits caused by meningiomas are generally related to direct local brain, cranial-nerve compression, and can be predicted from the site of origin of the tumor [20]. Thus, symptomatic meningioma may not be related to a high MIB-1 staining index. The growth rate of incidental meningioma may be similar to that of symptomatic meningioma. Our results demonstrated that skull base (OR, 0.540, p=0.035) is an independent risk factor for a high MIB-1 staining index. No relationship has been reported between the MIB-1 staining index and the location of meningiomas [15, 19]. In general, the surgical risk for meningiomas is higher in skull base. A low proliferative potential in skull base meningiomas should be taken into consideration especially when treating elderly patients with asymptomatic meningiomas [9, 16]. Although there was a significant difference of the MIB-1 staining index in tumor size by chi-square test [15], tumor size is not an independent risk factor for a high MIB-1 staining index. The tumor volume is associated with the annual growth rate but not with doubling time [13, 16, 21]. Assuming that a tumor shows a constant relative volume increase, larger tumors will show higher growth rates if the annual volume increase is expressed in absolute values. Large tumors should be carefully observed even though the initial volume is not a risk factor for a high MIB-1 index, and it is not correlated with doubling time. The etiology of peritumoral brain edema associated with meningiomas is multifactorial. Factors that may influence the etiology of peritumoral edema include tumor size, histological subtypes, vascularity, venous stasis, and brain invasion [1, 4]. Ide et al. [4] found a significant correlation of both the MIB-1 staining index and tumor size with the extent of edema. A high MIB-1 staining index itself did not seem to be directly responsible for perifocal edema, since our logistic regression analysis demonstrated that edema is not an independent risk factor for a high MIB-1 index [15]. Tumors with calcification grew significantly less than those without calcification [9, 13, 16]. Absence of calcification on CT correlated strongly with doubling time [5]. Diffusely calcified meningiomas had a low mean MIB-1 staining index of 0.57%. Focally calcified tumors showed a relatively low proliferative potential (0.92%) compared with that of noncalcified tumors (1.75%) [15]. The results are always consistent when the relationship between calcification and proliferative potential or growth rate is compared. We also confirmed this characteristic in meningioma (OR, 0.498, p=0.019). In conclusion, male, recurrence, non-skull base, absence of calcification are independent risk factors for a high proliferative potential. These should be taken into consideration when managing meningiomas.

Mech_Dev-2-1-2428104

Polydactyly in the mouse mutant Doublefoot involves altered Gli3 processing and is caused by a large deletion in cis to Indian hedgehog

The mouse mutant Doublefoot (Dbf) shows preaxial polydactyly with 6–9 triphalangeal digits in all four limbs and additional abnormalities including a broadened skull, hydrocephalus, and a thickened, kinked tail. The autopod undergoes a characteristic expansion between late embryonic day (E) 10.5 and E11.5, following the onset of ectopic Indian hedgehog (Ihh) expression in the entire distal mesenchyme, except for the zone of polarising activity (ZPA), at E10.5. We show here that limb prepattern, as indicated by expression of Gli3 and Hand2 at E9.5 is unaffected by the mutation. As both Sonic hedgehog (Shh) and Ihh expression are present in Dbf limb buds at E10.5, we generated Dbf/+;Shh−/− mutants to analyse the effects of different patterns of Hedgehog activity on the limb phenotype and molecular differentiation. Dbf/+ embryos lacking Shh showed postaxial as well as preaxial polydactyly, and the Ihh expression domain extended posteriorly into the domain in which Shh is normally expressed, indicating loss of ZPA identity. Differences in gene expression patterns in wild type, single and compound mutants were associated with differences in Gli3 processing: an increased ratio of Gli3 activator to Gli3 repressor was observed in the anterior half of Dbf/+ limb buds and in both anterior and posterior halves of compound mutant limb buds at E10.5. To identify the cause of Ihh misregulation in Dbf/+ mutants, we sequenced ∼20 kb of genomic DNA around Ihh but found no pathogenic changes. However, Southern blot analysis revealed a ∼600 kb deletion disrupting or deleting 25 transcripts, starting 50 kb 5′ of Ihh and extending away from the gene. The large deletion interval may explain the wide range of abnormalities in Dbf/+ mutants. However, we did not detect anologous deletions in cases of Laurin–Sandrow syndrome, a human disorder that shows phenotypic similarities to Dbf. 1 Introduction The Dbf mutant, which arose spontaneously in the 3H1 (C3H/HeH × 101/H F1 hybrid) genetic background at Harwell (UK), is a polydactylous mouse that exhibits semidominant inheritance. Mice heterozygous or homozygous for Dbf have 6–9 digits in all four limbs; the extra digits are all triphalangeal and arise preaxially (Lyon et al., 1996; Hayes et al., 1998a). Dbf/+ mice also show malformation of the tibia, a broadened skull, hydrocephalus, a thickened kinked tail, and reduced fertility and viability. Homozygotes additionally exhibit a midline facial cleft but cannot be recovered alive beyond embryonic day (E) 14.5. Polydactyly has been described in many mouse mutants, all except two of which show a discrete anterior domain of Sonic hedgehog (Shh) expression (Masuya et al., 1995; Hill et al., 2003). The Extra-toes (XtJ) mutant has an extended Shh domain due to functional inactivation of Gli3 (Hui and Joyner, 1993), whereas Dbf mice exhibit ectopic Indian hedgehog (Ihh) expression in the distal limb bud mesenchyme (Yang et al., 1998). Ectopic Ihh upregulation is first detectable at E10.5 (Crick et al., 2003), the stage at which hyperexpansion of the autopod begins; downstream targets of Shh signalling are ectopically up-regulated (Hayes et al., 1998b; Yang et al., 1998). However, the molecular mechanism by which the polydactyly arises from ectopic Ihh expression has not been investigated. The polydactylous phenotype of the XtJ mutant was originally thought to result from the enlarged Shh expression domain (Hui and Joyner, 1993). However, Shh−/−;Gli3−/− mutants exhibit polydactyly in a similar pattern to Gli3−/− mutants, suggesting that the polydactyly of Gli3-deficient mice is independent of Shh (te Welscher et al., 2002). In wild type (wt) limb buds, digital number and identity are regulated by interaction between Shh and Gli3 (Litingtung et al., 2002). In the presence of Shh, Gli3 remains as a 190 kDa activator species, Gli3A, that up-regulates Hedgehog (Hh)-responsive gene expression, while in the absence of Shh, Gli3A is processed to a smaller 83–86 kDa repressor form, Gli3R, which negatively regulates expression of Shh and its target genes (Dai et al., 1999; Shin et al., 1999; Sasaki et al., 1999). Litingtung et al. (2002) suggested that in wt limb buds the Gli3A:Gli3R ratio controlled by Shh limits the polydactylous potential of the autopod, imposing pentadactyl constraint. This is supported by the localization of Shh protein in wt limb buds, which extends anterior to the zone of polarising activity (ZPA) in a domain coincident with Patched1 (Ptc1) expression (Gritli-Linde et al., 2001), resulting in a posterior-to-anterior increase of the Gli3A:Gli3R ratio (Wang et al., 2000). Consistent with these observations, the Gli3 present throughout Shh−/− limb buds is mainly processed to Gli3R (Litingtung et al., 2002). Recently, the mutation underlying the polydactylous chicken talpid3 mutant has been reported to be in a novel gene and has also been shown to result in abnormal Gli3 processing (Davey et al., 2006). Given the evidence of involvement of abnormal Gli3 processing in the XtJ, Shh−/− and talpid3 mutants, it is possible that the polydactyly present in Dbf mice also results from aberrant Gli3 processing. This hypothesis is supported by evidence that Gli3 acts downstream of Ihh during endochondral skeletal development (Hilton et al., 2005; Koziel et al., 2005). To investigate the mechanism by which polydactyly arises in Dbf we have analysed gene expression in Dbf/+ limbs, where there is an excess of Hedgehog (Hh) signalling, and compared this to Shh−/− limbs, where there is none. Since Shh and Dbf are located on different chromosomes (5 and 1, respectively) (Blake et al., 2003; Hayes et al., 2001), we have been able to generate mutant mice that carry two copies of the disrupted Shh allele and are heterozygous for the Dbf mutation. To further dissect the mechanisms underlying the limb malformations in both Shh and Dbf mutants, we have analysed the effects of the ectopic Ihh expression associated with Dbf limb abnormalities in the Shh-null background by correlating altered patterns of gene expression with the phenotype of single and double mutants. Differences in Gli3 processing between each genotype suggest that Hh-Gli3 interactions govern the observed differences in digital number, and that postaxial polydactyly results from expression of Ihh, but not Shh, in the posterior ZPA mesenchyme. Previous attempts to identify the Dbf mutation have been unsuccessful. Hayes et al. (2001) constructed a high resolution genetic map and localized the mutation to a 0.4 cM interval on mouse chromosome 1. This region contained 35 genes including several plausible candidates for the Dbf mutation. However, despite the sequencing of three of these genes, the Dbf mutation remained unidentified. Based on the misregulation of Ihh expression in Dbf, we sequenced ∼20 kb of the surrounding genome but found no obvious pathogenic changes. To investigate whether a genomic rearrangement could be responsible, we used the mouse genome sequence to design a Southern blotting strategy to systematically screen the regions 5′ and 3′ of Ihh for copy number changes. We identified a ∼600 kb deletion starting ∼50 kb 5′ of Ihh, which removes or interrupts 25 known and predicted transcripts. This raises the possibility that additional abnormalities seen in Dbf/Dbf mice arise from loss of function of deleted genes, in addition to Ihh misregulation. 2 Results 2.1 The prepattern of Dbf limb buds is unaffected Expression of Hand2 and Gli3 has been implicated in patterning the limb bud prior to Shh expression, and has been shown to be affected later by the absence of Shh (Chiang et al., 2001; te Welscher et al., 2002). We assayed expression of these two genes before (E9.5) and after (E11.5) the onset of ectopic Ihh at E10.5 in Dbf/+ mutant embryos (Fig. 1). Gli3 expression is restricted to the anterior portion of the limb bud in wt embryos at E9.5 (Fig. 1A) and this expression pattern is not altered in the limb buds of Dbf/+ mutants (Fig. 1B). Hand2 is expressed throughout the flank of wt embryos prior to formation of the limb bud, then becomes limited to the posterior region of the limb bud as it is initiated (Fig. 1C); this pattern is not altered in Dbf/+ embryos at E9.5 (Fig. 1D). At E11.5, expression of Gli3 in Dbf/+ limb buds differs from that in wt embryos in extending more distally; the domain is also broader although this probably simply reflects the greater breadth of the limb bud (Fig. 1F). Expression of Hand2 is limited to the proximal posterior margin in wt E11.5 limb buds (Fig. 1G); in contrast, the Hand2 domain in Dbf/+ limb buds extends anteriorly and distally (Fig. 1H). Hence the limb prepattern as indicated by the expression of Hand2 and Gli3 at E9.5 is unaffected in Dbf/+ limb buds, but the expression domains of both genes are altered in association with the presence of ectopic Ihh expression at E11.5 (Fig. 3H). 2.2 Altered limb phenotype of Dbf mutants in the absence of Shh As Shh-null embryos die perinatally, gross morphological examination of wt, Shh−/−, Dbf/+ and Shh−/−;Dbf/+ embryos was conducted at E13.5 and alcian blue staining of the limb bones was carried out at E17.5 (Fig. 2). Both forelimb and hindlimb autopods of Shh−/−;Dbf/+ embryos resemble those of Dbf/+ except that the broadened digital plate is more regular, shows fewer bifurcations, and is more extensive posteriorly (compare Fig. 2B, F and J with D, H and L). At E13.5 the autopod forms a 180° fan, and the angle between the autopod and zeugopod on the postaxial side of the limb is decreased to 90° (Fig. 2D, arrow). 2.3 Ihh and Shh expression in compound mutant limbs is mutually exclusive The expansion that characterizes the Dbf/+ autopod takes place from late E10.5 to E11.5. We therefore analysed the expression domains of Shh and Ihh in limb buds immediately prior to (E10.5) and after (E11.5) the period of expansion. In both wt and Dbf/+ limb buds at E10.5, Shh is expressed at the posterior margin (Fig. 3A and B), defining this region as the ZPA (Riddle et al., 1993). In wt mice Ihh is not expressed in limbs prior to E12.5 (St-Jacques et al., 1999) while in Dbf/+ mutant mice, Ihh expression is present in the distal mesenchyme of the limb bud at E10.5 (Fig. 3C). This ectopic Ihh domain extends throughout the area anterior to the ZPA and may correspond to the progress zone. Its absence from the ZPA was confirmed by double in situ hybridization to show nonoverlapping juxtaposed Shh and Ihh expression (Fig. 3D). Expression of Ihh in E10.5 Shh−/−;Dbf/+ mutant limb buds extends throughout the distal mesenchyme including the posterior margin, i.e. the domain in which Shh is expressed in Dbf/+ embryos (Fig. 3E). At E11.5, Shh expression continues in the posterior margin of wt and Dbf/+ limb buds (Fig. 3F and G). Expression of Ihh in Dbf/+ mutant limb buds at E11.5 is progressively down-regulated from posterior to anterior, until it remains only in the anterior margin (Fig. 3H); in contrast, in Shh−/−;Dbf/+ mutant limbs, down-regulation of Ihh expression begins mid-distally, remaining strong in both the anterior and posterior mesenchyme (Fig. 3I). 2.4 Gene expression is altered in Dbf limb buds lacking Shh To gain insight into the mechanisms underlying the different patterns of polydactyly generated in the presence of different sources of Hh signalling in Dbf/+ and Shh−/−;Dbf/+ limbs, we examined the expression of genes implicated in Shh signalling and limb patterning in wt, Dbf/+, Shh−/− and Shh−/−;Dbf/+ limb buds at E10.5 (Fig. 4); as shown in Fig. 3, this is the stage at which Ihh expression is first detected. Expression of the transcriptional targets of Hh signalling, Ptc1 and Gli1, is expanded anteriorly in Dbf/+ limbs; interestingly, expression of these genes is broader in the proximal mesenchyme of Dbf/+limbs lacking Shh, suggesting expansion of the domain of Hh signalling in these limb buds. Conversely expression of Gli3, which is thought to be repressed by Hh signalling (Takahashi et al., 1998), shows a reduced expression domain in Dbf/+ limbs. As expected, Gli3 is expressed throughout Shh−/− limbs at E10.5, but in the presence of Ihh in Shh−/−;Dbf/+ mutants it is dramatically down-regulated and required a prolonged colour development time for detection. In wt and Dbf/+ limbs at E10.5 there is a strong expression of Hand2 in the posterior mesenchyme, with a graded lower expression anteriorly, similar expression is seen in Shh−/− limbs. However, in Shh−/−;Dbf/+ limbs there appears to be a second strong anterior domain of Hand2 expression, consistent with the extended expression seen at E11.5 (Fig. 1H). As reported previously (Hayes et al., 1998b; Yang et al., 1998), the Hoxd13 domain is expanded anteriorly in Dbf/+ limb buds; in Shh−/−;Dbf/+ limb buds, the domain shows even greater expansion, consistent with the more regular digital fan seen in these mutants. Expression of Fgf8 throughout the AER of expanded Dbf/+ and Shh−/−;Dbf/+ limbs indicates that in both mutants Hh signalling between the mesenchyme and ectodermal AER is intact. Ectopic anterior expression of Fgf4 in the expanded limb buds of both mutants is consistent with their ectopic Ihh expression. Bmp4 expression in the progress zone was slightly down-regulated in Dbf/+ limbs but up-regulated proximally; like wt limbs, it was absent from the AER. In contrast, Shh−/−;Dbf/+ limbs, which showed further down-regulation of Bmp4 in the mesenchyme of the progress zone, showed ectopic expression throughout the AER. Explanation for this pattern requires further investigation. 2.5 The Dbf mutation affects the limb bud Gli3 ratio The action of Gli3 protein as a transcriptional activator relies on its maintenance as Gli3A, which requires Hh signalling (Dai et al., 1999; Sasaki et al., 1999; Shin et al., 1999). To determine the effect of differential Hh signalling on Gli3 processing in mutant limb buds, we used a Gli3 antibody combined with Western blot analysis to assess the comparative levels of Gli3A and Gli3R in the anterior and posterior halves of E10.5 limb buds of all four genotypes (Fig. 5). As reported previously, wt limbs have a higher ratio of Gli3R to Gli3A anteriorly than posteriorly (Wang et al., 2000, and Fig. 5B and C). Dbf/+ limb buds have a reduced level of the repressor relative to the activator, especially in the anterior half, where levels of the two forms of Gli3 are similar. In Shh−/− limb buds the difference between the anterior and posterior halves is greatly reduced with relatively high levels of Gli3R to Gli3A in both halves of the limb bud mesenchyme (Litingtung et al., 2002, and Fig. 5B and C). In Shh−/−;Dbf/+ mutants, both halves of the limb bud show a decreased ratio of Gli3R to Gli3A compared with the wt result; the ratio is similar in both halves of the limb bud, in contrast to the Dbf/+ result which shows an A–P asymmetry. 2.6 A ∼600 kb deletion underlies Dbf To determine the cause of the ectopic Ihh expression in Dbf limbs, we initially searched for the genetic lesion by sequencing 20 kb of the region around Ihh in Dbf heterozygotes and both parental strains, but found no pathogenic changes (data not shown). Subsequently we sought genomic rearrangements using a systematic Southern blotting strategy to interrogate the mouse genome sequence (http://genome.ucsc.edu/), which initially identified the absence of a polymorphic 8 kb SpeI fragment in Dbf (see Section 4.5). Characterization of the breakpoint by Southern analysis and subsequently by inverse PCR led to identification of the centromeric breakpoint at position 75,098,488 bp on chromosome 1 (Fig. 6). Analysis of sequence 3′ to this in Dbf/+ DNA revealed the telomeric breakpoint to be at position 75,694,480 bp on chromosome 1. The deleted region therefore appears to be 595,992 bp; however this figure is not precise because the deletion encompasses a ∼16 kb gap in the current mouse genome sequence (mm9 assembly) present between 75,102,130 and 75,118,131 bp. We confirmed the deletion by PCR using primers flanking the breakpoint and further demonstrated that three different loci distributed within the putatively deleted region were present only in a single copy in Dbf/+ mutant DNA (see Section 4.5). Analysis of the wt sequences at the two breakpoints showed that the sequence at the centromeric breakpoint is unique, lying within the gene Non-homologous end joining factor 1 (Nhej1). However, a hexanucleotide motif CCAAAC present at the breakpoint is repeated 17 nucleotides upstream, separated by four copies of a trinucleotide CCT motif. The telomeric breakpoint resides within the 3′ terminal region of a B1 repetitive element at the endpoint of a very T-rich motif (35 thymine residues in 47 bases) which is likely to represent the complement of an ancestral poly(A) tract related to the B1 element and does not disrupt any known gene. There is a three nucleotide ambiguity in the position of the breakpoint as the sequence ACA is present on both sides of the deletion (Fig. 6). In addition to disrupting Nhej1, the deletion completely removes 24 known and predicted genes (Fig. 6, Supplementary Table 2 and Section 3). 2.7 Laurin–Sandrow syndrome does not result from large deletions 5′ of IHH Laurin–Sandrow syndrome (LSS) (MIM 135750) is rare human developmental disorder characterized by triphalangeal preaxial polydactyly of the hands and feet, with variable involvement of the proximal limb elements. It has been previously suggested that LSS shares many similarities with Dbf and may also arise from ectopic IHH expression (Innis and Hedera, 2004). To investigate the possibility that Dbf and LSS share a common etiology, we screened five patients diagnosed with LSS for copy number variation at 23 sites between IHH and EPHA4 using multiplex ligation-dependent probe amplification (see Supplementary Information). No copy number variation was detected (data not shown). 3 Discussion 3.1 Ectopic Ihh expression in Dbf/+ is modified in the absence of Shh and is associated with loss of ZPA identity Although we have previously shown that expression of ectopic Ihh in Dbf/+ limb buds coincides with the onset of limb bud expansion at E10.5 (Crick et al., 2003), it was not known whether the prepattern of Dbf limbs might be affected by the mutation prior to Ihh expression. However, no differences were detected in the expression of Gli3 or Hand2 in wt and Dbf/+ embryos at E9.5 or E10.5, consistent with the hypothesis that ectopic Ihh expression represents the primary pathogenic event. By E11.5, expression domains of both Hand2 and Gli3 were more extensive in Dbf/+ than wt limb buds, suggesting that Ihh signalling is able to modify their expression. In Dbf/+ limb buds, Ihh and Shh are expressed in discrete adjacent domains. Exclusion of Ihh from the Shh domain is reminiscent of the exclusion of the Hh-inducible gene Gremlin from this domain; Scherz et al. (2004) suggested that the effect may be due to high levels of intracellular autocrine Shh signalling. The loss of identity of the ZPA resulted in a abnormal expansion of the posterior limb bud mesenchyme in Shh−/−;Dbf/+ mice leading to the additional postaxial polydactyly seen in these mutants. 3.2 Abnormal gene expression leading to an aberrant Gli3 ratio underlies Dbf polydactyly To elucidate the limb patterning underlying Dbf/+ polydactyly and to investigate the generation of the broader, more regular fan of digits seen in Shh−/−;Dbf/+ mutants, we studied the expression of a range of limb patterning and development genes at E10.5. Dbf/+ mutant limbs show an anterior expansion of the positive regulators of Hh signalling Ptc1, Gli1 and the downstream targets Hoxd13 at E10.5 and Hand2 by E11.5. Dbf/+ limbs also show a reduction of Gli3 expression, which is thought to be negatively regulated by Hh signalling. Conversely, due to the complete lack of Hh activity in Shh−/−;Dbf/+ mutant limbs prior to E10.5, Gli3 is ubiquitously expressed in these limb buds until this stage, when it is down-regulated in Shh−/−;Dbf/+ but not Shh−/− mutants. Gli3R is thought to repress expression of Hoxd13 and Hand2 and Fgf4 in the anterior of wt limb buds while Gli3A induces the expression of Gli1 in the posterior region (reviewed in Tickle, 2006). Therefore, the postaxial polydactyly seen in Shh−/−;Dbf/+ mutants may be due to the loss of identity of the ZPA with concomitant posterior extension of the Ihh domain. In contrast, the preaxial polydactyly that is present in both Dbf/+ and Shh−/−;Dbf/+ mutants is correlated with ectopic Gli3A-induced Hh signal transduction together with lack of repression of posterior patterning genes by Gli3R in the anterior of the limb bud. We suggest that the discrepancy between the very low level of Gli3 mRNA (Fig. 4) and the Gli3 protein detected in Shh−/−;Dbf/+ limb buds at E10.5 (Fig. 5) indicates the perdurance of protein after the gene has been down-regulated. 3.3 Identification of the Dbf mutation The interpretation of the mechanism of the Dbf mutation has been hampered previously by the failure of attempts to identify the underlying mutation. Using a Southern blotting and inverse PCR strategy we have demonstrated that a ∼600 kb deletion underlies the Dbf phenotype. The presence of simple sequences at both breakpoints may have predisposed them to breakage; the lack of significant similarity between the breakpoints (except for a 3 nucleotide identity at the breakpoints themselves) suggests that the rearrangement is likely to have involved nonhomologous end joining (NHEJ). Further analysis of sequence at the breakpoints revealed that the distal breakpoint resides within the degenerate poly(A)n tract of a short retrotransposon (SINE) of the rodent B1 family, which, like human Alu repeats originate from 7Sl RNA (Vassetzky et al., 2003). The deleted region in Dbf is relatively gene-dense and completely deletes 24 known and predicted transcripts as well as interrupting Nhej1 at the centromeric breakpoint. Several of these genes have previously been implicated in abnormal mouse phenotypes or human disease; information on the known expression patterns and functions of these genes is summarized in Supplementary Table 2. Abnormalities associated with genes in the deleted region may contribute to additional aspects of the heterozygous Dbf phenotype such as the broadened skull, hydrocephalus, reduced viability and fertility, thickened tail and supernumerary hair follicles. However, none of the homozygous null phenotypes resulting from specific targeting of the Ptprn, Des, Inha or Slc4a3 genes is lethal in late embryogenesis so the cause of death at E14.5 in Dbf homozygotes remains unclear. This could be attributable to loss of function of any of the genes within the interval for which homozygous mice have not yet been described, and/or to homozygosity for the ectopic Ihh expression defect. Interestingly a recent study reported a human fetus with a balanced de novo translocation t(2;7)(q36;p22) with the chromosome 2 breakpoint interrupting the orthologue of Nhej1 at a position similar to the start of the Dbf deletion (Cantagrel et al., 2007). The consequence of this translocation, as in Dbf, would be to isolate the human IHH gene from possible regulatory sequences present on the opposite side of the NHEJ1 breakpoint. Although the terminated fetus exhibited syndactyly of all four limbs, polydactyly was not present, suggesting that the translocation did not result in ectopic IHH expression. We have presented evidence that the prepattern of Dbf limb buds is unaffected and that the preaxial polydactyly is attributable to a reduction in Gli3R resulting from ectopic Ihh expression. It is interesting that preaxial polydactyly, the most striking aspect of the Dbf phenotype, is unlikely to result directly from haploinsufficiency of any of the genes in the deleted region. Rather, the deletion appears to affect a cis-acting regulatory element of Ihh, which could be a repressor located within the deletion, or an enhancer beyond the deleted region. Other examples of regulatory mutations acting at a distance have been reviewed by Kleinjan and van Heyningen (2005). Pinpointing the regulatory sequences involved remains a major challenge, one notable success being the identification of the ZPA sequence regulatory sequence (ZRS) which lies ∼1.0 Mb upstream of Shh and regulates its expression in the ZPA; mutations in the ZRS lead to ectopic Shh expression resulting in preaxial polydactyly (Lettice et al., 2002). However, owing to the large size of the Dbf deletion and the large number of genes and highly conserved non-coding elements within it, it will be challenging to delineate the precise mechanism underlying ectopic Ihh expression in the Dbf mouse. 4 Materials and methods 4.1 Generation and identification of mutant mice Mice heterozygous for the Shh null allele (Chiang et al., 1996) on the C57BL/6J background were mated to Dbf/+ mice on the 3H1 background. The Shh mutant allele was detected as previously described (Chiang et al., 1996). Homozygous Shh−/− embryos were identified by their phenotype. To genotype Dbf/+ embryos (prior to the identification of the causative deletion), Dbf/+ mice were crossed with wt Mus musculus castaneus and the Dbf F1 progeny were bred with C3H wt mice. Embryos were genotyped using primers which amplify the marker D1Mit46 located ∼2.3 cM from Ihh (P1 5′-AGTCAGTCAGGGCTACATGATG-3′, P2 5′-CACGGGTGCTCTATTTGGAA-3′). This produces amplification products of 276 bp and 320 bp on the C3H and Mus musculus castaneus backgrounds respectively. 4.2 Whole mount in situ hybridization Doubly heterozygous Shh+/−;Dbf/+ were crossed with Shh+/− mice and embryos of all six possible genotypes were collected for analysis of gene expression domains and morphology: wt, Shh+/−, Shh−/−, Dbf/+, Shh+/−;Dbf/+ and Shh−/−;Dbf/+. To ensure consistency between developmental stages, only forelimb buds were analysed and a minimum of two samples were examined with each probe. Timing of embryos was by the vaginal plug method: 12.00 noon on the day on which the plug was observed was regarded as E0.5. Pregnant females were sacrificed on the appropriate day by cervical dislocation and the embryos were dissected from the uterus in ice cold phosphate buffered saline (PBS) (140 mM NaC1, 3 mM KCl, 10 mM Na2HPO4, 2 mM KH2PO4) followed by immersion in cold tissue fixative in accordance with the Animals (Scientific Procedures) Act, 1986. Where necessary, yolk sacs were removed for genotyping and embryos were fixed by immersion overnight at 4 °C in 4% paraformaldehyde in PBS. Embryos were dehydrated by sequential washing in 25%, 50%, 75% ethanol in PBT (PBS + 0.1% Tween 20) and finally by two washes in 100% ethanol. They were stored at −20 °C until required. Single stranded digioxygenin-UTP labelled antisense riboprobes were generated from linearized plasmids containing cDNAs. Whole mount in situ hybridization was carried out essentially as described by Wilkinson (1992). 4.3 Skeletal preparations Embryos for skeletal staining were dissected and fixed in 95% ethanol for 1–3 days. They were immersed in alcian blue stain (75% ethanol, 20% acetic acid, 3 mg/ml alcian blue) for 21–28 days at 37 °C. They were cleared in 0.8% KOH, 20% glycerol. Following clearing, they were sequentially dehydrated and stored in 50% ethanol/50% glycerol. 4.4 Western Blotting The polyclonal antibody specific for the amino terminus of Gli3 was a gift from Dr. Chin Chiang (Litingtung et al., 2002). Three μg of protein lysate derived from the anterior and posterior halves of ∼6 E10.5 forelimb buds were resolved on 4–12% polyacrylamide gels. Gli3 protein was detected using anti-N-terminal Gli3 (1:300) primary antibody and biotinylated anti-rabbit immunoglobulin-γ secondary antibody (1:1000). Protein bands were visualized by incubation with a streptavidin-peroxidase conjugate followed by an enhanced chemiluminescence detection method (Amersham). 4.5 Characterization of the Dbf deletion To determine whether genomic rearrangements were associated with Dbf, single copy probes labelled with α32P-dCTP were synthesized and used to hybridize Southern blots of DNA isolated from heterozygous 3H1 (C3H × H101 hybrid) Dbf mice and wt mice from both background strains. We used the mouse genome mm9 sequence release (July 2007) for all analyses presented in this paper. A probe corresponding to 75,099,314–75,099,651 bp revealed the absence of a polymorphic 8 kb SpeI fragment in Dbf, found in the C3H parental strain, suggesting the existence of a deletion. A further Southern blot using a probe corresponding to 75,097,987–75,098,255 bp revealed a 1.2 kb BspHI fragment present only in Dbf. This 1.2 kb breakpoint fragment was isolated by inverse PCR. Briefly, genomic DNA from Dbf was digested with BspHI, diluted to 10 ng/μl and T4 DNA ligase was added to promote intramolecular ligation. Religated DNA was used directly in an inverse PCR using the primer pair: 5′-GCATTTGAGATTGAGACAAGCACTCTCCACAC-3′ and 5′-ACAGCGCTAGACAGAAAGCCTGCTTGCT-3′. DNA sequencing revealed 228 bp of unknown sequence that was shown by BLAST analysis to originate from a region of chromosome 1, ∼596 kb telomeric from the breakpoint. PCR amplification with primers designed either side of the breakpoint (5′-TGGTCTGGAGAGACAGCTCGTCCAGAG-3′ and 5′-GAGTTGAAGAGTTGGCATAGTGGTGCACAC-3′) was employed to confirm the site of the deletion. To confirm the Dbf lesion was a true deletion, primers were designed to amplify three regions within the deletion predicted to contain polymorphisms variable between the C3H and H101 background strains. These regions were located at ∼150 kb intervals within the deleted region. The primer pairs used were site 1, 5′-GCCCTCATGCTTGAGTACCTTGCCTGTGAT-3′ and 5′-GTCCTCCCAGGGGCTGAGCAGAGTG-3′; site 2, 5′-TAGACTGAGCACCCGGCCTAACATGCTC-3′ and 5′-TGTGTCATCCACCCGGTGCCTCTGACT-3′; site 3, 5′-TAGAATTCCCACTGGGTCCACCCACTC-3′ and 5′-CATACATCCGTGTACATGTACTGACTGTCACTG-3′. Amplification products were digested with appropriate restriction endonucleases to discriminate between the alleles. Site 1 contained a novel TACC insertion polymorphism and was digested with HphI, site 2 contained a known C/T polymorphism (rs31657679) and was digested with AvaI and site 3 contained a known polymorphism (rs3049959) and was digested with Hpy8I. In each case only the H101 allele was present indicating that the C3H chromosome carried the deletion. The presence of both background strains was confirmed on the centromeric side of the Dbf deletion by sequence polymorphisms observed during Southern blotting (data not shown). Both strain backgrounds were shown to be present on the telomeric side of the deletion by AseI restriction digest of a fragment containing a novel informative C/T polymorphism which was amplified using the primer pair 5′-CAACAAAGCCCACATCAATTCACTCAGGCCGTG-3′ and 5′-CACCCTGCCTCAACCTCTCACCTGCTAG-3′.

Eur_Radiol-3-1-2077918

Multi-detector row computed tomography angiography of peripheral arterial disease

With the introduction of multi-detector row computed tomography (MDCT), scan speed and image quality has improved considerably. Since the longitudinal coverage is no longer a limitation, multi-detector row computed tomography angiography (MDCTA) is increasingly used to depict the peripheral arterial runoff. Hence, it is important to know the advantages and limitations of this new non-invasive alternative for the reference test, digital subtraction angiography. Optimization of the acquisition parameters and the contrast delivery is important to achieve a reliable enhancement of the entire arterial runoff in patients with peripheral arterial disease (PAD) using fast CT scanners. The purpose of this review is to discuss the different scanning and injection protocols using 4-, 16-, and 64-detector row CT scanners, to propose effective methods to evaluate and to present large data sets, to discuss its clinical value and major limitations, and to review the literature on the validity, reliability, and cost-effectiveness of multi-detector row CT in the evaluation of PAD. Introduction Before multi-detector row CT (MDCT) technology was available, the evaluation of peripheral arterial disease (PAD) using CT was restricted to imaging only a portion of the peripheral arterial tree [1–8]. With the introduction of four-detector row CT (4D-CT) in 1998, this major limitation was overcome. A complete coverage of the lower extremity inflow and runoff arteries was possible with one acquisition using a single-contrast bolus. With the launch of the 16-detector row CT (16D-CT), the spatial resolution increased to near isotropic voxels and the contrast medium efficiency improved [9–11]. True isotropic high spatial resolution of the entire volume was possible using the 64-detector row CT (64D-CT) scanner. In addition, improved X-ray tube capacity and scan speed allow submillimeter acquisition of a large coverage without limitations. These developments made multi-detector row CT angiography (MDCTA) an accurate alternative for the assessment of the peripheral arteries [12–25]. Using standardized scanning and reviewing protocols, peripheral CT angiography is a robust non-invasive technique for evaluating chronic and acute disease of the peripheral arteries. We present a review concerning our experience with 4-, 16-, and 64-detector row CT scanners in patients with PAD. Technique Preparation There are no specific prescanning preparations necessary for MDCTA of the peripheral arteries. The patient is placed comfortably to avoid movement, in the supine position with raised arms on the CT table. The legs are stabilized with cushions around the legs and slightly strapped with adhesive tape distally. It is important that the patient does not wear metal zippers or buttons on their clothing, since this can have a negative influence on the image quality, especially when using postprocessed images. Oral contrast should not be used, as this complicates postprocessing display (Table 1). Contrast material needs to be administered at body temperature to decrease the viscosity. The protocol can be completely programmed into the scanner. Table 1Prescanning preparationParameterDescriptionClothingNo metal parts on clothingOral contrastNoneI.V. cannula antecubitalMinimally 22 G (0.6 mm inner diameter, blue valve)PositioningSupine, stabilized and lightly strapped, feet-first and arms elevatedRespiratory phaseInspiration during abdominal-pelvic range Technical parameters The main challenge for peripheral CT angiography is the great range of the vascular system that needs to be depicted. Using a scanogram of approximately 1,500-mm length, the coverage of the acquisition is planned from the celiac trunk (T12 vertebral body) to the level of the talus using 4D-CT, or to the level of the feet using 16D-CT or higher (Fig. 1). Fig. 1(a) Scout image with three planned reconstruction batches of the abdomen, the upper legs, and the lower legs to preserve postprocessed image resolution. The frames 3-1, 3-2, and 3-3 depict the field of view of the three data sets, which need to be as narrow as possible to optimize pixel size. Whole-body volume maximum intensity projection (MIP) images after semiautomated bone removal of the abdominal data set (b), the femoral data set (c), and the crural data set (d) Scan duration The optimal scan duration for peripheral CT angiography varies between approximately 20 to 40 s, depending on the number of detector rows and the collimation (Table 2). The velocity of a contrast bolus to travel from the aorta to the popliteal arteries varies from 29 to 177 mm/s in patients with PAD [26]. This large variability is unpredictable and does not correspond to the severity of PAD. Based on these bolus travel times, it is recommended to limit the maximum table speed on faster scanners to 30 mm/s to avoid outrunning the bolus, leading to poor distal vessel opacification. This can be obtained, for example, by limiting the gantry rotation speed from 0.33 to 0.5 rotations per second or reducing the pitch (Table 2). Moreover, it is advised to program a second acquisition protocol into the scanner to start immediately if delayed distal enhancement is detected (Fig. 2). Because the time of the contrast bolus to travel from the aorta to the ankles varies from 7 to 40 s, a longer scan duration increases the risk of venous contamination, especially when there is critical ischemia and inflammation [26, 27]. Nevertheless, the discrimination of the arteries from the veins is often possible due to the stronger arterial enhancement and the anatomic 3D information [12]. Table 2Acquisition parameters for various multi-detector row computed tomography (MDCT) configurations for the angiography of peripheral arteriesType of scannerSection collimation width (mm)bRotation time (s)PitchcTable feed (mm/ rotation)Table speed (mm/s)Scan duration (s) dCharacteristics4D-CTa4 × 2.50.51.5153040Slow scan protocol, thick minimal slice width16D-CTa16 × 0.750.51.3153040Slow scan protocol, high resolution16 × 1.50.50.7173435Slow scan protocol, less resolution, better in obese patients16 × 1.50.51.0244825Fast scan protocol, less resolution, reduction of contrast media64D-CTa2 × 32 × 0.60.50.8153040Slow scan protocol, high resolution, isotropic voxel, double z-sampling, scanning of obese patients possible2 × 32 × 0.60.330.8164825Fast scan protocol, reduction of contrast media, high resolution, isotropic voxel, double z-sampling, scanning of obese patients possible2 × 32 × 0.60.331.019.86020Fast scan protocol, reduction of contrast media, high resolution, isotropic voxel, double z-sampling, scanning of obese patients possible, risk of outrunning the bolusaProtocols designed for Siemens CT scanners (Siemens Medical Systems, Erlangen, Germany) and should be modified appropriately for other models and manufacturersbValues are number of sections times section widthcPitch as the ratio of the table feed per rotation over the total width of the collimated beamdScan times representing a scanned range of 120 cmFig. 2Images from the first and second delayed acquisition of a 37-year old male with blue toe syndrome of the left hallux. (a and b) VRT images of the first acquisition show in the aneurysmatic abdominal aorta a short occlusion of the left femoral artery (white arrows) due to thrombo-embolism and an occlusion of the entire right superficial femoral artery. The anterior tibial arteries seem occluded in both legs. (c and d) VRT image of the feet with the first and the delayed second acquisition. The first acquisition (c) shows that the arteries of the feet are not enhanced yet due to slow flow (asterisk). The delayed acquisition (d) shows that both dorsal pedal arteries are patent and that the proximal arcuate artery and the first dorsal metatarsal artery (black arrows) of the left foot are occluded due to thrombo-embolism Using recent MDCT scanners, fast scans (25 s or less) can be performed of the peripheral arteries to reduce the amount of contrast media. To allow fast scan speed using 16-detector row CT (16D-CT) scanners, a wider collimation must be used (Table 2). The 64-detector row CT (64D-CT) even allows to perform fast scans while maintaining submillimeter collimation. However, to ensure distal opacification, the scanning delay must be increased appropriately in fast scans. Another difficulty of a fast scan is that there is a greater risk of asymmetric enhancement in patients with severe unilateral vascular disease. Therefore, it is safer to choose a slower scan speed. Collimation To aim for maximal spatial resolution, a thin section collimation width allows a narrow effective slice width. Furthermore, the partial volume effect and blooming effect of calcium will be reduced (Fig. 3) [10]. The collimation should be chosen as narrow as possible but still allowing for a table speed of 30 mm/s and depends on the number of detector-rows and heat capacity. On a 4-detector row CT (4D-CT), the collimation is limited to 4 × 2.5 mm, whereas the 16D-CT and 64-detector row CT (64D-CT) allow a submillimeter collimation of 16 × 0.75 mm and 32 × 2 × 0.6 mm, respectively. Fig. 3a, bImages of 16-detector row CT (16D-CT) acquired with a collimation of 0.75 mm showing the effect of slice width (SW) on the blooming of the arterial wall calcifications. (a) Reconstructed axial image of the right external and internal iliac artery with SW of 3.0 mm using a B46 reconstruction kernel shows more blooming of the calcifications than (b). (b) Reconstructed axial image of the right external and internal iliac artery with SW of 0.75 mm using a B46 reconstruction kernel with less blooming of calcifications Using 16D-CT in obese patients, the thin collimation protocol leads to unacceptable noise levels in the abdomen and pelvis because the tube is unable to deliver the necessary dose in this submillimeter configuration. In order to enable the tube to deliver a higher dose, a wider collimation (16 × 1.5 mm) with a reduced pitch factor of 0.7 (Table 2) is used to improve the image quality in obese patients. For the 64D-CT scanner, there is no longer a tradeoff between resolution and scan speed, and it allows, even in obese patients, a fast submillimeter scan protocol. Contrast injection It is important in peripheral CT angiography to obtain a high and homogenous enhancement of the arterial tree and to synchronize the acquisition with the enhancement. The optimization of acquisition timing and contrast medium delivery is essential for vascular assessment and image postprocessing. Normally, attenuation values higher than 200 HU in the arteries is considered suitable in MDCTA [12, 13]. For the intravenous injection of contrast medium in the antecubital vein, 22- and 20-gauge intravenous cannulas are needed for the maximal flow rates of 3.5 and 5.0 mL/s, respectively. Acquisition timing Due to the interindividual hemodynamic variability in peripheral CT angiography, reliable timing techniques are preferred over using a fixed delay. The test-bolus technique relies on the dynamic monitoring of small contrast boluses to measure the contrast arrival and travel time at the proximal and distal arteries, respectively. The bolus-triggering technique is a commonly used timing technique that is based on repetitive low-dose sequential scans at the level of the abdominal aorta, to monitor the arrival time of the contrast media. The acquisition starts automatically when the preferred threshold is reached, approximately 100 to 150 HU above the baseline value. During a transition delay, which is the time needed for the table to move and start the scan, of approximately 4 s, breathing instructions can be given to the patient. During this delay, the enhancement of the aorta will further increase to an absolute value of more than 200 HU. For a fast scan protocol, an extra delay must be added to the contrast arrival time to ensure distal arterial opacification [28]. This extra delay can be calculated as 35 s minus the scanning time. Thus, for a scan time of 25 s, a extra delay of 10 s. must be added. Another option is to monitor at the proximal level of the popliteal artery and to start the scan manually when enhancement is visualized. Consequently, the time of contrast arrival increases by approximately 8 s [26, 27] and the transition delay of the scanner increases to 11 s to travel from the knees to the diaphragm and then starting the acquisition. Contrast injection The volume of contrast material ranges from 120 to 160 ml for a typical scan duration of 40 s. The amount of contrast media depends on the scan duration and on the flow rate. Because the last volume of the bolus will not contribute to the enhancement when scanning below the knees, the injection duration can be shortened by 5 s, e.g., a 35-s injection time is used for an acquisition of 40 s. However, to ensure the enhancement of all arteries, the injection duration should not be shorter than 30 s and in fast scan protocols, a delay time needs to be added appropriately to prevent outrunning the contrast bolus. A flow rate of 3 to 4 ml/s is necessary for adequate arterial enhancement [12]. This corresponds to an iodine administration rate of 1.0 to 1.4 g/s using a contrast media concentration of approximately 320 to 350 mg I/mL. Based on the reported literature the average values of contrast media volume, concentration, injection rate, and administration rate are 134 ml, 341 mg I/mL, 3.5 ml/s, and 1.2 g/s, respectively [9–25, 29–36]. By increasing the iodine concentration to a concentration of 400 mg I /mL, the iodine administration rate can be increased to 1.6 g/s to increase the enhancement [37]. To optimize the enhancement, 20 to 60 mL of saline is injected immediately after the contrast media. A tighter bolus can be obtained to increase the attenuation. Using a monophasic injection rate, the arterial enhancement increases over time to decrease at the end of the bolus. Consequently, the Hounsfield values of the enhanced arteries start lower at the level of the aorta and increase at the level of the popliteal artery to the highest attenuation value, and, subsequently, decrease distally in the runoff arteries, especially for longer scan durations [26]. A more homogenous enhancement can be achieved using a biphasic injection rate using a higher rate (5–6 ml/s) at the beginning (during the first 5 s) of the injection and a lower rate (3 ml/s) for the remaining volume. In clinical practice, a monophasic injection rate is often used because it is a simple method and has resulted in adequate image quality [37]. Patient dose in MDCT A particular concern with MDCT scanners is delivering potentially higher radiation doses. To maintain the noise level in submillimeter slices, the dose needs to increase proportionally. On the other hand, with the increasing number of detector rows, the z-axis efficiency improves, since the overbeamed area decreases. Current MDCT scanners present an indication of patient dose on the scanner console for dose awareness and to help optimize the scan protocol. Useful in CT angiography is that, when reducing the X-ray energy, the contrast-to-noise ratio increases. Compared to a standard scan with 120 kVp, selecting 100 kVp, results in a dose saving of approximately 40% [38–40]. Furthermore, dose reduction can be achieved by decreasing the tube current using automatic tube current modulation. With angular tube current modulation, the tube current varies during the course of a rotation. The changing attenuation through different projections around the patient (e.g., at the level of the pelvis) can be used, to reduce unnecessary x-rays in the anterior–posterior projection without any substantial effect on image quality [9, 41, 42]. With longitudinal tube current modulation, the tube current varies along the z-axis based on the size, shape, and attenuation to maintain a predefined noise ratio. Compared with constant tube current, this technique results in acceptable image noise and a dose reduction of 20% or more without compromising diagnostic image quality [9, 41, 42]. The average patient dose reported in the literature in the assessment of PAD with CT angiography is 7.47 mSv [9, 12, 24, 31, 43]. The radiation risk from these doses is not a major concern in patients with PAD. Their life expectancy is shorter than the latency period of a radiation-induced fatal malignancy [44–46]. Display and evaluation Image reconstruction The raw data set is reconstructed using an increment with 50% to 70% overlap. Peripheral CT angiography generates more than 1,500 axial images, depending on slice width and reconstruction increment. It is recommended to reconstruct separate data sets. Routinely, we calculate three separate data sets of the peripheral runoff (Fig. 1). The first advantage is that it allows us to reconstruct thicker slices, e.g., of 1.5 mm for the abdominal and femoral data set, and thinner slices for the crural data set to optimize the resolution and to minimize the data load [9, 10, 16, 20, 23]. Secondly, longitudinal images that are calculated from the entire data set have a decreased resolution, due to the limited display matrix (e.g., 512 × 512) [32]. Images that are reconstructed from the separate data sets will preserve the initial longitudinal resolution. A smooth kernel (B20 for Siemens CT scanners) is generally used in CT angiography and leads to an accurate depiction of the diameter of the vessels and is very appropriate for postprocessing. A sharp kernel (B46) is used when stents or severe vessel wall calcifications are present, as it minimizes the blooming effect at the cost of some increase in the noise level [47]. The field of view (FOV) is selected as small as possible to optimize pixel size. A FOV of 380 mm, 350 mm, and 300 mm for the abdominal, femoral, and crural data sets, results in pixel sizes of approximately 0.74 mm, 0.68 mm, and 0.58 mm, respectively. Also, the FOV can be further decreased to 200 mm by including only one leg, leading to a pixel size of 0.4 mm. Advanced postprocessing and image evaluation Additional two-dimensional (2D) and three-dimensional (3D) postprocessing techniques are required to facilitate interpretation and presentation. Reviewing exclusively the transverse images is inefficient and less accurate than reviewing a combination of reformatted images. To preserve the study quality for clinical decision making, a standard set of postprocessed images needs to be included in the protocol. These include thin-slab maximum-intensity projections (MIPs) through visceral and renal arteries and the abdominal aorta, through femoropopliteal arteries, and through crural arteries (Fig. 4); whole-volume MIPs of the separate data sets after bone removal (Fig. 1) and when necessary, after removal of vessel wall calcifications (Fig. 5); and curved planar reformations (CPRs), e.g., through the iliac arteries. Volume-rendered (VR) images are fast and effortless created to present the pathology to clinicians, who normally do not have the possibility to review the data set interactively. Fig. 4a–dStandard slab MIP images in the postprocessing protocol make MDCTA of peripheral arteries on a routine basis feasible. Slab MIPs are easy and fast to create and to evaluate. The images depict the vasculature without superimposing bones (a and b). From the abdominal data set, MIP images are created in coronal projection to depict the renal arteries (b) and in sagittal projection to depict the celiac trunk and mesenteric arteries (d). The aorta is also depicted for evaluation. (c) Standard coronal slab MIPs from the data set of the upper legs are created, which are parallel to the superficial femoral and popliteal artery. (d) Standard coronal slab MIPS from the data set of the lower legs display the crural arteriesFig. 5a–dVolume MIP images in anteroposterior projection show the result of three different threshold levels used for the segmentation of arterial wall calcifications. (a) Volume MIP before removal of the calcifications shows that the lumen is not visible. (b) Volume MIP after removal of the calcifications shows that, still, many voxels of calcification are present, hampering lumen assessment (arrows). (c) Volume MIP shows angiogram after calcium segmentation using a correct threshold level allowing lumen assessment. The rest of the voxels of the burden of calcifications are just visible as unesthetical noise, which is, however, preferable to introducing pseudo-stenoses (d) (arrows) by using a too low threshold level The data sets are reviewed effectively by evaluating the standard set of postprocessed images and, interactively, exploring the data set using multiplanar reformations (MPRs) . The transverse images (or true cross-sectional images) need to be considered to verify diseased segments [4, 6, 19]. When extensive calcifications or stents are present, the vessel lumen visibility and the clinicians’ confidence in the CT images will decrease [25]. In whole-volume MIPs, superimposing calcifications can be selected to be removed digitally from the data set using thresholding and region-growing techniques (Fig. 6). However, the removal of the numerous arterial wall calcifications can be very time consuming. Another limitation of these segmentation techniques is that readers should be aware of artificial stenoses and occlusions. These can be introduced when voxels that represent lumen are inadvertently removed when in close contact with the bones (Figs. 7 and 8) or when a too low threshold value is used (Fig. 5). In addition, in VR images, the lumen is also obscured by vessel wall calcifications and, as a result, should not be used for the lumen assessment (Fig. 6). A more reliable technique for stenosis detection in extensive calcified arteries is CPR, which displays the lumen as a longitudinal cross-section (Fig. 6). When using an application that semi-automatically traces the vessel lumen, the risk of an inaccurately positioned central lumen line is minimized. The CPR projection should include at least two perpendicular longitudinal projections and true cross-sectional images can be viewed for lumen assessment [19]. Software tools are available for automatic quantitative evaluation of the traced lumen and to generate a graphical presentation of luminal diameter (Fig. 9). Multipath CPRs are under development and could enhance image evaluation. Fig. 6a–dInfluence of vessel wall calcifications on postprocessed images and the ability for lumen assessment. (a) VRT image (medial view) of right femoropopliteal segment showing arterial wall calcification; does not allow luminal assessment. (b) CPR image (anteroposterior view) shows the interior of blood vessels as a longitudinal cross-section, even in the presence of the arterial wall calcifications. This is the preferred imaging technique when extensive calcifications of the vessel wall are present. Volume MIP (anteroposterior view) after bone removal using region-growing and threshold techniques (c) does not allow lumen evaluation. Volume MIP after additional calcification removal (d) removes superimposing calcification to enable lumen evaluationFig. 7a–cImages of segmentation artifacts due to bone removal in 16D-CTA. (a) Volume MIP after bone segmentation of the lower legs showing a pseudo-occlusion of both distal anterior crural arteries (arrows), which is caused by segmentation of the bones. (b) MIP of the lower legs showing the anterior tibial arteries in close proximity to the tibia (arrows), which is the cause of the false positive pseudo-occlusion. (c) Axial image of the lower legs just caudal from the pseudo-occlusion, showing the patency of both anterior tibial arteries of both legs in close proximity of the tibia (arrows)Fig. 8a–cApplying blue color to the voxels selected for removal helps to identify the sites of segmentation artifacts in VRT images. (a) VRT image before bone segmentation of the lower legs showing patent proximal anterior tibial arteries. (b) VRT with blue bones to indicate the voxels to be removed shows the voxels of the bone which are in contact with the proximal anterior tibial artery are not selected for removal and shows the voxels of the artery which are selected for removal. (c) Segmented VRT image showing the pseudo-occlusion of the anterior tibial arteryFig. 9a, bResults of semiautomated quantitative lumen assessment in aortoiliac arteries of a patient with in stent thrombosis. (a) Graph (upper section) displaying the maximum and minimum diameters of the lumen to quantify stenosis. CPR (lower section) through the aortoiliac arteries, which can be rotated around its longitudinal axis, depicts the luminal obstruction (asterisk) due to a thrombus inside an iliac stent. (b) Corresponding transverse image confirms the occlusion of the iliac stent Wall calcification problem The depiction of vessel wall calcifications using MDCT can be valuable, since severely calcified arteries may have consequences for bypass surgery. On the other hand, these wall calcifications are known to hamper the assessment of the lumen [2, 10, 14, 19]. Approximately 20% to 50% of the vascular segments contain wall calcifications, of which, 10% severely calcified [11, 19]. Patients with a history of diabetes mellitus, cardiac disease, or elderly age are very likely to have extensive calcifications [48]. Furthermore, we found that patients with Fontaine stage III/IV have more infrapopliteal arterial wall calcifications compared to stage IIb. How can we deal with the vessel wall calcifications depicted with MDCTA? It is important to use a wider window width (WW) and higher window center (WC) level settings from the usual CT angiography level of around 150 WC ± 250 WW to 200 WC ± 1000 WW for a better differentiation of calcifications and stents from the enhanced lumen and to minimize the effect of blooming. A further minimization of blooming is reached by using a sharper reconstruction kernel and higher spatial resolution. Especially in MIP images, the lumen is hidden by the circumferential calcifications. In these circumstances, transverse images, CPR images, and the digital removal of the calcifications help to depict the lumen, at least for the larger arteries. Despite all of the available tools, in particular in the smaller crural arteries, the concentric calcifications still hamper lumen assessment [11]. Recent publications showed that a subtraction technique using two acquisitions is feasible in some patients with PAD using MDCTA [24, 49]. In the near future, automated 3D applications could help to minimize the impediment of the calcifications [50]. Whether dual-energy CT angiography can improve this limitation of CT needs to be evaluated. Clinical value Because MDCT angiography for the imaging of the peripheral arteries is a rather new non-invasive technique, there are a small number of studies published on its performance and reproducibility (Table 3). The majority report on 4D-CT; two authors report on 16D-CT. There are no reports of the assessment of PAD using 64D-CTA. In our meta-analysis soon appearing in Radiology, which included 436 patients and 9,541 arterial segments, a pooled sensitivity and specificity for detecting a >50% stenosis of 92% and 93% was estimated, respectively. Table 3Validity of CT angiography in peripheral arterial disease (PAD)AuthoraNo. of patientsNo. of analyzed segmentsNo. of detectorsReported sensitivity (%)eReported specificity (%)eAssessed segmentsStenosis category (%)hRichter et al. 199432ns184nsIliac>50Lawrence et al. 1995613419396Femorocrural>50Raptopoulos et al. 19963962419396Aortoiliac85–99Rieker et al. 199650400173–88b94–100bFemorocrural75–99Rieker et al. 19973021019399Aortoiliac75–99Kramer et al. 199810ns294nsIliocrural>90–99Ishikawa et al. 199949ns19795Bypass graftsnsBourlet et al. 20002231819590Aortoiliac>50Puls et al. 20013118648986Total tree50–99Willman et al. 20034676949199Aortoiliac graftsnsOfer et al. 20031841049192Total tree>50Heuschmid et al. 200318568491c92cTotal tree>50Martin et al. 2003411,31249297Total tree75–99Catalano et al. 2004501,14849693Total tree>50Mesurolle et al. 20041616829193Total tree>50Ota et al. 20042447049999Total tree>50Poletti et al. 2004d12144482/96gnsns>50Portugaller et al. 20045074049283Total treearea >70Romano et al. 2004423,40249395Total treensRomano et al. 2004221,78249294Total treensStueckle et al. 200452ns482100Total treensEdwards et al. 2005441,02447993Total tree50–99Fraioli et al. 2006751,425496–99h94–96hTotal tree50–99Schertler et al. 200517170169690Popliteocrural>50Willmann et al. 2005391,365169696Total tree>50Unpooled mean9194aBased on references [1–6, 8–10, 13–24, 31, 34, 35, 56, 57]bFor various anatomic levelscCalculated from the datadBased only on subtracted MDCTA images, the positive predictive value was 95%eSensitivity as published or calculated overall mean.fDiameter stenosis is mentioned unless specified (>50 means stenosis more than 50% including occlusion)gFor subtracted and nonsubtracted segments, respectivelyhDepending on the MDCTA protocol with varying mAs Publications on the reproducibility of CT angiography reported a good intertest agreement between MDCTA and DSA (Table 4) and a good to excellent interobserver agreement for 4D-CTA [12, 21, 23, 35] and 16D-CTA (Table 5) [9, 11]. A few studies provide stratified data on the aortoiliac, femoropopliteal, and crural tract and show that the accuracy and reproducibility of the crural tract is lower than for the aortoiliac and femoropopliteal tracts [9, 11, 20, 22, 23]. Table 4Intertest agreement between CT angiography and digital subtraction angiography in PADAuthoraNo. of patientsNo. of assessed segmentsNo. of detectorsReported intertest agreementdAssessed segmentsRaptopoulos et al. 199639624190%AortoiliacBeregi et al. 199720521100%PoplitealTins et al. 200135219184%AortoiliacWalter et al. 2001224564κ = 0.68 (0.50–0.97)cTotal treeRubin et al. 2001183514100%Total treeHeuschmid et al. 2003231,136486%Total treeOfer et al. 200318444478%Total treeRomano et al. 2004423,4024κ = 0.68; 90%Total treeRomano et al. 2004221,7824κ = 0.68; 90%Total treeaBased on references [3, 7, 12, 14, 16–18, 35, 55]bBased on 97% of the segmentscAverage of the reported kappa values (ranges) of the individual anatomical segmentsdAn unweighted kappa statistic (κ) is reported for percentage agreementTable 5Interobserver agreement of CT angiography in PADAuthoraNo. of patientsNo. of analyzed segmentsNo. of detectorsReported interobserver agreementbAssessed segmentsRieker et al. 1997302101ρ=0.95AortofemoralWalter et al. 2001224564κ = 0.71–0.76cTotal treeTins et al. 200135219178%AortofemoralMartin et al. 2003411,3124κw = 0.84Total treeRomano et al. 2004423,4024κ = 0.84;0.86dTotal treeRomano et al. 2004421,7824κ = 0.85, 0.88, κ = 0.80eTotal treeCatalano et al. 2004501,1374κ = 0.80Total treeOta et al. 2004244704κ = 0.88IliacPortugaller et al. 2004507404κ = 0.81Total treeKock et al.f732,2684κw = 0.84Total treeOuwendijk et al. 2005792,41916κw = 0.85Total treeWillmann et al. 2005391,36516κ = 0.85–1Total treeaBased on references [4, 7, 9, 11, 15, 17–19, 21, 23, 35]bAn unweighted kappa statistic (κ) is reported, unless indicated (κw=weighted kappa statistic; ρ=intraclass agreement coefficients as a measure of agreement for ordinal or quantitative data). A linear weighting was used, except in one paper [15], where a quadratic weighting was usedcRange of kappa values of the individual anatomical segmentsdFor reader one and two, respectivelyeFor intraobserver (two readers) and interobserver agreement, respectivelyfBased on unpublished data MDCTA leads to adequate decision making for treatment recommendations concerning both the anatomical level and the technique of revascularization [51]. A cost-effectiveness study showed that MDCTA is a cost-effective diagnostic strategy in the work-up of PAD [52, 53]. Randomized controlled trials confirmed that MDCTA in PAD is the optimal diagnostic imaging technique [25, 54] and reduces the diagnostic costs when compared to DSA and CEMRA with comparable clinical utility and patient outcomes. Besides these evidence-based results, local expertise and availability also define which modality to use in clinical practice (Table 6). Table 6Advantages and limitations of multi-detector row CT angiography (MDCTA), contrast enhanced MR angiography (CEMRA), and digital subtraction angiography (DSA) MDCTACEMRADSAIntermittent claudication (Fontaine II)+++Chronic critical ischemia (Fontaine III or IV)−++Short examination time+−−Short postprocessing time−++Outpatient setting++−Availability+−+Non-invasive technique/patient comfortb++−Low diagnostic imaging costs+−−Contrast media tolerance−+−Three-dimensional imaging++−Non-interference of stentsc+−+Radiation risk+(−)d−+(−)dAcute clinical setting+−+Hemodynamic assessment−−a+Extraluminal pathology visualization+−a−aIs only possible when using additional sequencesbFrom [58]cFrom [59]dNegligible risk in population with chronic obstructive PAD It is reported that arterial wall calcifications lead to false-positive interpretations and a decreased reproducibility in reading MDCTA [2, 14, 11]. We have to acknowledge this limitation with current technology. A preferential indication for MDCTA in patients with intermittent claudication (Fontaine stage IIb) is clearly justified. However, patients with critical limb ischemia (Fontaine stage III/IV), who are likely to have extensive calcifications of the smaller arteries, could be better off undergoing contrast-enhanced magnetic resonance angiography (CEMRA) or digital subtraction angiography (DSA). Finally, MDCTA is an accurate technique to evaluate the patency after revascularization procedures [43]. The technique can be used in the evaluation of acute ischemia, e.g., after a revascularization procedure or in thrombo-embolic disease (Fig. 10). For aneurysmatic popliteal artery disease or entrapment syndromes of the popliteal artery, MDCTA is the preferred imaging modality (Fig. 11) [55]. Fig. 10a, bAcute thrombosis of the crural arteries in a 53-year-old woman with an acutely cold left leg after stopping anticoagulation therapy. The patient refused angiography. (a) VRT image (posteroanterior view) of MDCTA at the level of the crural arteries shows abrupt stoppage of arterial opacification in the left peroneal, anterior, and posterior tibial artery (arrows). The contralateral right crural arteries are patent. (b) Selective anterograde DSA image (posteroanterior view) confirms the occlusions of the three left crural arteries (arrows) due to thrombo-embolismsFig. 11a–dA 56-year old male patient who had a history of deep venous thrombosis with intermittent claudication of the right lower extremity. (a and b) Thin MIP image shows an aneurysmatic right popliteal artery with a tight stenosis distally. (c) VRT and volume MIP (d) confirm these findings and show patent proximal crural arteries Conclusion Multi-detector row CT angiography (MDCTA) is an outstanding non-invasive imaging test in the evaluation of patients with peripheral arterial disease (PAD) and is currently the modality of choice in patients with intermittent claudication. The technique can be used in the evaluation of patency after revascularization procedures and in acute ischemia. MDCTA has been shown to have high diagnostic performance and reproducibility in evaluating peripheral arterial disease (PAD). MDCTA reduces diagnostic costs and provides adequate information for decision making. The most important drawback is the limited lumen evaluation of extensive calcified arteries. MDCTA appears to be clinically less valuable in critical limb ischemia because of extensive crural artery calcifications.

Graefes_Arch_Clin_Exp_Ophthalmol-4-1-2206250

Ophthalmologists, suicide bombings and getting it right in the emergency department

Background The number and extent of worldwide suicide attacks has risen sharply in recent years. The objectives of this retrospective study are: to determine the prevalence and outcome of the victims who sustained ocular injury, to describe the activities of ophthalmologists in the setting of an emergency department (ED) receiving mass casualties of a suicide bombing attack and to illustrate some of the treatment obstacles that they encountered and the protocol. The number and extent of worldwide suicide attacks has risen sharply in recent years [1, 2]. The perpetrators typically mingle among crowds of civilians and detonate an explosive device that is usually strapped on their bodies with the intent of sacrificing their own lives in order to cause the death of as many others as possible. The injuries sustained by survivors of these well-planned attacks combine the lethal effects of penetrating trauma, blast injury, and burns [3]. Suicide attacks on civilians were historically confined to a limited number of countries, but the outrageous and devastating destruction of the Twin Towers in New York City on 11 September 2001 and the bombings in London and Madrid established the universality of such terrorism. With the rise in terror-related activities in urban settings, ophthalmologists worldwide may find themselves treating ocular trauma under conditions unlike any they had experienced before, and certainly remote from the relatively orderly setting of an emergency room in which traumatic ocular injuries can usually be counted on one hand at any given time. Since the beginning of the latest Israeli-Palestinian Intifada on 29 September 2000, more than 4,600 people—mostly civilians—have been killed or injured by suicide bombings in that area [4]. The Tel Aviv Sourasky Medical Center (TASMC) is the largest hospital in the Tel Aviv metropolitan area. As such, most of the victims of the suicide bombing attacks in its catchment area are evacuated to TASMC, and a special treatment protocol has evolved for coping with the unusual logistic needs of such events. The objectives of this retrospective study are: to describe the activities of ophthalmologists in the setting of an ED receiving mass casualties of a suicide bombing attack, to illustrate some of the treatment obstacles that they encountered and the protocol that evolved for overcoming them, and to determine the prevalence and outcome of the victims who sustained ocular injury. Methods IRB approval was obtained for this retrospective interventional case series study. Clinical data on all casualties evacuated to the TASMC due to suicide bombing-related injuries were collected from the trauma registry records and reviewed. Their demographic data were obtained from the main admitting office records. The senior surgeon stationed at the ED entrance is rapidly provided essential information on the type/location of injuries from the arriving ambulance’s paramedical personnel. He/she pages the designated on-site specialist according to prioritization for urgent management. Ocular injuries are defined as any blunt, penetrating, or perforating trauma or blast-related damage to the eye, orbit, or ocular adnexa. The terms we used to describe ocular injuries conform to the recommendations the United States Eye Injury Registry and the International Society of Ocular Trauma [5]. Individuals with nonpenetrating or non-lacerating ocular injuries, non-penetrating debris in their eyelids, or superficial burns of the eyelids were excluded from this study, as were victims who died of their injuries before undergoing ophthalmologic treatment for whatever ocular damage had been suffered. The ocular injuries for each eye were categorized according to type: it was possible to have multiple occurrences of the same type (i.e., multiple corneal lacerations) and of several types (i.e., corneal laceration and retinal detachment) in the same eye. All records of ocular- and orbital-related trauma that were documented in the trauma registry were collected and analyzed together with hospital and outpatient clinic records. The analysis included age, gender, mechanism of injury, anatomic site of injury, Injury Severity Score (ISS) [6], length of stay, length of intensive care unit stay, and surgical procedures. Ophthalmic information included the initially diagnosed ocular condition, all surgeries performed during the ED stay and afterwards, and final ocular test findings. The data were entered using Excel spreadsheet (Microsoft Office) and a simple descriptive statistical analysis was performed. Results There were 13 suicide bombing attacks in the Tel Aviv metropolitan area between October 2000 and October 2004. A total of 352 patients were evacuated to the TASMC ED, and 198 of them were hospitalized. The other 154 patients suffered from minor injury or shock for which they were given appropriate treatment and instruction and sent home. The overall severity of suicide bomb-related trauma was very high: the mortality rate was 8.4% when the attack occurred in open spaces, 15.5% in closed spaces and 20.3% when the bomb exploded inside a bus. The ISS was 1–14 for 74% of the patients (non-hospitalized) and ≥16 for the remaining 26% of the patients (admitted to hospital). One of the prominent hallmarks of suicide bombing injuries is the extremely high prevalence of head injuries: among our patients, 49% suffered from head and neck injuries, 9% head and extremity injuries, 4% head and torso injuries, and 38% torso and other injuries (all data are taken from the experience with the 13 bombings in Tel Aviv). Seventeen patients (4.8%) were listed in the trauma database as having any ocular or periocular trauma, and several had more than one type of injury. The types of recorded trauma were: open globe injuries (n=7), closed globe injury consisting of severe subconjunctival hemorrhage (n=2), partial thickness lamellar laceration of cornea (n=8) (lamellar flap) of which five were burn-related and three were due to small foreign bodies, and extraocular injuries (n=6), which included three orbital fractures due to primary blast injury and three eyelid lacerations. Primary repair of open globe was performed in six eyes that underwent primary closure of laceration. One patient who was diagnosed during the initial triage as suffering from open globe injury died during the initial trauma surgery, thus no ocular procedure was performed. Two other eyes underwent primary exploration of subconjunctival hemorrhage that was suspected as being open globe due to massive subconjunctival: no laceration was found intraoperatively, and contusion was diagnosed. The one eye that was found to be unsalvageable underwent primary enucleation. After the initial eye surgery, two patients died from their other injuries within 24 h of the explosions. Of the eight patients with partial thickness lamellar laceration of cornea, three were discharged and given instructions to return as outpatients on the following day. Five other patients who required and received medical treatment for non-ocular-related medical problems were hospitalized and continued eye treatment as inpatients. All patients in which superficial burns were found (n=5) were treated by manual removal of corneal foreign body and antimicrobial drops, and their recovery was uneventful. Secondary surgical intervention was performed on four patients (all surviving patients who initially underwent primary closure of open globe) within 1 week of the initial trauma: large intravitreal foreign bodies were extracted in three of them, and an intra-lenticular foreign body was extracted from the fourth (Table 1). At final follow-up (≥2 years post-trauma), the visual acuity of the three patients who suffered from large intravitreal foreign bodies were finger counting (FM)-hand movement (HM). The silicone oil was removed in two of these patients and retained in the third; two of them are wearing large cosmetic contact lenses due to corneal opacities that were cosmetically disturbing. The patient who underwent surgery due to intra-lenticular foreign body and penetrating keratoplasty had final visual acuity of 6/12 (Table 1). Table 1Relevant data of suicide bombing survivors who sustained open globe injuryNo.Age, years/sexPrimary surgical interventionSecondary surgical interventionForeign body extractedFurther surgeryFinal outcomeFinal visual acuity117/fClosure of open globePatient died within 24 h of trauma216/fNonePatient died within 24 h of trauma362/mEnucleation417/fClosure of open globeLensectomy, vitrectomy, removal of foreign body, endolaser silicone oil injectionMetal shrapnell (5 mm)Silicone oil removalPreserved globe flat retinaFC518/fClosure of open globeLensectomy, vitrectomy, removal of foreign body, endolaser silicone oil injectionMetal ball (2 mm)Preserved globe flat retinaHM619/fClosure of open globeLensectomy, vitrectomy, removal of foreign body, endolaser silicone oil injectionGlass fragment (3 mm)Silicone oil removalPreserved globe flat retinaHM727/mClosure of corneal perforationPenetrating keratoplasty, lensectomy, vitrectomy endolaser, intraocular lens implantationGlass fragment (1 mm)6/12FC, finger counting; HM, hand movement Discussion The dynamics of an emergency receiving center of victims of a suicide bombing attack are alien to most ophthalmologists. Today, the setting is characterized by large numbers of victims who sustain injuries that are more complex and more severe than those that had occurred during earlier periods of terror activity [7, 8]. Importantly, suicide bombings are more likely to occur in closed spaces, unlike other mass trauma scenarios such as car bombs, train wrecks, and other outdoor explosions: over 62% of injuries that occur in closed spaces are to the face, head, and neck, thus posing a far greater risk to the ocular structures. In our experience, 17/352 (4.8%) of the survivors of suicide bombing attacks sustained eye injuries, and 9/17 (52%) required urgent attention. This rate is surprisingly lower when compared to previous reports, which documented that close to 10% of survivors of terrorist blasts have significant eye injuries [13]. We have no explanation for the low rate, but we can speculate that because of their close proximity to the hospital together with the efficiency of the Israeli Magen David (Red Cross), almost all survivors are speedily brought to the hospital for examination after terror-related episodes and are listed as admissions in the trauma records, even those with no complaints or only minor ones. Thus, the large number of admitted individuals artificially decreases the percentage of eye victims. Importantly, 41% of all of the reported ocular injuries [9–13] were severe in degree. The final visual outcome of all the surgeries we performed was poor: globe preservation was successfully achieved in most cases (6/7), but only one patient with an intra-lenticular foreign body had useful vision postoperatively. In order to provide the best treatment in such a complicated setting, special adaptations must be made to the treatment algorithm of the ophthalmology team. In terms of individual trauma cases, the victim of a suicide bombing attack is no different from any other eye trauma patient. The sudden presentation of large numbers of injured patients, however, presents two types of challenges: the logistical one of rapidly processing masses of casualties through the system and the medical one of providing the best possible trauma care to severely wounded patients [14]. According to our protocol, all patients who complain of eye symptoms and all unconscious patients who sustain head or face injuries must be checked by an ophthalmologist. This requires special disaster on-call lists of ophthalmologists who are able to arrive to the hospital on extremely short notice since, thanks to the highly efficient organization of our Red Shield ambulance facilities, blast victims usually arrive at the hospital within minutes and are hurried to either diagnostic tests or directly to operating theaters. Upon arrival to the ED, every victim of a terrorist attack is triaged by a senior surgeon who synchronizes the activities of the multifaceted operation. The ophthalmologists already present in the hospital and the ones on-call who arrive to the ED are in contact with that surgeon in order to expeditiously locate the victims with ocular injuries, examine them, and send them to the operation theaters, intensive care units, imaging studies or home. There is a directive in our department that all available staff members must contact the hospital immediately upon learning about any suicide bombing attack to check whether their services are required. In the event of large-scale attacks, they are prepared to be recruited to assist in triaging and in treating all the victims, not just those with ocular injuries. The triage procedure is the key to the successful management of large trauma events: the most important rule is that all patients must be checked by the ophthalmologist wherever they are located on the hospital premises. Trauma patients invariably require urgent treatment and some are sent directly from the ER to either imaging units or immediately to the trauma surgical unit. The senior ophthalmologist on the premises must contact trauma registration services, get a list of all admitted patients (usually assigned numbers upon admission) and make sure that each and every one of them is examined, even during emergency surgery or during imaging interventions for non-ophthalmological injuries. When an open globe is suspected, the eye is immediately patched, and the finding is reported to the surgeon in charge of the patient: the staff is instructed not to intervene in the treatment of the eye. Further evaluation is done only when it is certain that there is no danger of expulsive hemorrhage. Only patients suspected as suffering from open globe injury undergo urgent primary closure of the wound. Since that patient invariably presents with multiple injuries and may not be fit for transfer to the ophthalmology operating theater, however, special alterations to the surgical protocol may be required. For instance, no ophthalmological microscope is available in our trauma center because space is limited in the trauma room due to the concomitant performance of many surgeries and given the cumbersome structure of an ophthalmic microscope. There is, however, a high-quality neurosurgical microscope that has a long arm that can be placed at sufficient distance from the patient and the life-support machines so that it can be used without disturbing the anesthesiologists and other trauma teams as they work, and this microscope is used with great success during primary closure. Other eye surgery procedures are postponed, either until the patient has been stabilized or they are scheduled for a later date. This highlights the first critical responsibility of the ophthalmologist in the mass trauma setting, that of identifying which surgical procedures must be carried out immediately. The order of surgical intervention deserves special attention: due to the characteristic complexity of the injuries, most of the patients required multiple procedures immediately following the trauma. The established protocol adopted among our surgeons is (in descending order): trauma surgery (for life-threatening conditions, performed by either trauma surgeons or neurosurgeons), ophthalmologic interventions (immediate surgery or instructions for palliative care), and orthopedic and plastic surgery interventions. Finally, terrorist bombings present a danger to the ED staff members that is never associated with any other mass casualty situation: there is a very real chance of explosion by a second-hit, either by explosive material remaining on the perpetrator’s body, or, even more threatening, a second suicide bomber who infiltrates the ED disguised as one of the victims and detonates the bomb inside the crowded ED. Thus, a unique caveat in the ED protocol for terrorist bombing attacks is heightened vigilance, starting from the chaotic first minutes after the arrival of the victims. Providing medical assistance in an ED to victims of suicide bombing attacks is a harrowing experience: physicians who work in an urban hospital are more and more likely to be exposed to such events [14]. In his excellent editorial, Hirshberg wrote "Urban terrorism, the scourge of the 21st century, is already at our doorstep and surgeons are called upon to play leadership roles in shaping the emergency response in their hospitals. Learning from the experience of those for whom the unthinkable has become a daily reality can help us develop and implement more effective answers to the threats in our own communities" [14].

Int_J_Biochem_Cell_Biol-2-1-2267855

Site-directed mutagenesis of Arginine282 suggests how protons and peptides are co-transported by rabbit PepT1

The mammalian proton-coupled peptide transporter PepT1 is the major route of uptake for dietary nitrogen, as well as the oral absorption of a number of drugs, including β-lactam antibiotics and angiotensin-converting enzyme inhibitors. Here we have used site-directed mutagenesis to investigate further the role of conserved charged residues in transmembrane domains. Mutation of rabbit PepT1 arginine282 (R282, transmembrane domain 7) to a positive (R282K) or physiologically titratable residue (R282H), resulted in a transporter with wild-type characteristics when expressed in Xenopus laevis oocytes. Neutral (R282A, R282Q) or negatively charged (R282D, R282E) substitutions gave a transporter that was not stimulated by external acidification (reducing pHout from 7.4 to 5.5) but transported at the same rate as the wild-type maximal rate (pHout 5.5); however, only the R282E mutation was unable to concentrate substrate above the extracellular level. All of the R282 mutants showed trans-stimulation of efflux comparable to the wild-type, except R282E-PepT1 which was faster. A conserved negatively charged residue, aspartate341 (D341) in transmembrane domain 8 was implicated in forming a charge pair with R282, as R282E/D341R- and R282D/D341R-PepT1 had wild-type transporter characteristics. Despite their differences in ability to accumulate substrate, both R282E- and R282D-PepT1 showed an increased charge:peptide stoichiometry over the wild-type 1:1 ratio for the neutral dipeptide Gly-l-Gln, measured using two-electrode voltage clamp. This extra charge movement was linked to substrate transport, as 4-aminobenzoic acid, which binds but is not translocated, did not induce membrane potential depolarisation in R282E-expressing oocytes. A model is proposed for the substrate binding/translocation process in PepT1. 1 Introduction The proton-coupled di- and tri-peptide transporter PepT1 (SLC15a1) is the major route of uptake of dietary nitrogen from the intestine, and is also important along with the higher affinity gene product PepT2 (SLC15a2) in the re-absorption of filtered peptides in the kidney (Daniel & Kottra, 2004; Meredith & Boyd, 2000). In addition, PepT1 is the route of entry of a wide class of orally bio-available pharmaceutically important compounds, including the β-lactam antibiotics, angiotensin-converting enzyme (ACE) inhibitors, antiviral and anticancer agents (Terada & Inui, 2004). Although these therapeutic compounds are not di- or tri-peptides, they are carried by virtue of their similar 3D shape to endogenous substrates, i.e. they are peptidomimetic, and modelling of the substrate-binding site from the features in common of this huge and diverse range of substrates has led to predictions concerning which parts of the PepT1 protein may be important. For example, a substrate template model has been developed by several groups (Bailey et al., 2000; Bailey et al., 2006; Biegel et al., 2005) which allows prediction of binding affinity for a potential substrate. The rabbit PepT1 is a 707 amino acid protein, with twelve transmembrane spanning domains (TMDs) as confirmed by epitope mapping (Covitz, Amidon, & Sadee, 1998). In the absence of a crystal structure, attempts have been made to computer model the PepT1 transporter itself (Bolger et al., 1998), with site-directed mutagenesis used to test hypotheses generated by these models. One potential complication for these kind of studies is our recent report that PepT1 may form multimers in the plasma membrane (Panitsas, Boyd, & Meredith, 2006) although it is not clear how the subunits interact. Site-directed mutagenesis has been a useful tool to identify functionally important residues in PepT1. One such residue is arginine282 in the rabbit PepT1 sequence. The mutation of arginine282 to a glutamate produced a peptide transporter (R282E-PepT1) that was no longer driven by a proton-gradient but behaved more like a facilitated peptide transporter, whilst simultaneously exhibiting peptide-gated currents that were proposed to be through a non-specific cation channel activity (Meredith, 2004). Residue 282 is located approximately halfway down the predicted transmembrane domain 7 (TMD7), and is either an arginine or a lysine in all cloned mammalian PepT1 sequences to date. The presence of a charged amino acid residue in a TMD, along with its conservation, suggested a functional role. Here, we have systematically investigated the role of arginine282 in rabbit PepT1 by making further mutations to determine the requirement for the charge and have identified an interacting residue, aspartate341, located in putative TMD8. Some of these data have been previously published in abstract form (Pieri, Boyd, & Meredith, 2004). 2 Materials and methods 2.1 Site-directed mutagenesis of the PepT1 gene Oligonucleotides were custom synthesised (Sigma-Genosys, UK) for the following sequences (residues in bold are changed from wild-type PepT1):-R282-PepT1 mutants forward:where xxx was CAA for R282Q, AAG for R282K, GAT for R282D, CAT for R282H, and GCG for R282A.-D341-PepT1 mutants forward:where xxx was CGC for D341R. Reverse primers for the PepT1 mutant PCR reactions were the reverse compliment of the forward primers. The site-directed PepT1 mutants were generated using the Quikchange protocol (Stratagene), and the resulting constructs confirmed by DNA sequencing (Department of Biochemistry, University of Oxford, UK). 2.2 cRNA synthesis and oocyte injection PepT1 constructs were linearised with XbaI (New England Biolabs, UK) and cRNA generated by in vitro transcription (T7 mMessage mMachine, Ambion, Cambridgeshire, UK). X. laevis oocytes were obtained under MS222 anaesthesia (0.2%, w/v) in accordance with the UK Animals (Scientific Procedures) Act, 1986, and maintained at 18 °C in modified Barth's medium (88 mM NaCl, 1 mM KCl, 0.82 mM MgSO4, 2.4 mM NaHCO3, 0.42 mM CaCl2, 10 mM Hepes, 5 mM sodium pyruvate, 50 μg ml−1 gentamicin (Fluka, Poole, UK), adjusted to pH 7.6 with 1 M NaOH). Transport measurements were performed at least 72 h after micro-injection of oocytes with 27nl cRNA (1 μg/μl), with medium changed daily. 2.3 Transport experiments Zero-trans uptake of [3H]-d-Phe-l-Gln (17.4 Ci/mmole, custom synthesised, Cambridge Research Biochemicals, Stockton-on-Tees, UK) was performed as previously described (Meredith, 2004). Briefly, 5 oocytes were incubated in 100 μl of uptake medium (95 mM NaCl, 2 mM KCl, 1 mM CaCl2, 0.42 mM MgCl2, 10 mM Tris/Hepes pH 7.4 or Tris/Mes pH 5.5) with tracer (0.4 μM) [3H]-d-Phe-l-Gln. After incubation, the oocytes were washed sequentially five times in 1 ml of ice-cold 120 mM NaCl solution, lysed individually with 100 μl 2% (w/v) SDS and liquid scintillation counted. As a control non-injected oocytes were also incubated in uptake medium with [3H]-d-Phe-l-Gln as above. The affinity of wild-type and mutant PepT1 were assessed by competition studies with 0.4 μM [3H]-d-Phe-l-Gln and Gly-l-Gln present in the uptake medium in concentrations from 0 to 2 mM using the protocol above, and the Ki calculated using the method of Deves and Boyd (1989). Efflux studies were performed as previously described (Meredith, 2004), with the exception that the extracellular trans-stimulant Gly-l-Gln was used at 5 mM, and an efflux time-course was performed. Diethylpyrocarbonate (DEPC) inhibition of PepT1 was performed using a similar protocol to that of Terada, Saito, and Inui (1998). Briefly, PepT1 oocytes were preincubated with 1 mM DEPC for 15 min at pHout 5.5 in the absence or presence of the PepT1 substrates Gly-l-Gln, N-Acetyl-Phe (Meredith et al., 2000), and l-Ala-Tyramine (custom synthesised) and the non-substrate Tyr (all 5 mM). The oocytes were then washed in uptake medium before uptake assays were performed as detailed above. 2.4 Electrophysiology Measurements of membrane potential were made by impaling oocytes with a single glass microelectrode (Intra 767 amplifier, WPI, Stevenage, Hertfordshire, UK) perfused in uptake medium (as above) with or without 0.4 μM d-Phe-l-Gln (synthesised in house), 0.6 mM Gly-l-Gln (Sigma, Poole, UK) or 10 mM 4-aminobenzoic acid (4-AMBA, Sigma). Two-electrode voltage clamp (TEVC) was performed by placing oocytes in a 0.1 ml recording chamber and perfusing with uptake solution (pH 5.5 or 7.4) at a rate of 15 ml/min. Oocytes were impaled by two agarose-cushioned microelectrodes filled with 3 M KCl (0.5–2.0 MΩ) and voltage-clamped at −60 mV using a Geneclamp 500B amplifier and PCLAMP 8.1 software (Axon Instruments, CA, USA). The holding potential was stepped from −60 mV over the range of −150 to +50 mV in 10 mV steps, each pulse lasting 100 ms, and returning to −60 mV in between test voltage pulses. Typically traces were filtered at 1 kHz during recording and digitized at 0.5–5 kHz using the DigiData 1200 interface (Axon Instruments, CA, USA). All experiments were carried out at room temperature. 2.5 Data analysis All data are expressed as mean ± S.E.M., except for Fig. 6 where the error bars represent the maximum range of stoichiometry values when taking into account the errors for the uptake data and the currents. In order to calculate the apparent charge:substrate stoichiometry, the peptide induced current that was measured by two-electrode voltage clamp at the oocyte resting membrane potential was divided by the radiolabelled dipeptide uptake in oocytes from the same preparation. This value was normalised to 1:1 for wild-type PepT1, the accepted stoichiometry for a neutral dipeptide (Fei et al., 1994; Steel et al., 1997). 2.6 Statistical analysis Statistical analyses were performed using one-way ANOVA with differences considered significant if p < 0.05 when data were compared to the wild-type control, as detailed in the text and/or figure legend. 3 Results 3.1 pH dependence of d-Phe-l-Gln uptake into R282 mutants Fig. 1 shows the pH dependence of d-Phe-l-Gln uptake into oocytes expressing mutant PepT1 transporters where R282 has been changed into number of amino acid residues. Of the residues tested, R282K- and R282H-PepT1 behaved like the wild-type PepT1, in that the initial rate of uptake (1 h incubation time) of dipeptide was significantly faster (p < 0.05, one-way ANOVA) at an external pH of 5.5 than at 7.4. The other amino acid substitutions tested, R282E- (as previously reported, Meredith, 2004), R282D-, R282A- and R282Q-PepT1 all gave the same initial rate of uptake at pH 5.5 and 7.4, indicating that transport by these mutants is not stimulated by an inwardly directed proton gradient. These changes cannot be ascribed to changes in the affinity of the mutant PepT1 proteins for their substrate, as the Ki of Gly-l-Gln inhibiting 0.4 μM [3H]-d-Phe-l-Gln was unchanged in the mutants (Fig. 2). 3.2 Can the R282 mutant PepT1 transporters concentrate substrates? An earlier finding was that, unlike the wild-type, the R282E-PepT1 mutant was unable to concentrate substrate even in the presence of an inwardly directed proton gradient (Meredith, 2004). The ability to concentrate substrate was therefore tested for the other R282 mutants (Fig. 3a and b shows representative time-course experiments at pHout 5.5 and 7.4, respectively), and the mean accumulation levels are shown for 8 h uptakes in Fig. 3c for pHout 5.5. In contrast to R282E-PepT1, all were found to be able to concentrate peptide well above the equilibrium level when the external pH was 5.5 (an oocyte was assumed to have a volume of 1 μl, Petersen & Berridge, 1996; Yao & Tsien, 1997). In the absence of the proton driving force (pHout 7.4, Fig. 3d) a similar level of intracellular d-Phe-l-Gln concentration was reached by all the mutant PepT1 transporters, including R282E-PepT1 if the incubation time was increased to 24 h (accumulation of 2.3 ± 0.4-fold compared to 3.1 ± 0.5-fold for wild-type PepT1). Although at 24 h incubation times we observed that cell survival can be a limiting factor, there was no statistical increase in the accumulation for the wild-type PepT1 at pHout 5.5 or 7.4 between 8 and 24 h incubations, nor between 8 and 24 h for the wild-type PepT1 at pHout 7.4 (8.0 ± 1.4 vs. 8.3 ± 1.0, 2.5 ± 0.5 vs. 3.1 ± 0.5 and 1.0 ± 0.4 vs. 1.5 ± 0.3 respectively, all p > 0.05, one-way ANOVA). 3.3 Rates of d-Phe-l-Gln efflux from R282 mutants Fig. 4 shows the rates of efflux of [3H]-d-Phe-l-Gln from oocytes expressing either wild-type PepT1, R282 PepT1 mutants or non-injected controls. All of the PepT1 constructs showed a significantly faster efflux than the non-injected oocytes (one-way ANOVA, p < 0.05), whilst R282E-PepT1 was significantly faster than the wild-type (one-way ANOVA, p < 0.05) as previously described (Meredith, 2004). Interestingly, R282E-PepT1 showed a faster efflux than all the other R282 mutants (one-way ANOVA, p < 0.05), which were not significantly different to the wild-type (p = 0.14). 3.4 Apparent transport stoichiometry (charge to substrate) using two-electrode voltage clamp As well as uptake being pH independent and non-concentrative, the electrophysiological characteristics of R282E-PepT1 were strikingly different to the wild-type (Meredith, 2004). Since R282D-PepT1 is similarly pH independent but does accumulate substrate, the apparent proton to peptide stoichiometry was examined, where the uptake of 0.4 μM d-Phe-l-Gln was compared to the peptide-induced current in the same batch of oocytes. The current at the oocyte resting membrane potential (−27.0 ± 1.1 mV at pHout 5.5, and −36.8 ± 2.2 mV at pHout 7.4, n = 12) was taken to represent the membrane potential under uptake conditions, as the addition of 0.4 μM substrate does not produce a detectable change in membrane potential (Fig. 5). The ratio of proton to neutral dipeptide co-transported through wild-type PepT1 is 1:1 (Fei et al., 1994; Steel et al., 1997), yet in R282E-, R282D- and R282A-PepT1 the apparent stoichiometry is substantially higher (4, 5 and 5 respectively at pHout 5.5, Fig. 6). 3.5 Is the extra current measured dependent on substrate transport? In an attempt to see if the extra current carried by R282E-PepT1 was dependent on substrate translocation rather than simply substrate binding, the non-translocated PepT1 substrate 4-aminobenzoic acid (4-AMBA, Darcel, Liou, Tome, & Raybould, 2005; Meredith et al., 1998) was used. As found for wild-type PepT1, 4-AMBA failed to induce a depolarisation in R282E-PepT1 oocytes at 10 mM, three times its Ki (Meredith et al., 2000), in contrast to the known transported substrate Gly-l-Gln, also at three times Ki (Fig. 5). 3.6 Identification of an interacting residue for R282 The conservation of a positively charged amino acid in a transmembrane domain (TMD7), and the results above, strongly suggested that the presence of a positive charged residue was necessary for wild-type-like PepT1 transport function. In TMD8, predicted to be at approximately the same level in the membrane, there is a conserved aspartate (D341), and it was an appealing hypothesis that the two oppositely charged side chains might be forming a charge pair. To test this, double mutants were made, R282E/D341R- and R282D/D341R-PepT1, to swap the charges over. As can be seen in Fig. 7, these double mutants showed the same pH dependence of influx as the wild-type transporter, providing strong evidence to support the hypothesis. Both R282E/D341R- and R282D/D341R-PepT1 were also able to concentrate substrate like the wild-type (data not shown). 3.7 Diethylpyrocarbonate inhibition of PepT1 function Preincubation of wild-type PepT1-expressing oocytes with 1 mM diethylpyrocarbonate (DEPC) for 15 min completely inhibited the PepT1 mediated dipeptide uptake measured over 1 h, as shown in Fig. 8. This inhibition by DEPC was largely prevented by the presence of the known PepT1 substrates Gly-l-Gln and N-Acetyl-Phe (Meredith et al., 2000), but not by the amino acid non-substrate Tyr. Interestingly, despite being a good PepT1 substrate (Ki approximately 0.1 mM, data not shown), a modified peptide lacking a carboxyl terminus (l-Ala-Tyramine) only partially prevented DEPC inhibition (Fig. 8). 4 Discussion The R282K mutation of rabbit PepT1 is not only the most conservative one regarding the charge, but in a number of species, including dog, rat and mouse, lysine is the naturally occurring residue at this position. Therefore it was not surprising to find that this mutant behaves like the wild-type rabbit PepT1 (pig, sheep, rhesus and crab-eating monkeys and human also have R282). The finding that R282H also behaved like the wild-type was interesting, as histidine has a side-chain that can be titrated over the range used in the experiments (pK of ∼6 in free solution). Our findings could be interpreted in several ways, including the possibility that only at pHout 5.5 is there the formation of a positive charge by side-chain titration that gives a stimulation of uptake over that seen at pHout 7.4. A second possibility is that the protein environment surrounding H282 is such that its side chain pK is shifted away from 6 and it is therefore always protonated, and thus behaves more like an arginine or lysine. This effect has been shown for example in the enzyme protein tyrosine kinase (Tishmack, Bashford, Harms, & Van Etten, 1997), where histidine residues had a pK as high as 9.2 when analysed by NMR. In the original study on R282E-PepT1, it was concluded that the uptake of peptide by the mutant transporter was uncoupled from the movement of protons, and that in addition to acting as a facilitated peptide transporter, R282E-PepT1 also displayed a peptide-gated non-specific cation conductance (Meredith, 2004). However, it is possible that this conclusion needs updating in the light of the current findings that there are R282 mutants that, like R282E-PepT1, are not pH stimulated, yet are still able to accumulate substrate above equilibrium when an inwardly directed proton gradient is imposed. For a transporter to be able to accumulate substrate above equilibrium, an energy source must be involved, in this case the proton electrochemical gradient. Therefore, the mutant PepT1 proteins that can accumulate substrate but do not show pH stimulation (R282A and R282D) must still be coupled to the movement of protons down their electrochemical gradient. The lack of pH stimulation could be attributed to the fact that during the transport cycle these specific mutants have a different rate limiting step to the wild-type, and that for these mutants that step is not pH dependent. It has already been reported that R282E-PepT1 has a faster rate of efflux than wild-type, consistent with an uncoupling of peptide uptake from the proton driving force (Meredith, 2004); the finding here that the rates of efflux for the other R282 mutants are not different from that of the wild-type is in agreement with the hypothesis that they are still proton-coupled, as shown by their ability to accumulate substrate above the extracellular concentration at pHout 5.5 but not 7.4. The simplest hypothesis was that in R282E-PepT1, the extra inward charge movement associated with peptide uptake collapsed the membrane potential, which is known to be the major driving force for proton-coupled peptide uptake (Temple & Boyd, 1998), hence the apparent lack of substrate accumulation. Therefore by extension one would expect R282D- and R282A-PepT1 to have the same charge coupling as the wild-type, as they too accumulate substrate, but this was not the case: both R282E-, R282D- and R282A-PepT1 showed the same increased charge:peptide apparent stoichiometry, i.e. substantially larger than the wild-type. This stoichiometry itself showed pH dependence, with a lower value of around 2 for the mutants at pH 7.4, suggesting that the current is either carried by protons or is a pH-sensitive phenomenon. There was no difference in the uptake in the absence or presence of sodium, either for R282E-PepT1 (Meredith, 2004) or R282D-PepT1 (data not shown). The finding that the R282E/D341R- and R282D/D341R-PepT1 double mutants (the latter being a charge pair reversal of the naturally occurring residues in rabbit PepT1) had the same characteristics as the wild-type protein strongly suggests that these two residues do interact in the 3D protein (Pieri et al., 2004) as previously proposed (Meredith, 2004). During the preparation of this manuscript, Kulkarni et al. (2007) reported findings consistent with R282 and D341 forming a charge pair in human PepT1. The initial hypothesis that in R282E-PepT1 repulsion between E282 and D341 allowed the movement of extra ions through the protein when a peptide is transported was not supported however by the finding that the single D341R mutant also behaved like the wild-type, as one might have thought that R282 and R341 would repel in much the same way as E282 and D341. That D341R-PepT1 behaved like wild-type suggests that although R282 and D341 seem to form a charge pair, in rabbit PepT1 residue 341 being negatively charged is not crucial to PepT1 function; interestingly, in human PepT1 the D341R single mutant had reduced function (Kulkarni et al., 2007). The reason for this difference between rabbit and human PepT1 is not clear. The observation that the non-transported PepT1 substrate 4-AMBA did not induce a depolarisation in R282E-PepT1-expressing oocytes clearly shows that the charge movement is linked to substrate binding and translocation and not to binding alone. One explanation for the increase in proton–peptide stoichiometry is that in wild-type PepT1 the presence of a positively charged residue deep in the binding pocket at position 282 repels proton movement through the transporter protein during the translocation step. The data in Fig. 6 are consistent with this, as the apparent stoichiometry is lower at pHout 7.4 than it is at pHout 5.5, indicating that the proton electrochemical gradient is involved. In the case of mutations where R282 was replaced with a non-positively charged amino acid, the rate-limiting step of the transport cycle must be insensitive to extracellular pH, hence the lack of stimulation when the pHout is dropped from 7.4 to 5.5. Kinetic analysis of peptide transport by PepT1 in rat renal brush border membrane vesicles (Temple & Boyd, 1998) showed that at pHout7.4/pHin7.4, it was the protonation of the carrier protein that was the rate-limiting step (Temple, Bailey, Bronk, & Boyd, 1996), whereas at pHout5.5/pHin7.4 it was the return of the unloaded carrier. The rate of peptide uptake by R282E-PepT1 (corrected for protein expression in the intact oocyte plasma membrane by luminometry, Panitsas et al., 2006) is the same as for the wild-type at pHout 5.5 (data not shown), but, unlike for the wild-type, is not slower at pHout 7.4. This indicates that the mutations to arginine282 that abolish pH sensitivity (R to E, D, A, or Q) are affecting the rate limiting step, so when pHout is 7.4 the rate limiting step has the same magnitude as that at pHout 5.5, hence the lack of sensitivity to changing pHout. The rate limiting protonation of the outward-facing carrier protein at pHout 7.4, the first step in the transport cycle (Temple et al., 1996), was proposed by us (Bailey et al., 2000; Meredith & Boyd, 1995) and others (e.g. Steel et al., 1997; Uchiyama, Kulkarni, Davies, & Lee, 2003) to be protonation of histidine57 (H57). In the R282-PepT1 mutants, except for R282K and R282H, the lack of a positively charged residue might result in a conformational change of the protein that changes the local environment and increases the pKa of H57, such that it is more easily protonated at a higher pHout. Thus at pHout 7.4 the rate limiting step is no longer the protonation of H57, but the return of the unloaded carrier, as it is at pHout 5.5, hence the similar transport rates at pHout 5.5 and 7.4. A model for how proton–peptide transport might occur is shown in Fig. 9: the empty PepT1 is primed by protonation of H57, followed by the binding of a zwitterionic peptide, with the N-terminal co-ordinated at E595 and the C-terminal at H57-H+ (Meredith et al., 2000). Although the binding of the substrate C-terminal to His57-H+ is in disagreement with the conclusion of Terada et al. (1998), it is supported by the finding that N-Acetyl-Phe, a known PepT1 substrate which does not have a free amino terminal (Meredith et al., 2000), can protect PepT1 against inhibition by DEPC. Additionally, the finding that significant DEPC inhibition is still evident when oocytes were co-incubated with the PepT1 substrate l-Ala-Tyramine which lacks a carboxyl terminus further adds to the notion that H57-H+ binds the carboxyl terminus of the substrate (Fig. 8). H57-H+ then donates its proton to the C-terminal carboxyl group, and the transporter undergoes a conformational change that leads to the breaking of the salt bridge between R282 and D341, re-orientating the binding site to be inward facing as simultaneously the protonated N-terminal of the substrate binds to D341, neutralising the charge, and the R282 charge is stabilised by Y167 (the chemical properties of this tyrosine have been shown to be essential, Yeung et al., 1998). The peptide molecule is then released into the cytoplasm, whereby it returns to the zwitterionic state by releasing the proton from the carboxyl terminal. The transporter then undergoes the reverse conformational change to re-orientate the binding site to outward facing, and R282 reforms the salt bridge with D341. As R282E-PepT1 cannot accumulate substrate, the implication must be that the movement of peptide is no longer coupled to the movement of the protons, whereas in all the other mutants coupling must be maintained. If for R282E-PepT1 the pKa of H57 was raised to the point that it was no longer favourable for it to donate its proton to the carboxyl terminus of the substrate, then peptide transport would no longer be proton-coupled, and this would explain the failure of R282E-PepT1 to accumulate substrate. Intriguingly, it can be seen in Fig. 6 that both R282D- and R282A-PepT1 appear to carry approximately one more charge per substrate peptide than R282E-PepT1, which is consistent with the hypothesis that one of the charges carried is coupled with the substrate. In conclusion, the arginine at position 282 in rabbit PepT1 plays an intriguing role in the function of the transporter, with mutations to different residues revealing that a positive residue is required for pH dependence, whilst only R282E-PepT1 cannot concentrate substrate above equilibrium; this is despite other mutations, most notably R282D-PepT1, having a similarly increased charge:peptide stoichiometry. As previously proposed, R282 (TMD7) forms a charge pair with D341 (TMD8), with R282E/D341R-PepT1 showing normal transport characteristics. Further biological testing or a crystal structure of PepT1 will be required to establish the validity of the model proposed.

Exp_Brain_Res-4-1-2373863

Intramanual and intermanual transfer of the curvature aftereffect

The existence and transfer of a haptic curvature aftereffect was investigated to obtain a greater insight into neural representation of shape. The haptic curvature aftereffect is the phenomenon whereby a flat surface is judged concave if the preceding touched stimulus was convex and vice versa. Single fingers were used to touch the subsequently presented stimuli. A substantial aftereffect was found when the adaptation surface and the test surface were touched by the same finger. Furthermore, a partial, but significant transfer of the aftereffect was demonstrated between fingers of the same hand and between fingers of both the hands. These results provide evidence that curvature information is not only represented at a level that is directly connected to the mechanoreceptors of individual fingers but is also represented at a stage in the somatosensory cortex shared by the fingers of both the hands. Introduction The neural representation of haptic information can be investigated using different approaches. The representation of object shape perceived with the fingers has mainly been studied using neurophysiological tools. It has been found that especially slowly adapting type I (SAI) mechanoreceptors in the finger but also fast-adapting type I (FAI) receptors are sensitive to curvature (Goodwin et al. 1997; Jenmalm et al. 2003). In order to perceive curvature, a combination of responses from a population of receptors is required (Goodwin and Wheat 2004). This processing occurs along several stages up to at least the somatosensory cortex (SI) (Gardner and Kandel 2000). Taking a neurophysiological approach is useful to uncover the pathways underlying curvature processing, but is less appropriate to establish the levels at which perceived curvature is essentially represented. A psychophysical approach that has been successful in providing greater insight into the neural representation of perceived properties is the study of the aftereffect, and especially, the transfer of the aftereffect. In vision, for example, the finding of partial, interocular transfer of the motion aftereffect has been explained by the involvement of both monocular and binocular cells in the processing of motion information from the stimulus (Moulden 1980; Wade et al. 1993; Tao et al. 2003). In a similar way, establishing the transfer characteristics of a haptic curvature aftereffect would provide insight into the representation of shape information. Finding aftereffect transfer between different fingers would indicate that curvature is represented at a level shared by these fingers, whereas no transfer would imply that each finger has a separate representation of curvature. A curvature aftereffect is the phenomenon whereby a flat test surface feels concave following prolonged contact with a convex adaptation surface (see Fig. 1a). Curvature aftereffects have been found for different shapes and exploration modes. Gibson (1933) reported that a flat cardboard edge felt concave after the prolonged dynamic exploration of a convex cardboard edge. Vogels et al. (1996) demonstrated the existence of an aftereffect when the whole hand was placed on spherically curved shapes. They performed extensive experiments to examine the characteristics of this static curvature aftereffect. They found a linear relationship between the magnitude of the aftereffect and the curvature of the adaptation stimulus. Furthermore, they showed that the magnitude of the aftereffect increased with the adaptation time up to about 10 s. Finally, they found a decrease of the aftereffect with an increase of the interstimulus interval. In a follow-up study, they showed that the aftereffect also existed for alternative exploration modes, like touching a stimulus with only the five fingertips of the hand or performing small movements of the hand over the stimulus surface (Vogels et al. 1997). Given the strength and consistency of these findings, we supposed that curvature aftereffects should also occur for alternative ways of touching, such as the situation in which curved surfaces are statically being touched with only a single fingertip. However, this phenomenon has not yet been investigated, and consequently, any curvature aftereffect transfer between the fingers also remains unexplored.Fig. 1a Schematic overview of a haptic curvature aftereffect: when you first touch a convex (concave) surface for some time, say 10 s, and subsequently touch a flat surface, this latter surface feels concave (convex). b Schematic drawings of the cross-sections of a convex and a concave stimulus. The stimuli had a cylindrical shape with a spherical top (see illustration a). The distance from the bottom to the centre of the top (h) was consistently 30 mm. The diameter of the cylinders (d) was also 30 mm. c Examples of two psychometric curves. The circular data points and the fit through these points results from adaptation to the convex adaptation stimulus. The PSE is represented by PV. The square data points and the fit through these points result from adaptation to the concave adaptation stimulus. In this case, the PSE is represented by PC. The magnitude of the aftereffect (AE) is defined as the difference between PV and PC The purpose of the present study was to obtain a better understanding of the representation of haptically perceived shape information, by probing the transfer of the curvature aftereffect. In the first experiment, we established the existence of an aftereffect when a curved surface is touched by a single finger and measured whether this aftereffect transferred to other fingers of the same hand. The second experiment was set up to determine whether the aftereffect depended on the finger used. Finally, in the third and fourth experiments, we investigated the transfer of the aftereffect between fingers of both hands. Materials and methods Subjects A total number of 40 subjects participated [n = 8 for experiments 1, 2 and 4, n = 16 for experiment 3; 18 were male and 22 were female; the mean age was 22 years; 37 were right-handed, 3 were left-handed, according to a standard questionnaire (Coren 1993)]. Subjects in experiments 1 and 2 received course credit for their participation. Subjects in the third and fourth experiments received monetary compensation. Stimuli The stimuli comprised of a compound of polyurethane foam and artificial resin (Cibatool BM 5460). A computer-controlled milling machine was used to produce cylinders with a flat bottom and a spherically curved top. The top was either pointing outward (convex) or inward (concave). Both convex and concave adaptation stimuli were used, with curvature values of +36 and −36 m−1, respectively; the curvature of the nine test stimuli ranged from −16 to +16 m−1, in steps of 4 m−1. Illustrations of the stimuli and their cross-sections are given in Fig. 1a, b, respectively. Procedure A subject was seated behind a table. The preferred arm rested on a platform, which was 30 mm above the tabletop. In the third and fourth experiments, both arms rested on the platform. Only the fingertips projected over the platform. The experimenter placed the stimulus underneath a fingertip. A curtain prevented the subjects from seeing the stimulus. During a trial, the tip of one finger was placed on an adaptation stimulus for 10 s. Subsequently, the subject placed a finger on a test stimulus and had to judge whether this test stimulus felt convex or concave. Subjects were not allowed to move the finger over the stimulus surface, and the experimenter checked for this. No instructions were given on the force to contact the stimulus, nor was it measured. No feedback was provided on the response. Three conditions were measured in the first experiment. In all conditions, the adaptation stimulus was touched with the index finger. In one condition, the test stimulus was also touched with the index finger. In the other two conditions, the test stimulus was touched with the middle finger or the little finger of the same hand, respectively. Each condition consisted of 10 repetitions of a group of 18 trials (two adaptation stimuli × nine test stimuli) with trials randomized within a group. One complete condition was measured in a single session of about one and a half hours. The separate sessions were spread over different days. The order in which the conditions were conducted was counterbalanced for the first six subjects and randomly chosen for the last two subjects. In the second experiment, both the adaptation and the test stimuli were touched by the middle finger. In the third and fourth experiments, the adaptation stimulus was contacted by the index finger of the preferred hand; the test stimuli were touched with the index finger (third experiment) or middle finger (fourth experiment) of the non-preferred hand. Analysis The data for each subject and each condition were analyzed separately for the convex and the concave adaptation stimuli. The percentage of “convex” responses was plotted against the curvature of the test stimulus. The point of subjective equality (PSE) was determined by fitting a psychometric function (cumulative Gaussian) to the data. The PSE represents the curvature value that in 50% of the test cases was judged “convex” and in 50% of the cases was judged “concave”. The magnitude of the aftereffect is defined as the difference between the PSE resulting from the adaptation to a convex surface and the PSE resulting from the adaptation to a concave surface. Examples of psychometric curves for a convex and a concave adaptation are given in Fig. 1c. Results The mean results for the aftereffect values are shown in Fig. 2. The error bars indicate the standard errors of the mean.Fig. 2Mean results of the aftereffect. The indicated error bars are the standard error in the mean for each condition. Experiment 1: eight subjects participated. Adaptation was performed by the index finger of the preferred hand. Testing was done using the index finger, middle finger, or little finger of the same hand. Experiment 2: eight subjects participated. Adaptation and testing was performed by the middle finger of the preferred hand. Experiment 3: sixteen subjects participated. Adaptation was performed by the index finger of the preferred hand; testing was done by the opposite index finger. Experiment 4: eight subjects participated. Adaptation was performed by the index finger of the preferred hand; testing was done by the middle finger of the non-preferred hand Experiment 1 We tested the occurrence of an aftereffect in each condition by performing separate one-tailed t tests. A significant result was obtained in all conditions (t7 = 6.3, P < 0.001 for the index finger; t7 = 9.8, P < 0.001 for the middle finger; t7 = 3.4, P = 0.006 for the little finger). Subsequently, an ANOVA with a repeated measures design was performed to determine any differences between conditions. A significant main effect was found (F2,14 = 22.5, P < 0.001). Pairwise comparisons showed a significant difference between the index finger and the middle finger (P = 0.007) and between the index finger and the little finger (P = 0.004), but not between the middle finger and the little finger (P = 1.0). The P values were adjusted with a Bonferroni correction. Experiment 2 A one-tailed t test showed that there was a significant aftereffect (t7 = 8.0, P < 0.001). Inspection of Fig. 2 shows that the aftereffect of the middle finger condition of the second experiment was comparable to the index finger condition of the first experiment and was much higher than the middle finger condition of the first experiment. Independent samples t test confirmed that there was no significant difference in the first case (t14 = 0.6, P = 0.6), but that there was a significant difference in the second case (t7.4 = 6.1, P < 0.001). Experiment 3 A one-tailed t test highlighted a significant aftereffect (t15 = 2.7, P = 0.009). The magnitude of this aftereffect was much lower than for the index finger condition of the first experiment. This was confirmed by an independent sample t test (t22 = 5.0, P < 0.001). Experiment 4 A significant aftereffect was obtained, as was confirmed by a one-tailed t test (t7 = 7.4, P < 0.001). Discussion The first novel observation of this paper is that the perception of surface curvature by a single fingertip is influenced by preceding contact of this finger with another curved surface. The magnitude of this curvature aftereffect did not depend on the finger employed, as shown by a comparison between the results of the first and the second experiment. Previously, Vogels et al. (1996, 1997) reported the existence of a static curvature aftereffect, when the whole hand was used. We suppose that our finding of a one-finger aftereffect falls in the same class of phenomena. A quantitative comparison between the results of Vogels et al. (1996) and our finding can be made by calculating the relative magnitude of the aftereffect, i.e. the aftereffect divided by the difference between the adaptation stimuli. This value equals 0.17 ± 0.02 for the results of Vogels et al., whereas it was 0.15 ± 0.07 for the index finger condition of the first experiment and 0.17 ± 0.06 for the middle finger condition of the second experiment, respectively. These values are in the same order of magnitude, irrespective of the differences in manner of touching and curvature range of the stimuli. The second important finding of our study is that the aftereffect partially transfers between fingers of the same hand. This means that the sensation of shape with a certain finger influences the perception of a shape touched by another finger. This suggests that the sensations obtained by the different fingers share a common representation. However, the transfer is far from complete, indicating that curvature perception by each finger also yields a substantial, individual part in the representation. Interestingly, the aftereffect does not only transfer from the index finger to the neighboring middle finger, but also to the distant little finger. This result is unlike recently performed localization (Schweizer et al. 2000) and learning studies (Sathian and Zangaladze 1997; Harris et al. 2001), in which the reported transfer effects were obtained in the neighboring finger, but not in the distant fingers. This indicates that the processes involved in detecting the finger that is stimulated or increasing the skills to discriminate punctate pressure or roughness are quite different from those concerned in shape perception of an object. The third interesting result of this study is that there was a small, but significant transfer of the aftereffect between fingers of both hands, irrespective of whether opposite fingers (experiment 3) or different fingers (experiment 4) were employed. This result is different from the result reported by Vogels et al. 1997, who did not find intermanual transfer. However, in their experiments, whole hands were involved, whereas only single fingertips were used in our experiment. Moreover, their conclusion was based on the performance of only 2 subjects, whereas 24 participants provided the data for our study. The results of the third and fourth experiments suggest that the representation of shape information obtained with one hand is not completely distinct from the representation of shape information received by the other hand, but shares a common, bilateral component. How can our findings be interpreted in the context of neurophysiological literature? Firstly, our finding that the aftereffect only transfers partially between fingers of the same hand shows that a substantial part of the processing occurs at a stage where each finger is individually represented. On this stage, which spreads from the mechanoreceptors in the fingers up to area 3b in SI, no overlap occurs in signals from the slowly adapting receptors and the fast-adapting receptors (Gardner and Kandel 2000). Slowly adapting receptors respond with a sustained discharge when the finger is in contact with a surface, whereas fast-adapting receptors only respond at the onset and removal phase of the finger (Johansson and Vallbo 1983). Vogels et al. (1996) showed that the magnitude of their curvature aftereffect increased with an increase in adaptation time. These findings point to an important role for the slowly adapting receptors in the curvature aftereffect. Therefore, we suggest that the aftereffect at the stage related to an individual finger mainly originates from the processing of the slowly adapting receptors. Secondly, the fact that we found a transfer between the fingers of the same hand implies that a significant part of the processing of curvature information occurs at a level shared by the different fingers. In physiological terms, this indicates that at least area 1 or 2 of SI are involved, as receptive fields in these areas cover several fingers of a single hand (Gardner and Kandel 2000), but processing may also occur at an even higher stage. Thirdly, our finding of an intermanual transfer shows that the processing of curvature information also takes place on a higher, bilateral level. We can only speculate on the neural correlates of this bilateral processing. Possible candidates include area 2 of SI, areas 5 and 7 of the posterior parietal cortex, and the secondary somatosensory cortex (Iwamura 2000; Gardner and Kandel 2000). It is interesting to mention that the aftereffects in the intramanual transfer conditions (experiment 1) and the intermanual transfer conditions (experiment 3 and 4) are similar in magnitude. This suggests that no important curvature processing occurs at a level that is devoted to a single hand, but that all processing takes place at a higher stage. The similar results for experiments 3 and 4 provide further support that the hands and fingers are not somatotopically represented at this stage. From a previous study, it is known that subjects also performed similarly in intramanual and intermanual curvature discrimination tasks, but that higher performance was obtained when only a single finger was employed (Van der Horst and Kappers 2007). The analogy between that study and the current study is that curvature information is mainly represented at the level of the individual finger, but partly available at a higher, finger- and hand-independent level. We should be careful in ascribing a specific function to the involvement of the higher level areas in the processing of curvature information. The role of more cognitive aspects should not be excluded, since it is known that processes like tactile attention (Burton and Sinclair 2000; Spence and Gallace 2007), working memory (Burton and Sinclair 2000), and object recognition (Reed et al. 2004) also engage the somatosensory areas. The aftereffect that we found in the present study is a similar phenomenon as the aftereffect that was previously reported by Vogels et al. (1996, 1997). However, this does not entail that the representation of curvature is identical for touching with a single finger or with the whole hand. Vogels et al. (1997) already showed that, although similar aftereffects were found when either the whole hand or only the five fingertips were used, only a small transfer between these exploration modes was obtained, which points to a limited overlap in representation. Similarly, we suppose that there is a difference in representation between curvature that is perceived by a single finger and curvature that is perceived by the whole hand. In the single finger case, the representation is mainly at the level of the individual finger, whereas in the whole hand case, the representation is spread over all fingers and the palm of the hand. This study shows that establishing the intramanual and intermanual transfer of the aftereffect is a useful tool in obtaining more insight into the representation of object properties as perceived by the fingers. In general, studying aftereffect transfer is attractive, because it enables a connection between psychophysics and neurophysiology. The convergence of these approaches leads to a better understanding of human perception.

Eur_J_Appl_Physiol-3-1-1914221

Physical fitness, fatigue, and quality of life after liver transplantation

Fatigue is often experienced after liver transplantation. The aims of this cross-sectional study were to assess physical fitness (cardiorespiratory fitness, neuromuscular fitness, body composition) in liver transplant recipients and to explore whether physical fitness is related to severity of fatigue. In addition, we explored the relationship between physical fitness and health-related quality of life. Included were 18 patients 1–5 years after transplantation (aged 48.0 ± 11.8 years) with varying severity of fatigue. Peak oxygen uptake during cycle ergometry, 6-min walk distance, isokinetic muscle strength of the knee extensors, body mass index, waist circumference, skinfold thickness, severity of fatigue, and health-related quality of life were measured. Cardiorespiratory fitness in the liver transplant recipients was on average 16–34% lower than normative values (P ≤ 0.05). Furthermore, the prevalence of obesity seemed to be higher than in the general population (17 vs. 10%). We found no deficit in neuromuscular fitness. Cardiorespiratory fitness was the only fitness component that was related with severity of fatigue (rs = −0.61 to rs = -0.50, P ≤ 0.05). Particularly cardiorespiratory fitness was related with several aspects of health-related quality of life (rs = 0.48 to rs = 0.70, P ≤ 0.05). Results of our study imply that cardiorespiratory fitness and body composition are impaired in liver transplant recipients and that fitness is related with severity of fatigue (only cardiorespiratory fitness) and quality of life (particularly cardiorespiratory fitness) in this group. These findings have implications for the development of rehabilitation programs for liver transplant recipients. Introduction Liver transplantation (LTx) is the only definitive treatment for end-stage liver disease. Since 1988, patient survival and liver function after LTx have improved markedly, due to improved technical expertise, better selection of patients, improved post-LTx management of complications, and improved immunosuppressive therapy (Adam et al. 2003). In Europe, LTx has achieved a 1-year survival rate of 82% (Burroughs et al. 2006). Although studies have reported that quality of life improves after LTx (Gross et al. 1999; van der Plas et al. 2003), limitations in daily function still remain (Gross et al. 1999; van der Plas et al. 2003). Amongst these limitations, liver transplant recipients often experience fatigue (Aadahl et al. 2002; Belle et al. 1997; Leyendecker et al. 1993; van den Berg-Emons et al. 2006). Gross et al. (1999), and Belle et al. (1997) reported that, although the intensity of fatigue was reduced after LTx, fatigue remained the most distressing symptom 1 year after surgery. Leyendecker et al. (1993) found that 9 months after LTx, complaints of fatigue were more severe in the LTx group than in the general population. In a previous study, we found severe fatigue in 44% of patients up to 15 years after LTx, and these complaints did not decrease over time (van den Berg-Emons et al. 2006a). Rehabilitation programs might be effective in reducing severity of fatigue after LTx, but, to develop appropriate programs, knowledge on the factors associated with fatigue after LTx is necessary. However, data on these factors are scarce. Aadahl et al. (2002) and van den Berg-Emons et al. (2006a) suggested that the fatigue experienced by liver transplant recipients is primarily physical, rather than psychological. Furthermore, van den Berg-Emons et al. (2006b) found that severe complaints of fatigue in liver transplant recipients are associated with low levels of everyday physical activity. A hypoactive lifestyle may lead to a negative spiral: hypoactivity leading to a reduction in physical fitness and deterioration of complaints of fatigue, leading to further hypoactivity. Few studies have investigated the physical fitness of liver transplant recipients, but the limited data available suggest a reduced physical fitness after LTx (Beyer et al. 1999; Stephenson et al. 2001; Unnithan et al. 2001). Stephenson et al. (2001) found a 40–60% lower maximal oxygen uptake than predicted in liver transplant recipients. Beyer et al. (1999) reported that, although the cardiovascular and neuromuscular fitness in liver transplant recipients improved after a supervised exercise program during the post-operative year, maximal oxygen uptake and muscle strength remained 10–20% lower compared to healthy gender and age-matched individuals. Also in pediatric liver transplant recipients, deficits in cardiovascular fitness, and abdominal muscle strength have been reported (Unnithan et al. 2001). It may be hypothesized that deficits in physical fitness in liver transplant recipients are associated with complaints of fatigue. Furthermore, deficits in physical fitness may lead to impaired health-related quality of life (HRQoL). However, to our knowledge, no studies are available on the relationships between these parameters in liver transplant recipients. Because of the scarcity of studies on physical fitness and related parameters in liver transplant recipients, the present study assessed physical fitness (cardiorespiratory fitness, neuromuscular fitness, and body composition) in liver transplant recipients and explored whether physical fitness is related to severity of fatigue in this group. The relationship between physical fitness and HRQoL was also explored. Patients and methods Patients To obtain a representative sample of liver transplant recipients with respect to fatigue, we recruited liver transplant recipients with varying severity of fatigue, according to the distribution of severity of fatigue as found in our previous study in 96 liver transplant recipients (van den Berg-Emons et al. 2006a). Severity of fatigue in this previous study ranged from ‘no signs of fatigue’ to ‘most disabling fatigue’, and was assessed with the Fatigue Severity Scale of Krupp et al. (1989) (see below). Inclusion criteria for the present study were: LTx between 1 and 5 years ago, sufficient knowledge of the Dutch language, and age between 18 and 65 years. Exclusion criteria were: multiorgan transplant recipients, severe comorbidity, and contra-indication for a progressive maximal cycle ergometer test. Of the original sample of 96 patients in the previous study (van den Berg-Emons et al. 2006a), 4 patients had died, 1 patient had emigrated, 53 patients were transplanted more than 5 years previously, and 10 patients were not eligible because of contra-indications for a maximal cycle ergometer test. Of the 28 eligible patients, 18 patients (64%) agreed to participate. There were no significant differences in relevant characteristics between the patients who decided to participate and the non-participants. The study was approved by the Medical Ethics Committee of the Erasmus University Medical Center. Written informed consent was obtained from all subjects. Table 1 shows the characteristics of the study group. Table 1Characteristics of the study group (n = 18)Age (years)48.0 ± 11.8Gender Male11 Female7Primary disease (n) a Chronic17 Acute1Time since transplantation (years)3.3 ± 1.1Immunosuppressive agentsb 116 21 31Results are presented as mean ± SD or numbersaChronic primary disease: cholestatic (n = 6), viral (n = 6), miscellaneous (n = 5); acute primary disease: intoxication (n = 1)bImmunosuppressive agents: 1 agent, cyclosporine or tacrolimus; 2 agents, prednisone with tacrolimus; 3 agents, prednisone with tacrolimus and azathioprine Measurements Physical fitness: cardiorespiratory Cardiorespiratory fitness was measured with a progressive maximal aerobic test on a cycle ergometer (ER800, Jaeger Toennies, Breda, The Netherlands). The test was preceded by a 1-min warm-up period (20 W). The test started at 20 W, and resistance was increased every minute by 15 or 20 W, depending on the ability of the patients. Individual protocols were constructed such that the total exercise time ranged from 8 to 12 min. The pedal rate was 60 rpm and strong verbal encouragement was given during the test. The test was terminated when the subject voluntarily stopped due to exhaustion, or when the patient was unable to maintain the initial pedal rate. Gas exchange and heart rate (HR) were measured continuously using a breath-by-breath gas analysis system (K4b2, COSMED, Rome, Italy). Subjective strain was measured immediately after the final stage by the Borg Category Scale for Rating of Perceived Exertion (Borg 1982). Patients were asked to indicate how strenuous they had experienced the test by giving a number from 0 (no effort at all) to 10 (maximal effort). Cardiorespiratory fitness was defined as the mean oxygen uptake during the last 30 s of exercise [VO2peak, in ml kg−1 min−1 and in ml kg fat-free mass−1 min−1 (ml kgFFM−1 min−1)]. In addition, the ventilatory anaerobic threshold (VAT, expressed as percentage of predicted VO2peak) was estimated by the ventilatory equivalent method, when VE/VO2 and PetO2 increased while VE/VCO2 and PetCO2 remained stable (Reinhard et al. 1979; Wasserman et al. 1999). Finally, patients performed the submaximal 6-min walk test (6MWT) (Guyatt et al. 1985). Patients were instructed to walk, not run, as far as they could along a 30-m marked tape in a hall during a 6-min period. Standardized encouragement was provided with the following phrases: “You are doing well” and “Keep up the good work”. Patients were allowed to stop and rest during the test, but were instructed to resume walking as soon as they felt able to do so. The 6-min walk distance (6MWD) was registered. Physical fitness: neuromuscular Isokinetic muscle strength of the knee extensors was assessed in both legs by a Biodex® dynamometer (Shirley, New York, USA), recording strength as torque in Nm. The patients were seated against a back-rest, firmly strapped at the hip and thigh. The rotational axis was aligned with the lateral femoral epicondyle. After five familiarization repetitions, isokinetic strength was measured at 60°/s with 5 maximal contractions and at 180°/s with 15 maximal contractions. Strong verbal encouragement was given during the test. Peak torque (PT) was defined as the maximum torque generated by the patients throughout one series of repetitions at each velocity. Physical fitness: body composition Height and body mass were measured without shoes. Body mass was measured using a Cormier Paribel® weighing chair (FH Balances Cormier, Romainville, France). Body mass index (BMI, kg m−2) was calculated from height and body mass. Waist circumference (cm) was measured mid-way between the lowest rib and the iliac crest while standing. Thickness of four skinfolds (biceps, triceps, subscapular, suprailiaca region) was measured twice at the right side of the body with a Harpenden Skin-Fold Caliper (Burgess Hill, UK). The mean of the two measurements was used as representative for each site. Percentage body fat (BF) was predicted from skinfold thickness according to the method of Durnin and Womersley (1974). Severity of fatigue Severity of fatigue was assessed by the Dutch version of the Fatigue Severity Scale (FSS) (Krupp et al. 1989). The FSS is a self-administered questionnaire with answers ranging from 1 (‘strongly disagree’) to 7 (‘strongly agree’). The mean score of the nine inquiries ranges from 1 (‘no signs of fatigue’) to 7 (‘most disabling fatigue’). Internal consistency, reliability, validity, and sensitivity of the FSS have been established in several patient groups (Krupp et al. 1989; Merkies et al. 1999). In addition to the FSS, severity of fatigue was assessed with a horizontal visual analogue scale (VAS). Patients were asked to mark the 100-mm line according to how intense they had experienced fatigue during the last month (0 denotes ‘no fatigue experienced’ and 100 denotes ‘the most severe fatigue’) (ter Borg et al. 2004). Visual analogue scales have been found to yield reliable and valid data (Huskisson 1982; Scott and Huskisson 1979). Health-related quality of life HRQoL was assessed by the validated Dutch version (RAND-36) (van der Zee and Sanderman 1993) of the Medical Outcomes Study Short Form-36 (SF-36) (Ware and Sherbourne 1992). The SF-36 is a validated, self-administered questionnaire used internationally to measure health status with respect to different dimensions: physical functioning, social functioning, role limitations due to physical problems, role limitations due to emotional problems, pain, mental health, vitality, general health perception, and change in perceived health during the last 12 months. All raw scores were converted to a 0–100 scale, with higher scores indicating higher levels of functioning or well-being. Procedure On the day of the measurements, patients refrained from caffeine, nicotine, and heavy exercise. The order of the tests was standardized: patients started with the 6MWT, followed by the questionnaires (completed under supervision of the researcher), body composition measurements, strength test, and finally the progressive maximal aerobic test. Exercise tests were performed under supervision of a physician. There were sufficient rest periods between the tests. Statistics Statistical analysis was performed using SPSS 10.1 for Windows (SPSS Inc., Chicago, IL, USA). Results are presented as mean ± standard deviation (SD), range or numbers. Results on cardiorespiratory fitness were compared to normative values for sedentary persons and to normative values for people who exercise no more than 1–2 h a week (recreational) (Vos et al. 2001). Results on neuromuscular fitness were compared with the normative values of Akima et al. (2001) and 6MWD was compared with the normative values of Enright and Sherill in healthy adults aged 40–80 years (1998). When patients were younger than 40 years, the normative values of Gibbons et al. (2001) were used. Obesity was defined as a BMI ≥ 30 (World Health Organization 1989), a waist circumference ≥102 cm in men and ≥88 cm in women (Lean et al. 1995) or percentage BF ≥25% in men and ≥32% in women (Lohman 1992). Differences in physical fitness between patients and normative values were tested with the non-parametric Mann-Whitney U Test. Relationships between physical fitness and severity of fatigue and between physical fitness and quality of life were explored using the non-parametric Spearman correlation coefficient (rs). A probability value ≤0.05 determined statistical significance. However, because of the relatively small study sample, also results on the α = 0.10 level are presented (indicating a trend). Results Physical fitness Mean score on the Borg Scale was 6.3 ± 2.2, indicating that the patients experienced the maximal ergometer test on average as heavy to very heavy. The mean HRmax was 90 ± 10% of what was predicted (predicted HRmax = 220-age) (Fox et al. 1971). The individual results of the patients are presented in Table 2. Table 3 shows the cardiorespiratory fitness and neuromuscular fitness of the patients compared with normative values. VO2peak in ml kg−1 min−1 was 15 ± 22% (P = 0.07) and 34 ± 15% (P = 0.00) lower than normative values for, respectively, sedentary persons and those who exercise recreationally. VO2peak in ml kgFFM−1 min−1 was 16 ± 19.2% (P = 0.02) and 33 ± 14.8% (P = 0.00) lower than normative values for, respectively, sedentary persons and those who exercise recreationally. 6MWD was 16 ± 14% (P = 0.01) lower than normative values. There was no significant deficit in neuromuscular fitness. Table 4 shows the body composition of the LTx group. According to the cut-off points for obesity based on BMI, waist circumference and percentage BF, respectively 17, 36, and 41% of the patients were classified as obese. Table 2Individual results on physical fitness in the 18 liver transplant recipientsAge (year/Gender)BMI (kg m−2)Waist (cm)aBody fat (%)bVO2peak (ml kg−1 min−1)VO2peak (ml kgFFM−1 min−1)bVAT (% predicted VO2peak sed) VAT (% predicted VO2peak recr)6MWD (m)PT extension 60°/s (Nm)PT extension 180°/s (Nm)56/M28.3111.029.036.551.454.142.7652156.388.252/M29.032.716.724.849.138.9428101.873.142/F22.828.934.548.590.565.8530100.065.660/F38.4118.542.013.122.543.632.5495101.978.664/M29.7103.530.621.731.377.455.443099.962.624/F21.072.025.929.940.438.831.1660159.9111.546/M26.5100.025.828.238.048.638.7606151.688.563/M24.420.520.926.469.449.741074.444.842/M34.9132.022.830.925.8513145.9118.353/F23.977.530.123.133.0108.664.846540.035.753/M25.999.623.620.727.156.742.6510149.396.839/M23.787.017.431.237.851.543.5547169.0111.833/M24.088.019.835.143.767.959.9510155.9116.963/F33.0103.034.815.824.257.943.230367.239.445/F26.830.823.634.161.046.2514120.161.739/F18.775.022.730.238.156.246.8600115.967.932/M22.991.020.425.031.443.438.3576180.4124.658/M25.199.025.018.524.648.637.9495128.691.9Abbreviations: BMI (kg m−2), body mass index; Waist (cm), waist circumference; VO2peak (ml kg−1 min−1), peak oxygen uptake per kg; VO2peak (ml kgFFM−1 min−1), peak oxygen uptake per kg fat-free mass; VAT (% predicted VO2peak), ventilatory anaerobic threshold as percentage of predicted VO2peak; sed, sedentary normative values; recr, recreational normative values; 6MWD, 6-min walk distance; PT, peak torqueaWaist circumference: n = 14, because of thickness of the skin at the place of the cicatrice in four patientsbBody fat and VO2peak: n = 17, because the thickness of the subscapular skinfold could not be measured reliably in one patientTable 3Cardiorespiratory fitness and neuromuscular fitness in the liver transplant groupLTx groupNorm valuesaPCardiorespiratory fitness VO2peak (ml kg−1 min−1)24.8 ± 6.929.4 ± 7.4 (sed)0.07†37.5 ± 6.7 (recr)0.00* VO2peak (ml kgFFM−1 min−1)b34.0 ± 8.741.0 ± 7.8 (sed)0.02*50.9 ± 5.9(recr)0.00* 6MWD (m)513.6 ± 88.9609.6 ± 97.60.01*Neuromuscular fitness PT extension at 60°/s (Nm)123.2 ± 38.7138.5 ± 44.10.22 PT extension at 180°/s (Nm) 82.1 ± 27.986.1 ± 30.10.59Results are presented as mean ± SDAbbreviations: VO2peak (ml kg−1 min−1), peak oxygen uptake per kg; VO2peak (ml kgFFM−1 min−1), peak oxygen uptake per kg fat-free mass; VAT (% predicted VO2peak), ventilatory anaerobic threshold as percentage of predicted VO2peak; sed, sedentary normative values; recr, recreational normative values (people who exercise no more than 1–2 h a week); 6MWD 6-min walk distance; PT, peak torqueaVos et al. (2001), Enright and Sherill (1998), Gibbons et al. (2001), Akima et al. (2001)bn = 17, because the thickness of the subscapular skinfold could not be measured reliably in one patient* Significant (P ≤ 0.05) difference between patients and normative values† Difference between patients and normative values at the α = 0.10 level (trend)Table 4Body composition in the liver transplant groupBody compositionMean ± SDBody weight (kg)80.6 ± 18.3Height (m)1.74 ± .11Body mass index (kg m−2)26.6 ± 5.0Waist circumference (cm)a96.6 ± 16.8Body fat (%)b26.9 ± 6.4Results are presented as mean ± SDaWaist circumference: n = 14, because of thickness of the skin at the place of the cicatrice in four patientsbBody fat: n = 17, because the thickness of the subscapular skinfold could not be measured reliably in one patient Relationships Table 5 shows the correlation coefficients between the physical fitness parameters and severity of fatigue as assessed with the FSS and VAS. None of the parameters of neuromuscular fitness or body composition were significantly related with severity of fatigue, assessed with either the FSS or the VAS. Table 5Spearman correlation coefficients for the relationships between fitness parameters and severity of fatigue as measured with the Fatigue Severity Scale (FSS) and the Visual Analogue Scale (VAS) in 18 liver transplant recipientsPhysical fitnessFatigueFSSVASRsPRsPCardiorespiratory fitness VO2peak (ml kg−1 min−1)−0.400.10−0.520.03* % of sedentary norm−0.170.50−0.42.0.08† % of recreational norm−0.320.20−0.530.03*VO2peak (ml kgFFM−1 min−1)a−0.430.08†−0.510.04* % of sedentary norma−0.350.17−0.500.04* % of recreational norma−0.450.07†−0.610.01* VAT (% predicted VO2peak sed)0.220.380.030.92 VAT (% predicted VO2peak recr)0.170.49−0.060.82 6MWD (m)−0.440.07†−0.530.03* % of norm−0.250.32−0.520.03*Neuromuscular fitness PT extension at 60°/s (Nm)−0.390.11−0.310.22 % of norm−0.150.56−0.280.27 PT extension at 180°/s (Nm)−0.300.22−0.120.64 % of norm0.020.930.040.87Body composition Body mass index (kg m−2)0.030.910.060.82 Waist circumference (cm)b−0.070.82−0.060.83Body fat (%)a0.320.210.150.58Abbreviations: VO2peak (ml kg−1 min−1), peak oxygen uptake per kg; VO2peak (ml kgFFM−1 min−1), peak oxygen uptake per kg fat-free mass; VAT (% predicted VO2peak), ventilatory anaerobic threshold as percentage of predicted VO2peak; sed, sedentary normative values; recr, recreational normative values (people who exercise no more than 1–2 h a week); 6MWD, 6-min walk distance; PT, peak torquean = 17, because the thickness of the subscapular skinfold could not be measured reliably in one patientbn = 14, because of thickness of the skin at the place of the cicatrice in four patients* Significant (P ≤ 0.05) correlation† Correlation at the α = 0.10 level (trend) VO2peak in ml kgFFM−1 min−1 (absolute and expressed as percentage of recreational normative values) and 6MWD (absolute) were related with severity of fatigue as assessed with the FSS at the α = 0.10 level (trend). With respect to the VAS, VO2peak in ml kg−1 min−1 (absolute and expressed as percentage of recreational normative values), VO2peak in ml kgFFM−1 min−1 (absolute and expressed as percentage of both sedentary and recreational normative values) and 6MWD (absolute and expressed as percentage of normative values) were significantly (P ≤ 0.05) correlated with severity of fatigue. VO2peak in ml kg−1 min−1 as percentage of sedentary normative values was related with severity of fatigue as assessed with the VAS at the α = 0.10 level (trend). The VAT was not correlated with severity of fatigue, assessed with the FSS or the VAS. Table 6 shows the correlation coefficients between physical fitness parameters and HRQoL. There were several significant correlations between physical fitness and HRQoL and particularly between cardiorespiratory fitness on the one hand and ‘Physical functioning’, ‘Social functioning’ and ‘Vitality’ on the other hand. Table 6Spearman correlations coefficients for the relationships between physical fitness and health-related quality of life as assessed with the RAND−36 in 18 liver transplant recipientsPhysical fitnessRAND-36 domainPFSFRlpRleMHVTBPGHCHCardiorespiratory fitness VO2peak (ml kg−1 min−1)0.55–––––––– % of sedentary norm–0.42–––0.50––– % of recreational norm0.48––––0.51––– VO2peak (ml kgFFM−1 min−1)a0.570.510.41–––––– % of sedentary norma–0.56–––0.59–0.48– % of recreational norma0.460.58–––0.58––– VAT (% predicted VO2peak sed)––––––––– VAT (% predicted VO2peak recr)––––––––– 6MWD (m)0.670.57––0.53–0.54–– % of norm–0.70–––0.51––0.44Neuromuscular fitness PT extension at 60°/s (Nm)0.44–––.41–––– % of norm–0.68––––0.56–– PT extension at 180°/s (Nm)––––––0.44–– % of norm–0.51–0.54––0.50––Body composition Body mass index (kg m−2)––––––––– Waist circumference (cm)b–––0.49–0.41––––– Body fat (%)a–––––––To enhance the clarity of the table, only significant correlation coefficients (P ≤ 0.05, bold) or trends (P < 0.10, not bold) are presentedAbbreviations: VO2peak (ml kg−1 min−1), peak oxygen uptake per kg; VO2peak (ml kgFFM−1 min−1), peak oxygen uptake per kg fat-free mass; VAT (% predicted VO2peak) ventilatory anaerobic threshold as percentage of predicted VO2peak; sed, sedentary normative values; recr, recreational normative values (people who exercise no more than 1–2 h a week); 6MWD, 6-min walk distance; PT, peak torque; PF, physical functioning; SF, social functioning; Rlp, Role limitations physical; Rle, Role limitations emotional; MH, mental health; VT, vitality; BP, bodily pain; GH, general health perception; CH, changes in healthan = 17, because the thickness of the subscapular skinfold could not be measured reliably in one patientbn = 14, because of thickness of the skin at the place of the cicatrice in four patients Discussion This is the first study in which relationships between several aspects of physical fitness and severity of fatigue are explored after LTx. The study is an initial step in identifying factors that are associated with fatigue in liver transplant recipients. A limitation of this study may be that the sample was relatively small. However, we believe that this sample is representative for patients after LTx and that the study provides important information for the development of rehabilitation programs for liver transplant recipients. Physical fitness We found average deficits in VO2peak of 16–34%, when we compared the results in the liver transplant recipients with normative values for, respectively, sedentary persons and those who exercise no more than 1–2 h a week. We believe that a value in between the sedentary and recreational normative values is representative for healthy Dutch persons of the same age as the liver transplant recipients that participated in our study. 6MWD was 16% lower than normative values. This subnormal level of cardiorespiratory fitness (both VO2peak and 6MWD) is in agreement with findings of previous studies (Beyer et al. 1999; Stephenson et al. 2001; Unnithan et al. 2001). Furthermore, the prevalence of obesity was higher in this study than in the general Dutch population; 17% of the patients had a BMI of more than 30 compared to 10% in the general Dutch population (Gezondheidsraad 2003). Impaired cardiorespiratory fitness in liver transplant recipients may be due to the use of immunosuppressive medication, e.g. glucocorticoids and calcineurin inhibitors, which may influence both the cardiovascular system and skeletal muscles (Hokanson et al. 1995; Mercier et al. 1996). Immunosuppressive medication can also induce appetite stimulation and diabetes mellitus; this in association with emotional conditions can cause a change in eating habits (Correia et al. 2003; Hussaini et al. 1998). Furthermore, patients with primary biliary cirrhosis and some patients with other chronic liver disease are hypermetabolic. LTx in these patients may lead to a reduction in resting metabolic rate and can also cause an increase in BF (Hussaini et al. 1998; Richardson et al. 2001). Besides immunosuppressive medication and changes in appetite and metabolism, also deconditioning (both before and after transplantation) may contribute to the impaired cardiorespiratory fitness and body composition (Epstein et al. 1998; Grose et al. 1995). In a previous study we found that severe complaints of fatigue were associated with low levels of everyday physical activity in liver transplant recipients (van den Berg-Emons et al. 2006b). Because of the immunosuppressive medication and deconditioning, we also expected a deficit in neuromuscular fitness. However, this was not demonstrated in the present study, in contrast to Beyer et al. (1999), who found a 10–20% lower muscle strength in liver transplant recipients compared to age- and sex-matched sedentary individuals. They contributed this deficit to muscle weakness as a side-effect of glucocorticoids. The discrepancy between the findings of our study and those of Beyer et al. (1999) may be explained by the use of Tacrolimus instead of cyclosporine in the majority of our subjects. Tacrolimus is known to have less side effects than cyclosporine (Maes and Vanrenterghem 2004). Stephenson et al. (2001) reported an early VAT in liver transplant recipients (<45–50% of the predicted VO2max) compared to 50–60% of the VO2max in healthy persons found by Davis et al. (1997). They contributed this early anaerobic threshold in liver transplant recipients to the cyclosporine-induced decrease in mitochondrial oxygen consumption (2001). However, the European Respiratory Society (ERS Task Force 1997) reported that there is a wide range of normal predicted values (35–70%). Therefore, it is difficult to indicate whether the VAT in our patients was reduced compared to normal (58.6 ± 18.9 and 44.7 ± 11.1% of predicted VO2peak for, respectively, sedentary persons and those who exercise recreationally). Physical fitness and severity of fatigue Although relationships between severity of fatigue (as measured with the FSS and VAS) and physical fitness were not univocal, this study demonstrates a relationship between cardiorespiratory fitness and severity of fatigue after LTx. Patients with more severe complaints of fatigue had larger deficits in cardiorespiratory fitness than patients with less severe complaints of fatigue. Although this cross-sectional study does not allow us to conclude that a reduced cardiorespiratory fitness results in fatigue (or vice versa), there may be an interaction between parameters: complaints of fatigue leading to decreased physical activity (van den Berg-Emons et al. 2006b) and decreased physical fitness, leading to further deterioration of complaints of fatigue. It may then be hypothesized that rehabilitation programs, aimed at enhancing cardiorespiratory fitness, can be effective in breaking through this negative spiral and (partly) reduce complaints of fatigue in this population. However, this hypothesis has to be confirmed in future randomized trials. In contrast to our expectations, the present study indicates that other aspects of physical fitness, neuromuscular fitness and body composition, do not seem to be related with severity of fatigue after LTx. However, it should be realized that our study sample was relatively small and some of the studied relationships may have failed to show statistical significance. Physical fitness and HRQoL Previous studies on healthy persons indicate that physical activity, fitness, and body fatness are associated with HRQoL and mood (Han et al. 1998; Stewart et al. 2003). There are also indications that physical activity is related to HRQoL after LTx (Painter et al. 2001; van den Berg-Emons et al. 2006b). Therefore, we expected to find a relationship between physical fitness and HRQoL in our study group. In agreement with the study of Stewart et al. (2003) in healthy elderly subjects, we found several significant correlations between physical fitness (particularly cardiorespiratory fitness) and HRQoL in our liver transplant recipients. Patients with large deficits in physical fitness experienced worse HRQoL than patients with small deficits in physical fitness. However, in contrast with studies in healthy persons (Han et al. 1998; Stewart et al. 2003), we found only few relationships between body composition and HRQoL. Our results on the relationships between physical fitness and HRQoL imply that rehabilitation programs aimed at improving physical fitness (particularly cardiorespiratory fitness) may consequently result in improved HRQoL, particularly in improved physical and social functioning, vitality, and bodily pain. However, these implications have to be confirmed in future randomized trials on the effects of such rehabilitation programs in liver transplant recipients. Conclusion Cardiorespiratory fitness in the liver transplant recipients was distinctly impaired and the prevalence of obesity was higher than in the general population; there were no indications that neuromuscular fitness is impaired after LTx. Based on the relationships we found between cardiorespiratory fitness and severity of fatigue, a rehabilitation program aimed at enhancing cardiorespiratory fitness may help in reducing complaints of fatigue after LTx. Such rehabilitation programs may also result in improved HRQoL.

Clin_Rev_Allergy_Immunol-4-1-2243253

Anti-CCP2 Antibodies: An Overview and Perspective of the Diagnostic Abilities of this Serological Marker for Early Rheumatoid Arthritis

The literature of the last 4 years confirms that the anti-CCP2 test is a very useful marker for the early and specific diagnosis of rheumatoid arthritis (RA). The anti-CCP2 test is very specific for RA (95–99%) and has sensitivity comparable to that of the rheumatoid factor (70–75%). The antibodies can be detected very early in the disease and can be used as an indicator for the progression and prognosis of RA. In this review, these interesting properties and some future possibilities of this diagnostic test are discussed. The presence of autoantibodies in the serum of patients is a very typical phenomenon for autoimmune diseases. Most of these autoantibodies, however, can also be detected in patients with other conditions and are therefore not specific. A typical example is the rheumatoid factor (RF), which is present in most inflammatory conditions. However, in some cases, autoantibodies can give the clinician a more precise indication of the type of underlying disease because they occur specifically in a certain disease. For example, anti-Sm antibodies are linked almost exclusively to systemic lupus erythematosus (SLE); whereas anti-DNA topoisomerase-I antibodies are typically present in scleroderma patients. Among the most disease-specific autoantibodies described are the so-called ACPA (anti-citrullinated protein/peptide antibodies). These antibodies occur specifically in RA and can be measured most conveniently via the anti-CCP (anti-cyclic citrullinated peptide) antibody test. The first generation CCP test (CCP1) used in early diagnostic studies (2000–2001) contained a single cyclic citrullinated peptide derived from filaggrin as the substrate [1]. It could detect ACPA in 68% of patients with established RA with a very high specificity (98%). Because filaggrin is not expressed in the synovium, it is most likely not the natural citrullinated antigen for ACPA. Other peptides, not related to filaggrin, could therefore potentially provide better epitopes for detection of ACPA. Via screening of a number of peptide libraries, novel citrullinated peptides were obtained and incorporated into a second generation CCP test (CCP2). This test is commercially available, and as all companies use the same type of CCP2 peptides, standardization is achieved quite easily. The diagnostic properties of this test will be discussed below. The CCP2 Test is Sensitive and Highly Specific for RA Since its appearance on the market in the second half of 2002, the diagnostic properties of the CCP2 test have been studied by many laboratories. This resulted in more than 120 publications dealing with this subject. The accumulated data, including only papers that appeared in PubMed till December 2006, are given in Table 1. It is clear that the accumulated data confirm the earliest reports on specificity and sensitivity of the CCP2 test. The anti-CCP2 test demonstrates an RF-like sensitivity with a very high specificity for RA (see also recent reviews: [2, 3]). It is also commonly recognized that anti-CCP2 antibody may be present in up to 40% of RF-negative RA sera [4, 5]. Table 1Cumulative anti-CCP2 diagnostic data published between 2002 and 2006Patient groupNumberCCP2+Sensitivity (%)Specificity (%)RA total14,18310,13571.5 Early3,8762,36561.0 Established10,3077,77075.4Controls15,1566834.595.6 Non-RA11,5026475.694.4 Healthy3,654361.099.0In total, 122 independent studies were included. The selection for early RA has been adopted from the original reports. The anti-CCP2 test enables clinicians to distinguish RA patients from other arthritic diseases, especially in cases where the RF test is not discriminative. This is, for example, the case with chronic hepatitis C virus (HCV) infection, a disease that can easily be misdiagnosed as it often reveals RA-like arthropathies and, in many cases, is accompanied by a positive RF. Several examples of such studies are given in recent reviews on this subject [6]. Recently, there is also an interest to compare the diagnostic potential of anti-CCP2 with novel tests based on the use of a citrullinated antigen (for example, MCV = anti-mutated citrullinated vimentin [7], CPA = citrullinated protein antibodies, VCP = anti-viral citrullinated protein [8], antihuman fibrinogen alpha [9, 10], and CCP3 [11]). For a reliable comparison of these tests, it is essential that their performance is assessed under the same conditions, e.g., the sensitivities of the tests should be determined at the same level of specificity. Good examples of such stratified studies have been carried out recently by van der Cruijssen et al. [10] and Dejaco et al. [7]. For example, Dejaco et al. [7] showed, in a large cohort of patients (>600), that at a specificity of 98.7%, being the specificity of the anti-CCP2 test, the sensitivity of the anti-MCV test is 53.7% as opposed to 70.1% for the anti-CCP test. Coenen et al. [11] compared several commercial tests, including a very recent CCP3 test from Inova. At the cut-offs recommended by the various manufacturers, the positive predictive value of the three commercial CCP2 tests is about 90% with a specificity of around 96%. The specificity of the other tests (CCP3 = 88%, MCV = 90%, CPA = 94%) is lower as are their positive predictive values [11]. These numbers may improve a little bit when the cut-off values are adjusted to more realistic data; nevertheless, the data allow the statement that, in absolute percentages, none of the tests performs better than the anti-CCP2 test. They also seem to indicate that some tests detect RA patient groups that are negative in the anti-CCP test, illustrating again that the autoantibody repertoire of RA patients is very heterogeneous. Another risk for the specificity of a test that is based on a citrullinated antigen is the possibility that antibodies are not directed exclusively to the citrulline-containing epitope but also to other possibly overlapping epitopes present in the substrate antigen. This is particularly important when citrullinated versions of proteins like vimentin or fibrinogen are used. For example, it is known that antibodies to vimentin are present in several diseases different from RA [12, 13]. This particular problem has been addressed for CCP2 by Vannini et al. [14]. They used ELISA plates containing the control CCP2 antigens (Arg instead of Cit in the same peptide context), produced and made available by Euro-Diagnostica, Arnhem, The Netherlands, in parallel to the normal CCP2 test. The results of these comparative studies showed that in RA and most non-RA rheumatic disease sera, anti-CCP reactivity indeed is citrulline-dependent. However, in some patients, particularly autoimmune hepatitis patients, citrulline-independent reactivity with the antigen may occur. A positive CCP test in a rheumatic disease (almost always citrulline-specific) may thus suggest the future development of RA as has been suggested by several studies [15, 16]. A positive test in a nonrheumatic disease (very often not citrulline-specific), for example, liver disease, should be interpreted with care [14]. Anti-CCP2 Antibodies are Present Early in Disease and have Predictive Potential Because RA patients at first presentation often do not fulfill the criteria for the diagnosis/classification of RA, an early, highly predictive marker would greatly assist the clinician in reaching an early diagnosis. There are several studies indicating that the anti-CCP2 test provides this help (reviewed by [2]). In the recently published EULAR recommendations for the management of early arthritis [16], a list of factors has been proposed that predict persistent and erosive disease. These factors include: number of swollen and tender joints, ESR or CRP, level of RF and anti-CCP antibodies, and radiographic erosions. Most of these factors were also mentioned as being important in the prediction of early erosive RA (Visser et al. [17]). Subsequent studies by the same group gave an indication of the relative importance of these factors. When expressed as odds ratios (OR), the data was as follows: arthritis of three or more joints, 5.0; radiographic erosions, 8.7; positive IgM-RF, 1.7; and positive anti-CCP2, 38.6 [18]. These and other data (see also [19]) clearly show that the presence of anti-CCP antibodies is an important and independent prognostic factor for radiographic progression in not only early arthritis but also in early rheumatoid arthritis [16, 20]. Recently, it has also been shown that IgM-CCP is present in early samples from both patients with undifferentiated arthritis (UA) and patients with RA as well as in follow-up samples from patients with RA. These data indicate the development of the anti-CCP isotype repertoire into full usage early in the course of arthritis and a continuous (re)activation of the RA-specific anti-CCP response during the further development of the disease [21]. It is, however, evident that besides the clinical and laboratory parameters mentioned above, some genetic factors are important as well. The effect of the HLA shared-epitope alleles on the development of ACPA has been firmly established. Citrullination is typically a process that occurs in apoptotic cells. Because such dying cells are generally removed from the environment via clearance by phagocytes, a process that is regulated by many genes, the immune system will normally not encounter citrullinated proteins. However, it has been shown that during inflammation, citrullinated proteins are detectable in the inflamed tissue, both in RA and non-RA patients [22]. This is probably caused by inefficient clearance of the massive numbers of dying cells, a process already described to occur in SLE [23]. This inefficient clearing could, in principal, also be the consequence of altered genes, i.e., the genetic background. The mere presence of citrullinated antigens in inflamed synovial tissue does not necessarily result in the occurrence of anti-CCP antibodies in serum or synovial fluid, neither in humans nor in mouse models of arthritis [24, 22]. Hill et al. [25] showed that the generation of anticitrulline antibody in mice actually is linked to the expression of the RA shared epitope (SE), and a similar link was also found in patients because the combination of SE, and anti-CCP has a very high predictive value for the future development of RA [26]. The specific structure of HLA molecules obviously plays an important role in the induction of autoimmunity to citrullinated proteins. These studies were extended by Huizinga et al. [27] who found that HLA-DRB1 alleles encoding the SE were only associated with anti-CCP positive RA and not with anti-CCP negative RA. In contrast, anti-CCP negative RA appeared to be associated with HLA-DR3 [28]. Therefore, the presence of citrullinated antigens, together with the appropriate genetic background (the SE and probably other types of HLA), appears to be the minimum requirement for an immune response to citrullinated polypeptides to be generated [29]. Next to that, other sensitizing genetic settings resulting from polymorphisms (e.g., PTPN22 [30]) might, given their association, aid in triggering of ACPA. Conclusions and Future Perspectives The present literature describes and confirms that the anti-CCP2 test is a very useful marker for the early and accurate diagnosis of RA. Anti-CCP antibody is very specific for RA and has a sensitivity comparable to that of RF. These antibodies can be detected very early in disease and may be used as an indicator for the progression and prognosis of RA. Initially, the test was available principally as a manual ELISA method through Euro-Diagnostica and Axis-Shield and their partners. More recently, fully automated anti-CCP assays have also been made available from Phadia (UniCap Elia CCP) and Abbott Diagnostics (AxSYM anti-CCP). A very novel format is the CCPoint, a Point-of-Care test for the detection of these antibodies in whole blood in 10 min [31]. A small drop of whole blood is applied onto the sample well followed by the addition of four drops of running buffer. After 10 min, the result is read by visual inspection of the detection zone for staining of the antigen line. As no special equipment is required to perform the assay, this test will bring the detection of anti-CCP antibodies into the office of the family doctor, facilitating a very fast referral to the rheumatologist when the test gives a positive answer. Such new applications will undoubtedly further enhance the utility of the anti-CCP autoantibody system in clinical practice.

Histochem_Cell_Biol-3-1-2137947

Nucleolus: the fascinating nuclear body

Nucleoli are the prominent contrasted structures of the cell nucleus. In the nucleolus, ribosomal RNAs are synthesized, processed and assembled with ribosomal proteins. RNA polymerase I synthesizes the ribosomal RNAs and this activity is cell cycle regulated. The nucleolus reveals the functional organization of the nucleus in which the compartmentation of the different steps of ribosome biogenesis is observed whereas the nucleolar machineries are in permanent exchange with the nucleoplasm and other nuclear bodies. After mitosis, nucleolar assembly is a time and space regulated process controlled by the cell cycle. In addition, by generating a large volume in the nucleus with apparently no RNA polymerase II activity, the nucleolus creates a domain of retention/sequestration of molecules normally active outside the nucleolus. Viruses interact with the nucleolus and recruit nucleolar proteins to facilitate virus replication. The nucleolus is also a sensor of stress due to the redistribution of the ribosomal proteins in the nucleoplasm by nucleolus disruption. The nucleolus plays several crucial functions in the nucleus: in addition to its function as ribosome factory of the cells it is a multifunctional nuclear domain, and nucleolar activity is linked with several pathologies. Perspectives on the evolution of this research area are proposed. Introduction Brief history of the nucleolus An ovoid body visible in the nucleus was probably the first observation of the nucleolus more than two centuries ago by F. Fontana. Since that time, the nucleolus has been the object of intense investigation and interestingly our vision of the nucleolus has evolved with technical progress. During the nineteenth century, using light microscopy, numerous cytologists described the variability of nucleolar morphology with great precision (Montgomery 1898). In 1934, McClintock proposed that the “nucleolus is organized in the telophase through the activity of ... the nucleolar-organizing body” (McClintock 1934). Since the nucleolar-organizing body corresponds to a specific region of chromosome 6 in Zea mays, this was the first time the nucleolus was related to gene activity. In the 1950’s the presence of RNAs in the nucleolus was demonstrated, and in the 1960’s in situ hybridization techniques made it possible to identify ribosomal genes (rDNAs) in the nucleolar organizer region (NOR) (Caspersson 1950; Perry 1962; Ritossa and Spiegelman 1965). During the same period, mass isolation of nucleoli became possible leading to the biochemical characterization of nucleolar components. Based on these results it was proposed that ribosome biogenesis occurs in nucleoli. Given that the nucleolus became a subject of great interest, the “International symposium on the nucleolus—its structure and function” was organized in Montevideo in 1965 and the contributions published in Natl Cancer Inst Monogr no 23 (USA) in 1966. Since 1969, at the initiative of W. Bernhard and H. Busch, “Nucleolar Workshops” on nucleolar organization, the biochemistry of nucleolar proteins, rRNA processing as well as variability in cancer cells were regularly organized. Several books on nucleoli were published; among them, the famous “The nucleolus and ribosome biogenesis” is still a very useful source of information (Hadjiolov 1985). Between 1980 and 2000, the functional organization of the nucleolus was deciphered in large part due to the improvement of labeling by the electron microscopy (EM). Recently a new field of investigation was opened when molecules not involved in ribosome biogenesis were detected in the nucleolus (Carmo-Fonseca et al. 2000; Pederson 1998; Politz et al. 2002; Visintin and Amon 2000). In accordance with these nucleolar localizations, nucleolar mass spectrometry analyses identified ∼700 nucleolar proteins, some of them not related to ribosome biogenesis (Andersen et al. 2005). The area of plurifunctional nucleolus was opened. Consequently “The nucleolus”, a book that presents the state of the art was published (Olson 2004) as well as several reviews on the multiple functions of the nucleolus (Boisvert et al. 2007; Hernandez-Verdun 2006; Hiscox 2007; Raska et al. 2006). General information “The nucleolus: an organelle formed by the act of building a ribosome” (Mélèse and Xue 1995) reveals by its size and organization the efficiency of ribosome biogenesis. For example the nucleolus is a prominent nuclear structure in cycling cells but of limited size in the terminal stages of differentiation such as in lymphocytes or chick erythrocytes. If ribosome biogenesis is blocked, reorganization of the nucleolar components is visible in segregated nucleoli. In mammalian cells, the nucleolus is disorganized in prophase and reassembled at the end of mitosis using the nucleolar machineries from the previous cell cycle. On the contrary, in yeast the nucleolus is present and active throughout the cell cycle even though condensation of the rDNAs is necessary for transmission of the nucleolus in anaphase (D’Amours et al. 2004; Sullivan et al. 2004; Torres-Rosell et al. 2004). The nucleolus is the ribosome factory of the cell. In the nucleolus rDNAs are transcribed, the 47S precursor ribosomal RNAs (pre-rRNAs) are cleaved, processed and assembled with the 80 ribosomal proteins and the 5S RNA to form the 40S and 60S ribosomal subunits (selected reviews Gébrane-Younès et al. 2005; Hernandez-Verdun and Junéra 1995; Scheer et al. 1993; Scheer and Hock 1999; Shaw and Jordan 1995; Thiry and Goessens 1996). This complex series of maturation and processing events, presently better characterized in yeast than in higher eukaryotes is under the control of about 150 small nucleolar RNAs (snoRNAs) and 2 large RNP complexes: (1) the small subunit (SSU) processome containing the U3 snoRNAs and 40 proteins or Utps (U three proteins) required for the 40S ribosomal subunit, and (2) the large subunit (LSU) processome required for the 60S ribosomal subunit (de la Cruz et al. 2004; Fatica and Tollervey 2002; Fromont-Racine et al. 2003; Sollner-Webb et al. 1996; Tollervey 1996). The snoRNAs associated with proteins, function in the maturation of rRNAs creating two types of modified nucleotides (2′-O-methylation and pseudouridylation) and mediating endonucleolytic cleavages of pre-rRNAs (Gerbi and Borovjagin 2004). Our objective is to focus this review on the ribosome biogenesis processes occurring in the nucleoli that might help to decipher the global organization of nuclear functions. We describe nucleolar organization and dynamics, propose our view on nucleolar targeting, report the relationship between the nucleolus and the cell cycle, review particular relationships between nucleolus and virus, and nucleolus related to cancer. The nucleolus is a model of compartmentation Three main components in the active nucleolus The nucleolus has been proposed as the paradigm of nuclear functional compartmentalization (Strouboulis and Wolffe 1996). It is the site of ribosome biogenesis and in addition the nucleolar machineries are distributed in different compartments. When observed by EM, three main nucleolar components (compartments) can be discerned in mammalian cells: the fibrillar centers (FCs), the dense fibrillar component (DFC) and the granular component (GC) (Fig. 1a). The FCs are clear areas, partly or entirely surrounded by a highly contrasted region (Goessens et al. 1987), the DFC. The FCs and the DFC are embedded in the GC, mainly composed of granules of 15–20 nm in diameter. The most contrasted structures in the EM sections stained with uranyl and lead correspond to high concentrations of nucleic acids. The condensed chromatin surrounding part of the nucleolus is visible using standard or preferential staining methods and also as a network within the nucleolus (Fig. 1b). The global amount of intra-nucleolar chromatin is probably low since by light microscopy, DNA staining by DAPI excludes the nucleolus. Fig. 1The nucleolus of mammalian cells as seen by electron microscopy. a In the human HeLa cell, the three main nucleolar components are visible in a section of material fixed in glutaraldehyde and osmium tetroxyde, embedded in Epon and the section contrasted with uranyl acetate and lead citrate. FCs of different sizes are visible and the largest is indicated by an asterisk. The FCs are surrounded by the DFC and are embedded in the GC. b Preferential contrast of DNA using NAMA-Ur staining in a PtK1 cell (courtesy J. Gébrane-Younès). The nucleolus is the gray structure surrounded by highly contrasted chromatin (arrow). Some chromatin filaments are also visible inside of the nucleolus (Nu). c, d Nucleolus of rat neurones (courtesy M. J. Pébusque) in the day (c), and during the night (d) which is the active period for the nucleolus of the rat. In the nonactive period (c), the nucleolus is reticulated with small FCs (asterisk). In the active period, one giant FC is visible (d, asterisk). Bar in a: 0.5 μm and bars in b, c and d: 1 μm It has become apparent that nucleoli of different cell types exhibit a variable number of FCs of different sizes, with an inverse proportion between size and number (Hozak et al. 1989; Pébusque and Seïte 1981). Generally cells with a high rate of ribosome biogenesis possess numerous small FCs. On the contrary, cells with greatly reduced metabolic and transcription activities, present small nucleoli with one large-sized FC such as in lymphocytes and in inactive mammalian neurons (Hozàk et al. 1994; Lafarga et al. 1989). In the more active neurons, one giant FC (GFC) of 1–2 μm is observed together with small FCs (Fig. 1c, d). It was demonstrated that the GFC is enriched in the upstream binding factor, the UBF transcription factor, in a small ubiquitin-like modifier (SUMO)-1 and Ubc9 but lack ubiquitin-proteasome and 20S proteasome (Casafont et al. 2007). However, the possibility that only one FC might play a role in storage and become a GFC during intense nucleolar activity is still an open question. It is also remarkable that the tripartite nucleolar organization is not general since the nucleoli of Drosophila and insects lack FCs (Knibiehler et al. 1982; Knibiehler et al. 1984). It has been proposed that this difference in organization could be linked to the evolution of the rDNAs, in particular to the size of the intergenic sequences (Thiry and Lafontaine 2005). The localization of the nucleolar machineries is related to their function in the production of the small and large ribosome subunits. These findings have led to assigning specific functions to specific compartments of the nucleolus. Nascent transcripts appear at the junction between the FCs and DFC and accumulate in the DFC (Cmarko et al. 2000; Guillot et al. 2005; Hozàk et al. 1994; Puvion-Dutilleul et al. 1997; Shaw and Jordan 1995). This was recently confirmed in the GFC since no transcripts can be detected in these large structures (Casafont et al. 2007). Processing of the 47S pre-rRNA starts at the site of transcription in the DFC (Cmarko et al. 2000) and continues during the intra-nucleolar migration of the RNA towards the GC. The nucleolar proteins that participate in the early stages of rRNA processing, localize in the DFC, such as fibrillarin and nucleolin along with the U3 snoRNAs (Biggiogera et al. 1989; Ginisty et al. 1998; Ochs et al. 1985b; Puvion-Dutilleul et al. 1991), whereas proteins B23/NPM (nucleophosmin) and PM-Scl 100 (rrp6 in yeast) that are involved in intermediate or later stages of processing have been localized to the GC (Biggiogera et al. 1989; Gautier et al. 1994). Recent advances in the isolation of large RNP complexes by tandem affinity purification and the characterization of their constituents demonstrated that two largely independent processing machineries exist in yeast nucleoli, the SSU processome (Dragon et al. 2002; Grandi et al. 2002) and the LSU processing/assembly factors (Raué 2004). The SSU/90S processome is localized in the DFC and most of the 60S processing occurs in the GC. There is no particular domain characterized in the GC corresponding to the 43S subunit. This is most probably due to the limited events of 40S processing in the GC since the last step of processing occurs in the cytoplasm. In conclusion it seems that in the nucleoli, the vectorial distribution of the machineries successively involved in ribosome biogenesis correlates with the different processing steps of the biogenesis of the ribosome subunits. When ribosome biogenesis is active, the confinement of certain machineries in the FCs, DFC or GC makes it possible to reveal these subnucleolar constituents by immunofluorescence as illustrated for FCs (Fig. 2A), DFC (Fig. 2Ba, b), and GC (Fig. 2Bc, d). The factors associated with the rDNA transcription machinery are distributed in several foci, most frequently inside the nucleolar volume as illustrated for UBF. These foci correspond to FCs. A distribution within the network inside the nucleolus is typical of the DFC as demonstrated for fibrillarin. Labeling of the nucleolar volume excluding small areas contained within the volume is typical of the GC as illustrated for B23/NPM. These labeling patterns (FCs, DFC, GC) in the nucleoli provide a good indication of the step of ribosome biogenesis concerned and also reveal the blockage of ribosome biogenesis when this organization is disturbed (see below). Fig. 2The subnucleolar constituents revealed by fluorescence microscopy. (A) The rDNA transcription machinery, illustrated by UBF labeling, is localized in several foci corresponding to FCs in HeLa cells (a). rDNA transcription sites detected by in situ BrUTP incorporation (b), mainly colocalize with UBF as seen by the merge (c). The nucleus is visualized by Dapi staining (d). (B) HeLa cells expressing either fibrillarin-GFP fusion or DsRed-B23 fusion. Fibrillarin decorates the DFC (a, b) whereas B23 decorates the GC (c, d). In ActD-treated cells nucleoli are segregated, fibrillarin localizes in caps (e, f) contrary to B23 that localizes in the central body and outside the caps (g, h). Arrowheads point the caps. Bars: 10 μm Signature of impaired ribosome biogenesis Transient association of functionally related components appears necessary to generate a morphologically defined nucleolus with its three distinct components, thereby maintaining the nucleolus in its usual organization. This suggests that such an organization results from the activity of ribosome biogenesis. Indeed nucleolar reorganization is induced when ribosome biogenesis is impaired either by inhibiting rDNA transcription, or inhibiting rRNA processing and/or transport. Inhibition of transcription Nucleolar segregation is observed upon rDNA transcriptional arrest either in physiological conditions or induced by low doses of actinomycin D (ActD). The segregation of nucleoli is characterized by the separation of the nucleolar components that remain close to each other but no longer intermingle (Fig. 2Be–h) (for reviews see Hadjiolov 1985; Hernandez-Verdun and Junéra 1995; Scheer and Benavente 1990). The effect of ActD on nucleolar organization follows sequential changes: first the fibrillar components (FCs and DFC) condense and migrate towards the periphery of the nucleolus, after which the nucleolar components segregate to finally form a central body associated with caps (Hadjiolova et al. 1995). In the caps are several proteins related to the RNA polymerase (pol) I transcription machinery such as UBF, close to fibrillarin-containing caps. In the central body are proteins derived from the GC, some of which are progressively released, such as PM-Scl 100. It was recently demonstrated that certain nucleolar caps of segregated nucleoli could recruit factors involved in mRNA splicing. In this case, localization is induced by inhibition of both RNA pol I (rRNA transcription arrest) and RNA pol II (mRNA transcription arrest) (Shav-Tal et al. 2005). This is not observed when only RNA pol I is inhibited indicating that the composition of a segregated nucleolus can be more complex when induced by general transcription inhibition. One question that remains unanswered is how nucleolar components continue to be maintained in segregated nucleoli in spite of the absence of transcription or pre-rRNA processing. Nucleolar proteins may still be capable of forming complexes during inhibition of transcription, but why these complexes remain juxtaposed is presently unknown. Recently, it was reported that re-localization of proteins in specific caps of segregated nucleoli (after inhibition of RNA pol I and II transcription) is an energy-dependent repositioning process that requires active metabolism of the cells (Shav-Tal et al. 2005) most probably also ATP and GTP. Degradation of rRNAs A clue to the question of rRNA degradation in the nucleolus was recently proposed in yeast: a surveillance pathway that eliminates defective 60S pre-ribosomal subunits after addition of poly(A) tails was described (LaCava et al. 2005). RNA degradation appears to occur preferentially within a subnucleolar structure, the No-body, and is mediated by the exosome (Dez et al. 2006). Similarly, when the nuclear protein of the exosome rrp6 was deleted, poly(A) rRNAs and poly(A) U14 snoRNAs colocalized in one focus with Nop1 (fibrillarin in human), most probably the No-body (Carneiro et al. 2007). This body is distinct from the nucleolar body that functions in snoRNA maturation in yeast and could be a compartment where polyadenylation and degradation of nucleolar RNAs take place. This compartmentation would promote efficient recognition of rRNAs in view of degradation by the exosome (Carneiro et al. 2007). During nucleolar segregation induced by ActD in human cells, rRNAs are degraded. However, the formation of one focus containing PM-Scl 100 has not been described; it could be either an early event that was not carefully examined, or rRNA degradation could be different in yeast and mammalian cells. Disconnection of the nucleolar component Disconnection between rDNA transcription sites and the late rRNA processing proteins can be induced either by kinase inhibitors or by modifications of snoRNA domains (Chan et al. 1996; Colau et al. 2004; David-Pfeuty et al. 2001; Rubbi and Milner 2003; Sirri et al. 2002). The separation of the DFC and GC can be reversed by removal of a CK2 inhibitor, restoring nucleolar organization. The CK2 kinase is known to phosphorylate several nucleolar proteins (Meggio and Pinna 2003). We postulate that the connection between DFC and GC is controlled at least in part by phosphorylation of these proteins. This hypothesis was verified for B23 by mutation of the major site of CK2 phosphorylation (Louvet et al. 2006). In conclusion, the rRNA processing proteins can be disconnected from the rRNA transcription sites indicating that rRNA transcripts are not sufficient to attract the processing proteins in these conditions. The dynamics of nucleolar reformation and the connection between DFC and GC is ATP/GTP dependent, sensitive to temperature, and is CK2-driven. Traffic and dynamics of nucleolar actors The analysis in living cells of intranuclear dynamics has recently become possible using fluorescent fusion proteins. Time-lapse videomicroscopy can track the movement of large fluorescent complexes in the cell, and fluorescent recovery after photobleaching (FRAP) can measure the intracellular mobility or the residency time of fluorescent proteins (Lippincott-Schwartz et al. 2001). Inverse FRAP (iFRAP) quantifies the loss of fluorescence of the region of interest (ROI) after complete bleaching outside this region (Dundr et al. 2004). This constitutes a direct evaluation of the residency time of the proteins in the ROI. Another approach is the use of photoactivatable GFP (PA-GFP) to follow the traffic of the activated proteins (Patterson and Lippincott-Schwartz 2002). This process is similar to pulse-chase experiments since it makes it possible to follow a pool of labeled proteins starting at time zero. These technologies applied to nuclear dynamics have introduced new dimensions and unexpected concepts concerning nuclear functional compartmentation. The mobility of several GFP-tagged nuclear proteins (nucleolar proteins, histones, DNA binding proteins, transcription factors, splicing factors, nuclear receptors) has been estimated by FRAP and the recovery of fluorescence was slower than predicted for isolated diffusing proteins of similar size. FRAP recovery rates change with inhibition of transcription, decreased temperature and depletion of ATP indicating that recovery is correlated with nuclear activity. It was demonstrated that nucleolar proteins rapidly associate with and dissociate from nucleolar components in continuous exchanges with the nucleoplasm (Phair and Misteli 2000). The recovery curve of GFP-fibrillarin (DFC marker) in the nucleolus reached a plateau, 60 s after bleaching and the plateau indicated an immobile fraction of ∼15%. The diffusion coefficient of fibrillarin (estimated between 0.02 and 0.046 μm2 s−1) was 10 times lower in the nucleolus than in the nucleoplasm (Chen and Huang 2001; Phair and Misteli 2000; Snaar et al. 2000). This value is proposed to reflect the time of residency of fibrillarin engaged in nucleolar activity, and could explain the fact that the time of residency of fibrillarin is shorter in the Cajal body than in the nucleolus (Dundr et al. 2004). The nucleolar proteins engaged in rRNA transcription and processing (respectively UBF, B23, Nop52, nucleolin and Rpp29) also move with rapid recovery rates in the nucleolus as does fibrillarin (Chen and Huang 2001; Louvet et al. 2005). Conversely the recovery rates of ribosomal proteins are slow (∼3 times slower than nucleolar proteins). This could reflect a slower mechanism for ribosome protein assembly compared with transcription and processing (Chen and Huang 2001), or alternatively, more stable associations of ribosomal proteins with the pre-rRNAs. B23 (also designated NPM) is a multifunctional protein, abundant in the GC of the nucleolus that undergoes different phosphorylation events during the cell cycle. It was recently demonstrated by FRAP that the kinetics of B23 depends on its phosphorylation status (Negi and Olson 2006). During interphase, the half-time (t1/2) recovery of B23 is 22 s in nucleoli but when the CK2 phosphorylation site is mutated (S125A) the t1/2 increases to 44 s, and when a mutant mimicking phosphorylation charges of the four sites of mitotic CDK1 phosphorylation, the t1/2 decreases to 12 s. This could indicate that the S125A-B23 protein has a higher affinity for the nucleolar components (Negi and Olson 2006). Alternatively this could correspond to a decreased turner-over in the nucleolar complexes in correlation with the disconnection of the DFC and GC occurring by overexpression of S125A-B23 (Louvet et al. 2006). Overexpression during interphase of B23 mimicking four sites of mitotic phosphorylation increased the mobility of B23. It is tempting to propose that this results from a defect of affinity for rRNAs of these B23s as demonstrated for mitotic phosphorylation of B23 (Okuwaki et al. 2002). Inhibition of pol I transcription by ActD does not prevent traffic of nucleolar proteins. However, if the diffusion coefficient of nucleolar proteins in the nucleoplasm is similar for active and repressed pol I transcription, the traffic in segregated nucleoli changes differently for different nucleolar components. Traffic of UBF in the nucleolus is decreased by ActD, whereas it is similar for nucleolin or increased for ribosomal proteins (Chen and Huang 2001). In contrast to the well-defined nucleolar structures visible by EM, all the nucleolar proteins involved in ribosome biogenesis that have been examined, cycle between the nucleolus and the nucleoplasm in interphase cells. To summarize, it is now established that rapid diffusion of nucleolar proteins occurs in the nucleoplasm and recruitment to the nucleolus is permanent. Moreover, the difference in kinetics of several proteins shared between the nucleolus and the Cajal body suggests the existence of compartment-specific retention (Dundr et al. 2004). Targeting to the nucleolus To be localized or retained within the nucleolus In eukaryotic cells, once imported or diffused into the nucleus, some proteins distribute throughout the nucleoplasm and others are targeted to specific nuclear compartments such as nucleoli. Proteomic analyses revealed that at least 700 proteins are localized in nucleoli (Andersen et al. 2002, 2005; Leung et al. 2003). Whereas the rules and signals that govern the nuclear localization and nuclear export of proteins are now well defined, those concerning nucleolar localization are still debated. Contrary to the nuclear localization signals (NLSs), nucleolar localization signals or sequences (NoLSs) are not well characterized. Although several NoLSs have been described, no obvious consensus sequence has emerged. Nevertheless all NoLSs reported for nucleolus localizing virus proteins, such as HIV-1 Rev (Kubota et al. 1989), HIV-1 Tat (Dang and Lee 1989) and human T-cell leukaemia virus type 1 Rex (HTLV-1 Rex) (Siomi et al. 1988), and for cellular proteins such as the nucleolar protein p120 (Valdez et al. 1994), Survivin-deltaEx3 (Song and Wu 2005) and HSP70 (Dang and Lee 1989) are rich in basic residues. The capacity of numerous proteins to adopt nucleolar localization has been correlated with interaction of these proteins with B23. Owing to the ability of numerous nucleolar proteins to interact with B23 and because this major nucleolar protein shuttles constantly between the nucleus and the cytoplasm (Borer et al. 1989), it was frequently suggested that B23 might be a transporter for nucleolar proteins possessing a NoLS (Fankhauser et al. 1991; Li 1997; Valdez et al. 1994). Even if this tempting hypothesis has never been demonstrated, recent results obtained using stable U2OS-derived cell lines with reduced B23 expression levels showed that the nucleolar localization of ARF is linked to B23 (Korgaonkar et al. 2005). Indeed, reduced expression of B23 induced a partial delocalization of ARF from nucleoli to nucleoplasm. The authors therefore concluded that B23 targets ARF to nucleoli in a dose-dependent manner. Nevertheless, this result does not allow discriminating between a role for B23 in the transport of ARF from nucleoplasm to nucleoli and/or in the retention of ARF in nucleoli. A NoLS, i.e. a sequence essential for nucleolar localization, is most probably a sequence involved in nucleolar retention by interacting with a nucleolar molecule such as B23 (Lechertier et al. 2007). Indeed, recent analyses of the intranuclear dynamics of proteins in living cells revealed that nuclear proteins could diffuse within the nucleoplasm (Phair and Misteli 2000; Sprague and McNally 2005). As for the nucleolus, it was demonstrated that nucleolar proteins rapidly associate with and dissociate from nucleolar components in a continuous exchange with the nucleoplasm (Chen and Huang 2001; Dundr et al. 2004; Phair and Misteli 2000; Snaar et al. 2000). There probably exist compartment-specific retention mechanisms for proteins in nuclear bodies, implying that the residency time of a particular molecule in a given nuclear body depends on its specific interactions (Misteli 2001). In support of this possibility, we have recently shown that the fusion of a B23-interacting sequence with fibrillarin makes it possible to re-localize fibrillarin from the DFC to the GC of nucleoli where B23 is mainly localized (Lechertier et al. 2007). Similarly, by fusing the B23-interacting sequence to MafG (part of the nuclear transcription factor NF-E2 composed of both MafG and p45), NF-E2 is redirected from the nucleoplasm to the GC. Therefore, interactions most probably govern the nuclear distribution of proteins and a NoLS is very likely a nucleolar molecule-interacting sequence. However, nucleolar localization of a protein is most probably governed by several factors and the presence of a NoLS in its sequence is not sufficient to predict nucleolar localization of the protein. In particular, a nucleolar protein must first be localized in the nucleus, and consequently all mechanisms that interfere with nuclear import and/or nuclear export of a nucleolar protein will modify its localization at the steady state. A good illustration is provided by the major nucleolar protein, B23. This multifunctional protein is normally mainly located in the GC of nucleoli but exhibits an aberrant cytoplasmic localization in one-third of acute myeloid leukemias due to mutations in its C-terminal coding exon that causes a frameshift and the formation of an additional CRM1-dependent nuclear export signal (NES) (Mariano et al. 2006). Another example showing the difficulty encountered in predicting nucleolar localization is provided by the box C/D snoRNPs: it seems clear that the nucleolar localization of box C/D snoRNPs is linked to their biogenesis (Verheggen et al. 2001; Watkins et al. 2002). Indeed, by modifying the conserved stem II of the box C/D motif present in the U14 snoRNA, both the specific assembly of the box C/D snoRNP and nucleolar localization are lost (Watkins et al. 2002). Moreover, genetic depletion of one of the four core proteins, namely 15.5kD, Nop56, Nop58 and fibrillarin, also inhibits the nucleolar localization of box C/D snoRNPs (Verheggen et al. 2001). However, targeting of box C/D snoRNPs to nucleoli is not yet fully understood. Indeed, unexpectedly two nuclear export factors, PHAX and CRM1 appear to be stably associated with the U3 pre-snoRNPs (Boulon et al. 2004; Watkins et al. 2004). Boulon and coworkers proposed that U3 precursors bind PHAX, which targets the complex to the Cajal body, and that subsequently CRM1 further targets the U3 complexes to the nucleolus. Even if PHAX and CRM1 play an important role in the transport of box C/D snoRNPs to the nucleus, the possibility that these proteins may also function in the nuclear export of snoRNPs cannot be excluded (Watkins et al. 2004). This possibility is reinforced by a recent study showing that in addition to nuclear export factors, the nuclear import factor Snurportin 1 is involved in U8 box C/D snoRNP biogenesis (Watkins et al. 2007). Nucleolar localization of the components of the box C/D snoRNPs would therefore depend on the biogenesis of the box C/D snoRNP complexes, which would imply nuclear export. Control of rDNA transcription during cell cycle rDNA transcription machinery rDNAs are found in multiple, tandem, head-to-tail arrayed copies in the nucleoli of eukaryotic cells (Hadjiolov 1985). In mitotic human cells, rDNA clusters are localized on the short arm of the five pairs of chromosomes 13, 14, 15, 21 and 22 and are termed NORs. Each rDNA unit consists of a transcribed sequence and an external non-transcribed spacer (Hadjiolov 1985; Liau and Perry 1969) in which all the sequences necessary for proper RNA pol I transcription such as proximal promoters, spacer promoters and terminators are located (Hadjiolov 1985). In the rDNA promoter two important elements have been described, a CORE element and an upstream control element (UCE) (Haltiner et al. 1986; Windle and Sollner-Webb 1986; Xie et al. 1992) that function synergistically to recruit a transcriptionally competent RNA pol I complex. This complex contains in addition to RNA pol I, the upstream binding factor (UBF) (Pikaard et al. 1989; Voit et al. 1992), the selectivity factor protein complex SL1 (Learned et al. 1985) also called TIF-1B in mouse cells (Clos et al. 1986), consisting of the TATA-binding protein (TBP) and four transcription activating factors [TAFIs110, 63, 48 and 41 (Comai et al. 1994; Gorski et al. 2007; Zomerdijk et al. 1994)], the transcription initiation factor TIF-IA, the mouse homolog of Rrn3p (Bodem et al. 2000; Moorefield et al. 2000) and the transcription termination factor TTF-1 (Bartsch et al. 1988). The UBF containing HMG boxes (Bachvarov and Moss 1991; Jantzen et al. 1990) that confer a high affinity for DNA structures plays an architectural role on the rDNA promoter (Mais et al. 2005). It was proposed that UBF activates rDNA transcription because it stabilizes binding of SL1 required to recruit the initiation-competent subfraction of RNA pol I. This recruitment is achieved by interaction of UBF with the RNA pol I-associated factor PAF53 (Schnapp et al. 1994), and by interaction of SL1 with TIF-1A/Rrn3p (Miller et al. 2001). TIF-1A/Rrn3p interacts also with the RPA43 subunit of RNA pol I and thus facilitates linking between RNA pol I and SL1 complexes (Peyroche et al. 2000; Yuan et al. 2002). Following initiation, TIF-1A/Rrn3p is released and can associate with another preinitiation complex. Recycling of TIF-1A/Rrn3p requires a post-translational phosphorylation event that appears to play a role in its initiation activity (Cavanaugh et al. 2002; Zhao et al. 2003). Moreover, it was proposed that TTF-1 is not only involved in termination of transcription in cooperation with the release factor PTRF (Jansa and Grummt 1999), but also in the remodeling of ribosomal chromatin by recruiting ATP-dependent remodeling factors to the rDNA promoter (Längst et al. 1997). The nucleolar remodeling complex (NoRC) (Strohner et al. 2001), which acts in repression at the rDNA promoter level (Li et al. 2006; Santoro et al. 2002), and the transcription activator CSB (Cockayne syndrome group B protein), a DNA-dependent ATPase, interact with TTF-1 (Yuan et al. 2007). The finding that TTF-1 interacts with both CSB and NoRC suggests that competitive recruitment of CSB and NoRC may determine the epigenetic state of the rDNA. Regulation during the cell cycle It is now established that the presence of a fully active nucleolus depends on cell cycle regulators. rDNA transcription is maximum in the S and G2 phases, silent in mitosis, and slowly recovers in G1. Post-translational modifications of the RNA pol I machinery are required for the formation of a productive preinitiation complex. The phosphorylation status of several components of the RNA pol I machinery can modify the activity and interactions of these proteins and thus can modulate rDNA transcription during the cell cycle. Concerning silencing of rDNA transcription during mitosis, it is well established that some components of the rDNA transcription machinery such as SL1 (Heix et al. 1998) and TTF-1 (Sirri et al. 1999), are mitotically phosphorylated by CDK1-cyclin B. As shown in vitro, CDK1-cyclin B-mediated phosphorylation of SL1 abrogates its transcriptional activity (Heix et al. 1998). Moreover CDK1-cyclin B is necessary not only to establish repression but also to maintain it from prophase to telophase. Indeed, in vivo inhibition of CDK1-cyclin B leads to dephosphorylation of the mitotically phosphorylated forms of components of the RNA pol I machinery and restores rDNA transcription in mitotic cells (Sirri et al. 2000). On the other hand, rDNA transcription also appears regulated by CDK(s) during interphase: the increase of rDNA transcription during G1 progression depends on phosphorylation of UBF by G1-specific CDK–cyclin complexes (Voit et al. 1999), and CDK inhibitor treatments partially inhibit rDNA transcription in interphase cells (Sirri et al. 2000). Modifications of the phosphorylation status of UBF and/or TAFI110 affect the interactions between UBF and SL1 necessary for recruitment of RNA pol I (Zhai and Comai 1999). In addition to phosphorylation, it has been speculated that acetyltransferases might also regulate the activity of RNA pol I transcription factors. Indeed, two studies have demonstrated that UBF and one of the SL1 subunits are acetylated in vivo (Muth et al. 2001; Pelletier et al. 2000). Functional studies indicated that acetylated UBF is transcriptionally more active than deacetylated UBF. However, acetylation of UBF does not affect its DNA binding activity as shown for other transcription factors, and it is unclear how this post-transcriptional modification modulates UBF activity. The TAFI68 subunit of SL1 is specifically acetylated by recruitment of PCAF (p300/CBP associated factor) to the rDNA promoter. In vitro analyses indicate that acetylation of TAFI68 is likely to increase the activity of SL1 facilitating interaction of the complex with DNA. Sirtuins, the human homologues of the yeast Sir2 (silent information regulator) with NAD-dependent deacetylase and ADP-ribosyltranferase activity, have recently been implicated in the regulation of the RNA pol I machinery. In particular, nuclear sirtuin1 deacetylates TAFI68 and represses RNA pol I transcription in vitro (Muth et al. 2001). Conversely, the nucleolar sirtuin7 is described as activator of rDNA transcription by increasing RNA pol I recruitment to the rDNA, but no substrates of such activity have as yet been identified (Ford et al. 2006). Additional in vivo approaches are necessary to better understand the role of sirtuins in the regulation of rDNA transcription. SUMO modification is reported to influence the assembly of transcription factors on promoters and the recruitment of chromatin-modifying enzymes, and is often associated with transcriptional repression (Gill 2004). Recently, the colocalization of SUMO-1 and UBF in the GFC (Casafont et al. 2007) of neuronal cells and the nucleolar localization of the sentrin/SUMO-specific proteases, SENP3 and SENP5 (Gong and Yeh 2006; Nishida et al. 2000) suggest a potential role of sumoylation on the regulation of rDNA transcription. Further studies of the identification of sumoylated nucleolar transcription factors will be necessary to verify this possibility. Nucleolar assembly and disassembly In higher eukaryotic cells at the beginning of mitosis when rDNA transcription is repressed, the nucleoli disassemble and are no longer observed throughout mitosis. Conversely nucleoli assemble at the exit from mitosis concomitantly with restoration of rDNA transcription and are functionally active throughout interphase. Disassembly in prophase In late prophase when mitotic repression of rDNA transcription occurs, the rDNA transcription machinery remains associated with rDNAs in the NORs as revealed by the analysis of different components at the steady state (Roussel et al. 1993, 1996; Sirri et al. 1999). Nevertheless, more recent quantitative kinetic analyses have revealed that some RNA pol I subunits, including RPA39, RPA16 and RPA194, transiently dissociate from the NORs during metaphase and reappear in anaphase (Chen et al. 2005; Leung et al. 2004). As for the mechanism that governs disassembly of nucleoli in prophase, it may be assumed that it is linked to repression of rDNA transcription, most probably caused by CDK1-cyclin B-directed phosphorylation of components of the rDNA transcription machinery (Heix et al. 1998; Sirri et al. 1999). At the beginning of prophase, the components of the pre-rRNA processing machinery do not remain in the vicinity of the rDNAs (Gautier et al. 1992) but become partially distributed over the surface of all the chromosomes (reviewed in Hernandez-Verdun et al. 1993). The nucleolar proteins that relocate to the chromosome periphery are components of the DFC and GC of the active nucleolus. In living cells, nucleolar proteins tagged with GFP are concentrated around the chromosomes during mitosis and migrate with the chromosomes (Savino et al. 2001). However, the mechanisms maintaining interactions of nucleolar processing proteins with chromosomes during mitosis have not been characterized. The colocalization of the different factors involved in pre-rRNA processing suggests that processing complexes are at least to some extent maintained during mitosis. It is as yet unknown whether migration of the nucleolar processing proteins occurring at the onset of mitosis (Fan and Penman 1971) takes place as a consequence of the arrest of pre-rRNA synthesis or whether it is also regulated. Indeed, it is noticeable that (1) during prophase, the components of the rRNA processing machinery appear to be delocalized before total repression of rDNA transcription occurs, and (2) the most recently synthesized pre-rRNAs accumulate as partially processed 45S pre-rRNAs (Dousset et al. 2000) suggesting that total repression of pre-rRNA processing could occur prior to total repression of rDNA transcription. These observations therefore raise the possibility that rDNA transcription and pre-rRNA processing are both repressed during prophase by distinct mechanisms. Assembly in telophase Nucleoli assemble at the exit from mitosis concomitantly with restoration of rDNA transcription at the level of competent NORs (Roussel et al. 1996). Until recently it was admitted that transcriptionally active rDNAs, serving as nucleation sites, possessed by themselves the ability to organize the nucleoli (Scheer and Weisenberger 1994). Results obtained in the laboratory showed that (1) reactivation of rDNA transcription in mitotic cells does not lead to the formation of nucleoli (Sirri et al. 2000), (2) initiation of nucleolar assembly occurs independently of rDNA transcription (Dousset et al. 2000), and (3) at the exit from mitosis nucleologenesis is impaired in the presence of a CDK inhibitor even if rDNAs are actively transcribed (Sirri et al. 2002). Consequently, the formation of functional nucleoli at the exit from mitosis is not governed solely by the resumption of rDNA transcription. Based on previous studies (Sirri et al. 2000, 2002), we propose that the formation of nucleoli is a process regulated by CDK(s) at two levels: resumption of rDNA transcription but also restoration of rRNA processing. In anaphase, early and late processing proteins (respectively fibrillarin, and Bop1, B23, Nop52) are homogeneously distributed around the chromosomes. During telophase and early G1, along the translocation pathway between chromosome periphery and transcription sites, processing proteins concentrate in foci designated prenucleolar bodies (PNBs), first described in plant cells (Stevens 1965). PNB formation is a general phenomenon occurring during the recruitment of the nucleolar processing proteins at exit from mitosis (Angelier et al. 2005; Azum-Gélade et al. 1994; Dundr et al. 2000; Jiménez-Garcia et al. 1994; Ochs et al. 1985a; Savino et al. 2001). This appears to be a cell cycle regulated process since when the nucleolar function is established during interphase, recruitment of processing proteins is not associated with PNB formation. Inactivation of CDK1-cyclin B occurring at the end of mitosis induces the first events of nucleologenesis. Strikingly, fibrillarin concentrates in PNBs and rDNA clusters when decrease in CDK1-cyclin B activity overcomes the mitotic repression of RNA pol I transcription (Clute and Pines 1999), while Nop52 and other GC proteins are recruited later on transcription sites. This late recruitment is under the control of cyclin-dependent kinases since CDK inhibitors block this process (Sirri et al. 2002). Thus, it seems that recruitment of the processing machinery at the time of nucleolar assembly is a regulated process most probably dependent on cell cycle progression. This provides a physiological situation to investigate the formation, control and dynamics of nuclear bodies. The dynamics of the processing nucleolar proteins was analyzed at the transition mitosis/interphase using rapid time-lapse video microscopy (Fig. 3). The first detectable assembly of proteins in foci occurred on the surface of the chromosome during telophase (Savino et al. 2001), followed by the progressive delivery of proteins to nucleoli ensured by progressive and sequential release of proteins from PNBs (Dundr et al. 2000). Based on the observations of different fixed cells, it was concluded that the early processing proteins are recruited first on transcription sites while the majority of the late processing proteins are still in PNBs (Fomproix et al. 1998; Savino et al. 1999). This sequence of events was confirmed in living HeLa cells. Fibrillarin resides briefly in PNBs (∼15 min) before recruitment to the nucleolus, while Nop52 is maintained longer in PNBs (∼80 min) (Savino et al. 2001). The relative dynamics of early and late rRNA processing proteins at the time of PNB formation was examined using co-expression of GFP-fibrillarin and DsRed-B23 (Angelier et al. 2005). Once near the poles, 1–2 min after the onset of telophase, numerous bright fluorescent foci containing both GFP-fibrillarin and DsRed-B23 appeared almost simultaneously. For about 10 min, the relative amount of B23 in foci was five to six times higher than that of the dispersed proteins whereas the amount of fibrillarin in the same foci was three to four times higher than that of dispersed proteins. Subsequently, fibrillarin was released while B23 was still present in the foci. This clearly illustrates the presence of the two types of nucleolar processing proteins in the same PNBs and suggests differential sorting of these proteins. Conversely in the same conditions, similar dynamics and flows of GFP-Nop52 and DsRed-B23 were observed. Thus the processing proteins passed through the same PNBs and were released simultaneously suggesting that these proteins could form complexes in PNBs. Fig. 3At the exit from mitosis, the dynamics of DsRed-B23 is followed in living cell. In telophase (0 min), the B23 signal is visible in small foci. These foci corresponding to PNBs are clearly visible 20 min later. The B23-containing PNBs are distributed in the nucleoplasm and B23 is progressively recruited in the incipient nucleolus (40 min). Nu nucleolus Time-lapse analysis of fluorescence resonance energy transfer (FRET) was chosen to determine whether nucleolar processing proteins interact along the recruitment pathway. The apparatus used to determine FRET performs tdFLIM (time domain fluorescence lifetime imaging microscopy) by the time and space-correlated single-photon counting method (Emiliani et al. 2003). This technique directly yields the picosecond time-resolved fluorescence decay for every pixel by counting and sampling single emitted photons. Positive FRET between GFP-Nop52 and DsRed-B23 in nucleoli indicates that the distance and most probably the interactions between the proteins can be evaluated by this approach (Angelier et al. 2005). Since it is possible to detect FRET between B23 and Nop52 in nucleoli, FRET was tracked during the recruitment of these proteins into nucleoli from anaphase to early G1. FRET was never detected during anaphase at the periphery of the chromosomes whereas it was registered in 20% of the PNBs at the beginning of telophase, in about 40% at the end of telophase, and in 55% in early G1. Thus, interaction between GFP-Nop52 and DsRed-B23 was established progressively in PNBs, as the number of PNBs exhibiting FRET increased. Such data indicate that Nop52 and B23 did not interact until they were recruited in PNBs. It is noteworthy that a given PNB can alternatively present or not present FRET. Based on the behavior of these two proteins, one possibility is that late rRNA processing proteins already interact in PNBs. Were this to be confirmed for other rRNA processing complexes, PNBs could be proposed as assembly platforms of processing complexes at this step of the cell cycle. It would be very interesting to establish whether this role can be extended to the early rRNA processing machinery (Angelier et al. 2005). In conclusion, assembly of the nucleolus requires reactivation of the rDNA transcription machinery, and also recruitment and reactivation of the pre-rRNA processing machinery. Indeed cells exiting from mitosis in the presence of a CDK inhibitor exhibit neither relocalization of the late pre-rRNA processing components from PNBs to rDNA transcription sites, resumption of proper rRNA processing, nor formation of functional nucleoli. Nucleolus and cancer The link between cell proliferation, cancer and nucleolar activity has been well established during the past several decades (more than 5,000 references). Half of the studies related to the nucleolus and cancer are dedicated to the prognostic value of AgNOR staining, a technique revealing the amount of nucleolar proteins. The aim of this technique is to evaluate the proliferation potential of cancer cells by measuring nucleolar activity. B23, nucleolin, UBF and subunits of RNA pol I were found to be the argyrophilic proteins responsible for the silver-staining properties of nucleolar structures (Roussel et al. 1992; Roussel and Hernandez-Verdun 1994). In interphase cells, the amount of major AgNOR proteins, B23 and nucleolin, is high in S–G2 and low in G1 phases and thus a higher value of AgNOR corresponds to actively cycling cells (Sirri et al. 1997). Standardization of the AgNOR staining method permits routine application of this technique for clinical purposes. The size of the nucleolus is generally enlarged in cancer cells, and this has been correlated with cell proliferation. A new field of research was recently opened by the discovery that several tumor suppressors and proto-oncogenes affect the production of ribosomes (Ruggero and Pandolfi 2003). rRNA synthesis is enhanced by c-Myc (Arabi et al. 2005) and it was proposed that this stimulation is a key pathway driving cell growth and tumorigenesis (Grandori et al. 2005). On the contrary, the decrease of ribosome production induces apoptosis in a p53-dependent or independent manner (David-Pfeuty et al. 2001; Pestov et al. 2001) and the disruption of the nucleolus mediates p53 stabilization (Rubbi and Milner 2003). The cross talk between the p53 pathway and the nucleolus is at least in part mediated by localization of Mdm2 in the nucleolus, an E3 ubiquitin ligase involved in p53 degradation. Nucleostemin, a nucleolar protein discovered in stem cells and in cancer cells interacts with p53 (Tsai and McKay 2005, 2002). It was proposed that nucleostemin might regulate p53 function through shuttling between the nucleolus and the nucleoplasm. The major nucleolar protein B23 is directly implicated in cancer pathogenesis as demonstrated by mutation of the gene in a number of hematological disorders (Grisendi et al. 2005). Importantly, in acute promyelocytic leukemia, the fusion protein NPM/RARα localizes in the nucleolus indicating a role of this nucleolar protein in this disease (Rego et al. 2006). Nucleolus and virus Within the last few years, increasing evidence has revealed that viruses require the nucleus and in particular the nucleolus to target proteins indispensable for their replication. An increasing number of key proteins from both DNA- and RNA-containing viruses are localized in the nucleolus: viruses of the family Herpesviridae, Adenoviridae, Hepadnaviridae, Retroviridae, Rhabdoviridae, Orthomyxoviridae, Potyviridae, Coronaviridae and Flaviviridae, encode such proteins. Viruses have developed different strategies to facilitate targeting of their proteins to the nucleolus: (1) it was reported that the sequences of certain viral proteins harbor NoLS and NES (Harris and Hope 2000; Hiscox 2007; Kann et al. 2007). Recently (Reed et al. 2006) it was demonstrated by mutagenesis that the nucleocapsid (N) protein of infectious bronchitis virus (IBV), presents an 8 amino acid-long motif that functions as NoLS, and is necessary and sufficient for nucleolar retention of the N protein and colocalization with nucleolin and fibrillarin; the NoLS is required for interaction with cell factors. (2) Other viral proteins present sequences rich in arginine–lysine (Ghorbel et al. 2006; Reed et al. 2006) known to be nucleolar retention signals; generally, these sequences overlap the NLS. (3) Some viral proteins that target the nucleolus present motifs with affinity for double-stranded RNA (dsRNA), for RNA binding or for DNA binding (Melen et al. 2007). (4) Other studies showed that nucleolar localization of viral proteins, is cell cycle-dependent (Cawood et al. 2007); using synchronization studies coupled to live cell confocal microscopy, the authors demonstrated that the concentration of N protein in the nucleolus was higher in the G2/M phase than in other phases, and that in this phase the protein was more mobile in the nucleoplasm. In all the cases examined, the viral proteins depend on cell factors to successfully shuttle between the nucleolus and the cytoplasm. Why must viral proteins target to the nucleolus? The answer to this question is not clear; however, different authors had reported that such viral proteins are involved both in replication of the viral genome, and in transcriptional and post-transcriptional regulation of viral genome expression (Dang and Lee 1989; Pyper et al. 1998). For example, some plants viruses are known to encode a protein designated movement protein, responsible for long-distance movement of the viral RNA through the phloem (Ryabov et al. 1999). Movement strictly depends on the interaction of the viral movement protein with the nucleolus and the Cajal bodies, which contain snRNPs and snoRNPs (Kim et al. 2007a, b). The open reading frame (ORF) 3 of Groundnut rosette virus is one such protein; it is first localized in Cajal bodies and forms Cajal body-like structures, it is then localized in the nucleolus when the Cajal body-like structures fuse with the nucleolus, and finally it exits to the cytoplasm (Kim et al. 2007b). Another study showed that this shuttling is indispensable to form the RNPs essential for systemic virus infection (Kim et al. 2007a). In this process, the interaction of the viral ORF3 with fibrillarin is absolutely required. Interestingly, silencing of the fibrillarin gene blocks long-distance movement of the virus but does not affect virus replication and movement via plasmodesmata. Because the mobility of nucleolar components depends on the interactions and functions of the components (Olson and Dundr 2005), we suggest that targeting of viral proteins to the nucleolus could help viral protein traffic and diffusion of viral infection. The activity of the human immunodeficiency virus (HIV)-1 Rev protein is essential for virus replication. Its subcellular localization is nucleolar, but it has the ability to shuttle continuously between the nucleus and the cytoplasm (Felber et al. 1989; Kalland et al. 1994). Rev possesses both an NES and an NLS; the NLS is associated with importin-β as well as with B23 (Fankhauser et al. 1991; Henderson and Percipalle 1997). Rev-GFP movement in the nucleolus is very slow, implying that it is attached to affinity binding sites in this subcellular compartment (Daelemans et al. 2004). In addition, the transport of Rev from the nucleolus to the cytoplasm can be affected negatively by NF90, a cellular protein that colocalizes with Rev in the nucleolus (Urcuqui-Inchima et al. 2006) (Fig. 4). This indicates that the transport of HIV transcripts by Rev to the cytoplasm is a regulated process. Because Rev is concentrated in the nucleolus, it was suggested that the passage of Rev to the nucleolus is an indispensable step for Rev function, and hence for HIV-1 replication. Indeed, based on HIV-1 RNA trafficking through the nucleolus, this organelle is an essential participant of HIV-1 RNA export (Michienzi et al. 2000). Fig. 4HIV Rev-GFP and NF90-RFP fusions were expressed in HeLa cells. Both proteins colocalize in nucleoli as seen by the merge. The nucleus is visualized by Dapi staining. Bars: 10 μm As discussed above for Rev, it has been shown that the Herpes virus saimiri ORF57 protein is required for nuclear export of viral intronless mRNAs (Boyne and Whitehouse 2006). In addition, the expression of ORF57 induces nuclear trafficking, which is essential for nuclear export of such RNAs; the human transcription/export protein involved in mRNA export, is redistributed to the nucleolus in the presence of the ORF57 protein. Based on these findings, the authors concluded that the nucleolus is required for nuclear export of the viral mRNAs. What are the consequences for the cells of the passage of viral proteins via the nucleolus? It is known that all viruses whether with DNA or RNA genomes interfere with the cell cycle, affecting host-cell functions and increasing the efficiency of virus replication. The data obtained suggest that targeting of virus proteins to the nucleolus not only facilitates virus replication, but may also be required for pathogenic processes. Recent studies following infection by IBV, revealed a change in the morphology and protein content of the nucleolus (Dove et al. 2006). This included an enlarged FC and an increase in protein content; interestingly, the tumor suppressor protein p53, normally localized in the nucleus in virus infected cells, was redistributed mainly in the cytoplasm. The Hepatitis B virus (HBV) core antigen (HBcAg) is responsible for export of the virus with a mature genome (Yuan et al. 1999a, b). Indeed a change from isoleucine to leucine in position 97 (I97L) of HBcAg causes the cell to release virus particles with immature genomes. HBcAg with a mutation in position 97 (I97E or I97W) has been detected in the nucleolus colocalizing with nucleolin and B23, and this colocalization was often related with binucleated cells or apoptosis (Ning and Shih 2004), suggesting that the localization of HBcAg in the nucleolus could perturb cytokinesis. The authors propose that this event may be associated with liver pathogenesis. Some factors expressed by west nile virus (WNV) such as NS2B and NS3 and the WNV capsid (WNVCp) participate in WNV-mediated apoptosis (Oh et al. 2006; Ramanathan et al. 2006). It is well known that p53 is activated in response to oncogenic or DNA damaging stresses, inducing cell cycle arrest and apoptosis (Harris and Levine 2005). In normal conditions, HDM2 targets p53 and blocks abnormal accumulation of p53 by HDM2-mediated ubiquitinylation, followed by 26S proteasome-dependent degradation of p53 (Haupt et al. 1997; Kubbutat et al. 1997). Recently it was demonstrated that WNVCp could bind to and sequester HDM2 in the nucleolus, blocking p53-HDM2 complex formation (Yang et al. 2007). This phenomenon causes stabilization of p53 and Bax activation and thereafter apoptosis. In addition, the authors show that WNVCp is able to induce apoptosis-dependent processes, suggesting that the viral protein mediates apoptosis through p53-dependent mechanisms by retention of HDM2 in the nucleolus. Remarks and perspectives The conclusions are based on the perspectives and the tendency that can be anticipated from the present research in the field of the nucleolus. We propose that in the future, a better understanding of the complexity and variability of ribosome biogenesis will need to be established. For example, the difference between the information available in yeast and mammalian cells is of major importance. The different steps of ribosome biogenesis and protein complexes are well characterized in yeast due to easy access of mutants. Similarly, Miller chromatin spreading for electron microscopy in yeast strains carrying mutations reveals the coupling of RNA pol I transcription with rRNA processing (Schneider et al. 2007). Additionally, the compaction into SSU processomes of pre-18S ribosomal RNA before cleavage was observed on Miller spreads (Osheim et al. 2004). There is presently no comparable information for the mammalian genes. Yet the tendency is to generalize and suppose that the information is similar in the two models. In the future, differences will most probably be revealed as well as the complexity of the regulation in differentiated cells. Along this line, it was demonstrated that basonuclin, a cell-type-specific rDNA regulator transcribes only one subset of rDNAs of a cell (Zhang et al. 2007b). In such a differentiated cell, it remains to be established how the subset of rDNA repeats is selected. The nucleolus was proposed to be a domain of the sequestration of molecules that normally operate outside this organelle, mainly in the nucleoplasm. Sequestration in the yeast nucleolus of the phosphatase cdc14 and its release into the cytoplasm at anaphase was demonstrated to be a key event in cell cycle progression (for a review see Cockell and Gasser 1999; Guarente 2000; Visintin and Amon 2000). However, it is important to recall that there is no nucleolus during mitosis in mammalian cells. In mammalian interphase cells, the nucleolus is a domain of retention of molecules related to cell cycle, life span, and apoptosis, and is in particular an actor of the p53-dependent pathway. Recently nucleolar retention of the Hand1 transcription factor was observed in trophoblast stem cells (Martindill et al. 2007). Phosphorylation of Hand1 induced nucleolar to nucleoplasm translocation of Hand1 and commitment of stem cells to differentiate into giant cells. Hand1 translocation to the nucleoplasm might regulate a crucial step of stem cell differentiation into polyploid giant cells but the targets of Hand1 in the nucleoplasm are still undefined. The nucleolus is generally surrounded by highly condensed chromatin first described in rat hepatocytes and presently known as heterochromatin. By following the movements of chromosome sequences introduced in different sites in chromosomes of living cells, it was demonstrated that loci at nucleoli periphery and nuclear periphery are less mobile than in other sites. Disruption of the nucleoli by a CK2 inhibitor increases the mobility of the perinucleolar loci. It was proposed that the nucleolus and nuclear periphery could maintain the three-dimensional organization of chromatin in the nucleus (Chubb et al. 2002). Recently the perinucleolar ring of chromatin was brought to the fore when its role in the maintenance of inactive X (Xi) was demonstrated (Zhang et al. 2007a). During middle and late S phase, Xi contacts the nucleolar periphery when it is replicated during the cell cycle. It was discovered that the perinucleolar chromatin is enriched in Snf2h, the catalytic subunit of a remodeling complex required for replication of heterochromatin. These observations demonstrate the role of the perinucleolar compartment in maintaining the epigenetic state of Xi (Zhang et al. 2007a). The presence of inactive rDNA repeats in perinucleolar heterochromatin is known in many plant cells and in Drosophila. It was recently demonstrated that disruption in Drosophila of histone H3K9 methylation, a marker of heterochromatin, induced nucleolar disorganization and decondensation, and disorganization of rDNA repeats (Peng and Karpen 2007). The authors suggest that condensation of a part of the rDNA copies into heterochromatin could be a general strategy against recombination of these highly repeated genes. For long, interest concerning the nucleolus was to establish how efficient ribosome biogenesis occurs and the link of this function with the cell cycle. More recently the effect of the disruption of ribosome biogenesis appeared very important when it was proposed that the nucleolus is a sensor of stress (Rubbi and Milner 2003). Indeed disruption of ribosome biogenesis releases ribosomal proteins from the nucleolus that bind to MDM2 and inhibit p53 degradation (Lindstrom et al. 2007). A connection between ribosomal stress and p53-dependent cell cycle arrest is now proposed (Gilkes et al. 2006). Considering the diversity of the recent information gathered on the nucleolus, it is clear that this is a very dynamic and rapidly progressing research area. The most promising aspect is the contribution of new models (pseudo-NORs, Prieto and McStay 2007), new species (not only yeast), new approaches (Miller spreads using mutants, proteomics) and new questions (for instance the role of siRNAs or antisens RNAs in the activity of the nucleolus).

Calcif_Tissue_Int-3-1-1914229

Use of β-Blockers and the Risk of Hip/Femur Fracture in the United Kingdom and The Netherlands

Data from in vivo studies have indicated a role for β-blockers in the prevention of bone loss. Some epidemiological studies have found protective effects of β-blockers on fracture risk. However, there is limited information on the association with cumulative dose and type of β-blockers used. We conducted two case-control studies using data from the UK General Practice Research Database (GPRD) and the Dutch PHARMO Record Linkage System (RLS). Cases were patients with a first hip or femur fracture; controls were individually matched on practice/region, gender, year of birth, and calendar time. Current use of β-blockers was defined as a prescription in 90 days before the index date. We adjusted for medical conditions and drugs associated with falling or bone mineral density. Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated using conditional logistic regression analysis. The study population included 22,247 cases and controls in the GPRD and 6,763 cases and 26,341 controls in the PHARMO RLS. Current use of β-blockers was associated with a reduced risk of hip/femur fracture in both the GPRD (adjusted OR = 0.82, 95% CI 0.74–0.91) and PHARMO RLS (adjusted OR = 0.87, 95% CI 0.80–0.95) study populations. However, this reduction of risk was not associated with cumulative dose, lipophilicity, or receptor selectivity of β-blockers. The protective effect of β-blockers was only present among patients with a history of use of other antihypertensive agents (GPRD adjusted OR = 0.72, 95% CI 0.64–0.83; PHARMO RLS adjusted OR = 0.76, 95% CI 0.67–0.86) but not in patients using β-blockers only (GPRD adjusted OR = 0.97, 95% CI 0.82–1.14; PHARMO RLS adjusted OR = 1.01, 95% CI 0.90–1.14). Also, in patients with a history of use of other antihypertensive agents, no dose-response relationship with β-blocker use was found. The effect was constant with cumulative dose and the OR was below 1.0 even among patients who just started treatment with β-blockers. As the mechanism by which β-blockers could influence bone mineral density is likely to need some time to exert a clinically relevant effect, all these finding suggests that the association between β-blockers and fracture risk is not causal. Bone remodeling comprises a phase of resorption by osteoclasts and a phase of formation by osteoblasts. Recent studies have shown bone metabolism to be mediated through the autonomic central nervous system. Leptin, a hormone produced in fat cells to signal energy insufficiency, regulates bone remodeling by modulating osteoblast proliferation and subsequent osteoclast activation via the osteoclast differentiation factor receptor activator of nuclear factor κB ligand (RANKL). The antiosteogenic effect of leptin is not present in β2-adrenergic receptor-deficient mice, which had actual increases in bone mineral density (BMD) [1–5]. Data from these in vivo studies indicate a role for β-blockers in the prevention of bone loss. In the early 1990s, propranolol was found to increase bone formation [6]. Some observational studies have reported that use of β-blockers was associated with a decreased risk of fractures [7–9], conflicting with other studies which found no association with fractures [10–12]. Studies on the effects of β-blockers on subclinical endpoints, like BMD or biochemical markers of bone resorption, have also yielded inconsistent results [7, 10, 12–14]. A possible role for β-blockers in the prevention of fractures is of major clinical interest, given that fractures are a major source of morbidity, disability, hospitalization, and mortality. One of the most serious fractures resulting from accidental falls is hip fracture [15]. However, there is still a lack of knowledge with respect to the effects of cumulative dose and type of β-blockers used. Thus, the objective of this study was to assess the strength of the association between use of β-blockers and risk of hip/femur fractures using data from two different large population-based databases in the United Kingdom and The Netherlands. Materials and Methods Setting Data for this study were obtained from the UK General Practice Research Database (GPRD) and the Dutch PHARMO Record Linkage System (RLS). The GPRD contains the computerized medical records of general practices across the United Kingdom (http://www.gprd.com). Approximately 6% of the total registered population of England and Wales is represented in the database, and it includes a cumulative total of over 5 million adult patients. The age and sex distribution of patients enrolled is representative of the general English and Welsh populations. Patient details accrued in the GPRD include demographic information, diagnoses, prescription details, preventive care provided, referrals to specialist care, hospital admissions, and related major outcomes [16]. Clinical data are stored and retrieved by means of Oxford Medical Information Systems and Read codes for diseases or causes of morbidity and mortality that are cross-referenced to the International Classification of Diseases (ICD-9). Several independent validation studies have shown that the GPRD has a high level of completeness and validity, including for hip fractures [17, 18]. The PHARMO RLS includes the demographic details and complete medication history of 950,000 community-dwelling residents of more than 25 population-defined areas in The Netherlands from 1985 onward. It is further linked to hospital admission records as well as several other health registries, including pathology, clinical laboratory findings, and general practitioner data (http://www.pharmo.nl). Since the majority of all patients in The Netherlands are registered only with one community pharmacy, independently of prescriber, pharmacy records are virtually complete with regard to prescription drugs. For this study, drug-dispensing and hospitalization data were used. The computerized histories record information on the type of drug dispensed, dispensing date, prescriber, amount dispensed, and prescribed dosage regimen. Hospital discharge records include detailed information on the primary and secondary discharge diagnoses; diagnostic, surgical, and treatment procedures; type and frequency of consultations with medical specialists; and dates of hospital admission and discharge. All diagnoses are coded according to the ICD-9-CM [19]. Definition of Cases and Controls GPRD A case-control study was conducted using GPRD data collected from January 1987 to July 1999. The details of this study have been described elsewhere [20, 21]. Briefly, cases were defined as patients aged 18 years and older with a first record of a hospital admission for a hip/femur fracture (ICD-9 codes 820–821) recorded in their medical records between the enrollment date of their practice in the GPRD and the end of data collection. The date of the occurrence of the hip/femur fracture was the index date. Each case was matched by year of birth, sex, medical practice, and calendar time to one control patient without a history of a fracture. If no eligible control was available, the age criterion was expanded consecutively at 1-yearly intervals to a maximum of 10 years. If no eligible control patient could be found, then an age- and sex-matched control patient from another practice was selected. PHARMO RLS Cases were patients aged 18 years and older with a first admission for a hip/femur fracture between January 1, 1991, and December 31, 2002. The date of the hospital admission was the index date. Up to four control patients, who did not have a history of any type of fracture, were matched to each case by year of birth, gender, region, and calendar time. Exposure Assessment For each patient, we identified all prescriptions for (GPRD) and dispensing of (PHARMO RLS) β-blockers prior to the index date. Current users were defined as patients who had a prescription/dispensing for β-blockers within 3 months prior to the index date. Recent users received a last prescription/dispensing 3–6 months before the index date, past users 6–12 months before the index date, and distant past users more than 12 months before the index date. The last prescribed daily dose prior to the index date was obtained from the written dosage instructions. For each patient, cumulative exposure to β-blockers ever before the index date was calculated. The effect of cumulative dose was assessed both regardless of the timing of use and stratified to current, recent, past, and distant past use. Thus, current users could be classified to a low previous cumulative exposure (e.g., one or two 30-day prescriptions) or a high previous cumulative exposure (e.g., 100 30-day prescriptions). The estimated daily dose for each class of β-blocker was expressed as a fraction of the World Health Organization defined daily dose (DDD). A DDD is defined as the assumed average maintenance dose per day for a drug if used for its main indication in adults [22]. DDD equivalents can be used to compare drugs within a certain therapeutic group. In this study, we converted the DDD equivalents to milligrams of metoprolol, similar to the approach used in previous studies [21]. Furthermore, β-blockers were categorized according to receptor selectivity and lipophilicity based on data in handbooks on clinical pharmacology and therapeutics. Statistical Analysis The strength of the association between use of β-blockers and risk of hip/femur fractures was estimated using conditional logistic regression analysis (SAS version 9.1.3, PHREG procedure; SAS Institute, Cary, NC) and expressed as odds ratios (ORs) and 95% confidence intervals (CIs). Final regression models were determined by stepwise backward elimination using a significance level of 0.05. Smoothing spline regression plots were used to visualize the effect of cumulative β-blocker dose on risk of hip/femur fractures [23]. In our analysis, we controlled for a wide range of clinical variables that have been associated with risk of falls or fractures. In the GPRD study, we included the following variables in the final model: history of diabetes mellitus, rheumatoid arthritis, hyperthyroidism, congestive heart failure, seizures, anemia, dementia, depression, psychotic disorder, cerebrovascular accident, chronic obstructive pulmonary disease, osteoporosis, and a record of back pain or falls in the year before the index date. Furthermore, prescriptions, in the 6 months prior to the index date, for anticonvulsants, nonsteroidal anti-inflammatory drugs (NSAIDs), methotrexate, hormone replacement therapy, other hypertensive drugs (low-ceiling diuretics, renin-angiotensin-aldosterone system [RAAS] inhibitors, calcium channel blockers), anxiolytics/hypnotics, antipsychotics, antidepressants, anti-Parkinson drugs, oral and inhaled glucocorticoids, and bronchodilators as well as information on body mass index (BMI) were retained in the model. In the PHARMO RLS study, variables included in the final model were a dispensing of benzodiazepines in the 3 months prior to the index date, or, in the 6 months prior to the index date, a dispensing of oral glucocorticoids, inhaled glucocorticoids, bronchodilators, statins, BMD-modifying drugs, hormone replacement therapy, antipsychotics, antidepressants, opioids, antiepileptics, other hypertensive drugs (low-ceiling diuretics, RAAS inhibitors, calcium channel blockers), antidiabetics, laxants, disease-modifying antirheumatic drugs, NSAIDs, or metoclopramide. A history of hospital admission for cerebrovascular disease, cancer, endocrine disorders, inflammatory bowel disease and other bowel diseases, obstructive airway disease, musculoskeletal and connective tissue diseases, anemia, and skin diseases prior to the index date were also included in the final model. Results The study population in the GPRD comprised 22,247 cases and 22,247 controls, whereas in the PHARMO RLS 6,763 cases and 26,341 controls were identified. The characteristics of both populations are displayed in Table 1. The sex and age distributions of cases were similar in the two case-control sets, although more cases in the GPRD set were aged over 80 years. Table 1.Characteristics of hip/femur fracture cases and controls in the GPRD and PHARMO RLSCharacteristicGPRDPHARMO RLSCases (n = 22,247)Controls (n = 22,247)Cases (n = 6,763)Controls (n = 26,341)Gender  Women75.8%75.8%72.9%72.7%Age (years)  <6513.9%13.9%15.8%16.2%  65–7930.8%32.2%36.6%37.3%  ≥8055.2%53.9%47.6%46.5%  Smokinga22.1%20.6%No data availableBMIb  20–2546.6%42.4%  <2018.0%9.7%No data available  >2535.4%47.9%Antidepressants13.0%7.2%9.5%5.1%Oral glucocorticoids7.2%4.4%5.4%3.5%Thiazides11.9%12.9%12.1%11.4%Nitrates6.9%7.4%9.4%9.1%Hormone replacement0.6%1.2%1.1%1.3%a No data on smoking status for 45% of GPRD study populationb No data on BMI for 58.5% of GPRD study population Current use of β-blockers was associated with a significantly decreased risk of hip/femur fracture in both databases, whereas recent and past use was not (Table 2). Adjusted ORs for current β-blocker users were 0.83 (95% CI 0.75–0.92) in the GPRD and 0.87 (95% CI 0.80–0.95) in the PHARMO RLS. There was no strong effect of cumulative dose among current users of β-blockers in either data set (Fig. 1), while assessing the effect cumulative dose irrespective of timing of use yielded similar results. Table 2.Use of β-blockers and risk of hip/femur fracture in the GPRD and PHARMO RLSβ-blocker exposureGPRDPHARMO RLSCases (%)Controls (%)Crude OR (95% CI)Adjusted OR (95% CI)aCases (%)Controls (%)Crude OR (95% CI)Adjusted OR (95% CI)aTiming of use  Current use4.55.90.70 (0.64–0.77)0.82 (0.74–0.91)12.413.70.91 (0.83–0.98)0.87 (0.80–0.95)  Recent use0.60.60.89 (0.70–1.14)0.88 (0.68–1.15)1.71.71.01 (0.82–1.24)0.93 (0.75–1.15)  Past use0.70.61.10 (0.86–1.40)0.99 (0.76–1.28)1.41.31.11 (0.88–1.40)0.99 (0.78–1.26)  Distant past use3.72.81.10 (0.98–1.22)1.06 (0.94–1.20)6.86.31.07 (0.96–1.20)0.97 (0.87–1.09)Among current users of β-blockersSelectivity  Low1.01.20.75 (0.63–0.90)0.86 (0.71–1.05)3.53.11.12 (0.97–1.30)1.04 (0.89–1.21)  Medium0.10.20.79 (0.48–1.32)0.81 (0.47–1.41)0.30.50.40 (0.20–0.84)0.38 (0.18–0.79)  High3.34.40.69 (0.62–0.76)0.77 (0.69–0.87)8.710.10.86 (0.79–0.95)0.84 (0.76–0.93)Lipophilicity  Hydrophile3.24.20.69 (0.62–0.77)0.78 (0.70–0.88)5.15.20.98 (0.86–1.11)0.94 (0.83–1.06)  Intermediate0.20.20.77 (0.50–1.20)0.88 (0.56–1.39)0.50.41.15 (0.77–1.73)1.04 (0.69–1.58)  Lipophile1.11.40.72 (0.60–0.85)0.81 (0.68–0.98)6.98.00.85 (0.77–0.95)0.83 (0.74–0.92)First prescription  Yes0.20.20.93 (0.58–1.48)1.18 (0.69–1.99)0.40.60.63 (0.42–0.95)0.62 (0.41–0.94)Last prescribed daily dose (DDD)b  <0.670.91.20.72 (0.59–0.87)0.81 (0.65–1.00)8.79.80.89 (0.81–0.98)0.87 (0.79–0.96)  0.67–1.332.02.60.73 (0.64–0.83)0.85 (0.74–0.99)3.23.40.95 (0.82–1.11)0.90 (0.77–1.06)  >1.331.42.00.64 (0.55–0.75)0.81 (0.69–0.97)0.40.40.95 (0.61–1.49)0.85 (0.54–1.35)a Adjusted for use of other antihypertensive drugs and general risk factors for falls and fractures (see Materials and Methods)b One DDD is equivalent to 150 mg metoprololFig. 1.Spline visualization of cumulative dose among current ß-blocker users and risk of hip/femur fractures (GPRD, dashed line, solid circles; PHARMO RLS, solid line, open circles). Cumulative dose is expressed in DDDs (1,000 DDDs are equivalent to 150 g of metoprolol). Odds ratios were adjusted for the same confounders as in Table 2. The most frequently prescribed β-blocker in the GPRD was atenolol (3.0% among cases vs. 4.0% among controls, adjusted OR = 0.80, 95% CI 0.71–0.90). Current use of other β-blockers was infrequent (propranolol 0.4% vs. 0.5%, metoprolol 0.2% vs. 0.2%) and not associated with a decreased risk of hip/femur fracture (adjusted OR = 0.90, 95% CI 0.76–1.06). In the PHARMO RLS, the most frequently used β-blockers at the index date were metoprolol (4.6% among cases vs. 5.6% among controls), atenolol (3.2% vs. 3.6%), sotalol (1.9% vs. 1.5%), and propranolol (0.9% vs. 0.9%). The adjusted ORs for current use of metoprolol and atenolol were 0.79 (95% CI 0.69–0.90) and 0.89 (95% CI 0.77–1.04), respectively. Current use of propranolol (adjusted OR = 0.98, 95 CI 0.74–1.21) and sotalol (adjusted OR = 1.15, 95% CI 0.93–1.42) were not associated with a protective effect of hip/femur fracture. Categorizing β-blockers according to their lipophilicity, receptor selectivity, or last prescribed daily dose did not reveal major differences in effect (Table 2), nor did stratification according to age and gender (Table 3). Table 3.Current use of β-blockers and risk of hip/femur fractures in patient subgroupsCurrent use of β-blockersGPRDPHARMO RLSCases (%)Controls (%)Crude OR (95% CI)Adjusted OR (95% CI)aCases (%)Controls (%)Crude OR (95% CI)Adjusted OR (95% CI)aGender  Men3.44.50.68 (0.55–0.84)0.77 (0.60–0.98)10.112.00.83 (0.70–0.98)0.77 (0.64–0.93)  Women4.86.30.71 (0.64–0.78)0.83 (0.74–0.93)13.314.40.93 (0.95–1.02)0.90 (0.82–1.00)Age (years)  <652.93.10.93 (0.68–1.26)0.91 (0.62–0.91)5.07.21.04 (0.80–1.35)0.94 (0.70–1.27)  65–807.09.10.71 (0.62–0.80)0.84 (0.73–0.98)14.516.80.84 (0.58–1.12)0.80 (0.70–0.91)  80+3.54.60.65 (0.57–0.75)0.77 (0.66–0.91)12.413.50.94 (0.84–1.06)0.94 (0.83–1.07)History of any use of other antihypertensive drugs  No2.22.40.89 (0.77–1.04)0.97 (0.82–1.14)8.38.11.02 (0.85–1.23)1.01 (0.90–1.14)  Yes9.714.70.60 (0.53–0.67)0.73 (0.64–0.83)20.926.90.73 (0.65–0.83)0.76 (0.67–0.86)a Adjusted for use of other antihypertensive drugs and general risk factors for falls and fractures (see Materials and Methods) Percentages represent the proportion of current beta-blocker within each subcategory (e.g. male gender) The protective effect of β-blockers was only present among patients who had been treated with other antihypertensive agents (e.g., low-ceiling diuretics, calcium antagonists, RAAS inhibitors), either concurrently or in the past (Table 3). This finding was consistent in both the GPRD (adjusted OR = 0.73, 95% CI 0.64–0.83) and PHARMO RLS (adjusted OR = 0.76, 95% CI 0.67–0.86). Among patients using only β-blockers, the adjusted ORs were close to unity. In both data sets, the interaction term between current use of β-blockers and use of other antihypertensive drugs was statistically significant (P < 0.05). Furthermore, no effect of cumulative dose was found in either current users of β-blocker only or in current β-blocker users with a history of using other antihypertensive drugs (Fig. 2). Further stratification according to high or low average daily dose during the study period indicated no effect of the intensity of β-blocker use on the risk estimates. Fig. 2.Spline visualization of cumulative dose among current ß -blocker users, stratified according to patients not having (A) or having a history of other antihypertensive drugs (B) (GPRD, dashed line, solid circles; PHARMO RLS, solid line, open circles). Cumulative dose is expressed in DDDs (1,000 DDDs are equivalent to 150 g of metoprolol). Odds ratios were adjusted for the same confounders as in Table 2, except for the use of antihypertensive drugs. Discussion In both the GPRD and PHARMO RLS data sets, current use of β-blockers was associated with a decreased risk of hip/femur fractures. However, there was no reduced risk of hip/femur fractures among patients who did not have a history of using other antihypertensive drugs: a protective effect of β-blockers was only observed for patients with current or prior use of other antihypertensive agents. Even within this group of patients, no dose-response relationship with β-blocker use was found. The effect was constant with cumulative dose and the OR was below 1.0 even among patients who had just started treatment with β-blockers. As the mechanism by which β-blockers could influence BMD is likely to need some time to exert a clinically relevant effect, this finding suggests that the association between β-blockers and fracture risk is not causal. Based on in vivo and in vitro studies and the discovery that the central nervous system is involved in the regulation of bone, β-blockers have been implicated in a preventive role in patients with osteoporosis. Central effects of leptin have been found to be mediated by the sympathetic nervous system, acting via β receptors on osteoblasts [3]. β agonists stimulate bone-resorption activity in osteoclasts [24]. Previous data have shown that systemic application of β agonists had a negative effect on bone mass in mice [25], whereas β antagonists stimulated bone formation in rats [6]. Several epidemiological studies have reported discrepant results on the association between use of β-blockers and risk of fracture [7–12], including a study that also used data from the GPRD [8]. The reason for the discrepancy is unclear. Emerging data from randomized controlled trials also support a lack of effect of β-blockers on the risk of fracture. Recently, a clinical trial among 41 normal postmenopausal women found no evidence that propranolol stimulates bone formation, as measured by bone turnover markers [26]. Furthermore, pooled data from nine clinical trials investigating the nonselective β-blocker carvedilol in the management of congestive heart failure did not provide evidence in support of an effect of β-blockers on fracture risk reduction [10]. The data from this meta-analysis are consistent with the results of our study as no effect on fracture was observed in patients treated with β-blockers without history of other antihypertensive drug use. Therapeutic uses of β-blockers include the treatment of hypertension as well as heart failure and secondary prevention post-myocardial infarction, -cardiac dysrhythmia, and -angina pectoris. Cardiovascular disease, heart failure, and hypertension have all been linked with decreased BMD [27–30]. Because thiazides, calcium channel blockers, and inhibitors of the renin-angiotensin system have also been associated with beneficial effects on bone [31], we stratified in our study on history of use of other antihypertensive drugs to separate the effect of current use of β-blockers from that of other antihypertensive agents. This method of stratification was not applied in earlier epidemiological studies on β-blockers and fractures. The strength of our study is that it was population-based. Furthermore, we found the same results in both the UK and Dutch data sets. The prevalence of β-blocker use was nearly three times as high in the PHARMO RLS compared to the GPRD. In 2002, 5.1 million prescriptions for β-blockers were issued in The Netherlands (population 16 million) [32] compared to 22.4 million prescriptions in the United Kingdom in the same year (population 59 million) [33]. Taking into account that prescriptions in the United Kingdom are usually for 30 days and those in The Netherlands for 90 days, we can conclude that the observed difference in exposure prevalence is in line with prescribing data volumes in both countries. Observational studies like ours have potential for bias and confounding and can fuel debates on study interpretation and credibility [34–36]. Various drugs with effects on the central nervous system are known to increase the risk of falls and thereby fracture risk. Also, there are likely to be complex interactions between vascular disease and fracture risk, operating through falls risk, BMD, or common genetic or lifestyle factors. In this study, we had no information on BMD and we cannot exclude the possibility that cases and controls were different with respect to BMD. However, given that initiation of antihypertensive treatment in daily clinical practice will usually be independent of patient BMD, major confounding seems unlikely. Variables included in the final regression models were slightly different between the two data sets due to the nature of the data collections. However, multivariate adjustment had only modest effects on the OR of the exposure of interest. Information on smoking and BMI was not available in the PHARMO RLS database, but adjustment for these factors in the GPRD had no influence on risk estimates (smoking was not retained in the model). Data on BMI were missing for 58% of the GPRD study population, so we cannot exclude the concern that adjustment was suboptimal. We had no data on physical activity, diet, or socioeconomic status as such information is not available in the databases used in our studies [37]. Furthermore, we cannot exclude the possibility that residual confounding can explain part of our results. An alternative explanation for our findings could be that the protective effect of β-blockers on fractures is an artifact caused by selective underuse in patients with an unmeasured comorbidity, a problem that has been described by Glynn et al. in a study on cardiovascular drug use and mortality [38]. In their study, among elderly subjects in the United States, they found that users of drugs from seven commonly prescribed therapeutic classes, including β-blockers, thiazide diuretics, and angiotensin-converting enzyme inhibitors, had reduced rates of death compared to nonusers, which was more likely to be explained by selective prescribing and nonadherence. A potential limitation is that we confined our study to hip/femur fractures and did not evaluate other type of fractures. It is possible that potential beneficial effects of β-blockers are present only at sites other than the hip/femur, but there is no evidence to support this. In conclusion, the reduction in hip/femur fracture risk was not related to cumulative dose of β-blockers and was only present in patients using β-blockers with a history of using other antihypertensive drugs as well. This suggests that the effect of β-blockers on hip/femur fracture is not causal.

Purinergic_Signal-4-2-2377322

A case of serendipity*

An account is given of how a sensitive bioassay system for measurement of the neurotransmitter acetylcholine serendipitously led to the identification of adenosine triphosphate (ATP) released in vitro from active skeletal muscle. Subsequent application of the identification procedures to exercising human muscle in vivo, cardiac muscle cells in vitro, and human erythrocytes exposed to hypoxia gave rise to the general concept of ATP as a molecule that could influence cell function from the extracellular direction. Mechanisms of ATP release from cells in terms of “trigger” events such as mechanical distortion of the membrane, depolarization of the membrane, and exposure to hypoxia are discussed. Potential therapeutic uses of extracellular ATP in cancer therapy, radiation therapy, and a possible influence upon aging are discussed. Possible roles (distant and local) of extracellular ATP released from muscle during whole body exercise are discussed. Introduction In 1961 Boyd and Eadie [5] developed a highly sensitive bioassay for acetylcholine using an in situ perfusion technique of frog heart. Based on the great sensitivity of the frog heart to acetylcholine (Fig. 1) a project was suggested by B. Katz to I. A. Boyd; they thought it possible that the amounts of acetylcholine released from stimulated motor nerve endings in frog skeletal muscle could be directly measured. The skeletal muscle to be used was the frog sartorius, about which a lot was known regarding the distribution and population of the motor nerve terminals. At that time the association of miniature end-plate potentials (MEPPs) with the release of acetylcholine in packets (“quanta”) together with the anatomical existence of synaptic vesicles at the motor nerve terminals was highly plausible. Any accurate measurement of acetylcholine release from motor nerve stimulation would provide valuable evidence for (or against) the association and the establishment of the “quantal” theory of neuromuscular transmission. In essence the experimental approach mimicked the classic experiment of Otto Loewi, where fluid perfusing a first frog heart was used to perfuse a second heart. When the vagus nerve supplying the first heart was stimulated cardiac arrest occurred, and shortly after this the second heart also stopped beating. Loewi termed the vagal inhibitory transmitter Vagusstoff. An isolated frog sartorius was to be stimulated via the motor nerve in a bathing solution and the solution then perfused through a frog heart. When a stimulated muscle solution was perfused a surprising stimulatory response was elicited, quite unlike the anticipated inhibitory effect of acetylcholine (Fig. 1). Clearly this response would preclude any accurate assay for acetylcholine (as well as any successful conclusion to a thesis project aimed at substantiating the quantal theory of neuromuscular transmission!). Fig. 1The response of a very sensitive frog heart to graded concentrations of acetylcholine (g.ml−1). A concentration of 2.5 × 10−13 g.ml−1 (1.54 × 10−12 mmol.ml−1) produced a threshold response [19] It was evident that for successful measurement of acetylcholine, the identification of the interfering substance, and its removal from the solution, had to be achieved. At this stage the basic question became: what substance(s) released from active skeletal muscle could possibly stimulate the frog heart? A “blunderbuss” approach was adopted, by perfusing through the heart as many available substances, known to exist in muscle, to see if any of them produced the stimulatory effect. The purine monophosphates and adenosine were eliminated since they did not stimulate the heart. Electrolyte changes, including an elevated potassium level, were also ruled out. Changes in calcium levels were hardly detectable in the bathing solution. One obvious candidate as a stimulatory substance was catecholamine material. However, when the frog heart response to adrenaline was blocked by an ergot/pronethalol perfusion, the stimulatory effect remained (Fig. 3). Fig. 2Effect of graded concentrations of ATP (g.ml−1) perfused through a frog heart [6]. Note that low concentrations have a pure inotropic effect, while a triphasic response was produced by higher concentrations. Further analysis of this triphasic response has since been made [17] Indirect versus direct muscle stimulation The distribution of motor end plates in the sartorius muscle (there are no end plate regions at the pelvic end of the frog sartorius) allowed muscle fibers to be stimulated directly, without involvement of the neuromuscular junction in the activation process. Solutions bathing muscles stimulated directly produced the same stimulatory effect as the solutions from muscles stimulated via the motor nerve (indirect stimulation). This indicated that it was highly unlikely that the stimulatory substance had as its major source the activated motor nerve terminal. Identification procedures Identification of the stimulatory substance was established using three procedures: gel filtration, enzyme action of apyrase, and firefly tail luminescence. Gel filtration chromatography The use of a gel filtration chromatography technique (Sephadex G-25, fine grade) enabled removal of protein from the stimulated muscle solution, allowing spectral absorbance analysis. An absorbance peak obtained at 265λ was the first hint that a purine compound was present in the solution. The gel filtration technique also established a molecular weight of the stimulatory substance close to that of ATP.In the face of prevailing dogma that ATP could not cross cell membranes, perfusion with ATP was reluctantly performed. A stimulatory effect was obtained—the author had become a citizen of the Republic of Serendip! A dose response of the frog heart to ATP is shown in Fig. 2. Fig. 3Comparison of the stimulatory action of a stimulated muscle solution (‘X’) on a frog heart before and after adrenergic blockade. Vertical interrupted line, 2-h period during which heart was perfused with pronethalol hydrochloride (10−6 g.ml−1) and ergotamine tartrate (10−6 g.ml−1). The stimulatory action of the stimulated muscle solution was unaffected by adrenergic blockade. Reproduced from Boyd and Forrester [6], with permission of the Physiological Society Incubation with apyrase In 1962 Traverso-Cori and Cori [61] demonstrated that an enzyme purified from potatoes (“apyrase”) could selectively cleave the terminal phosphate of ATP. The stimulated muscle solution was incubated with apyrase, as were the stimulatory fractions from the gel filtration technique, and it was shown that the stimulatory effect was abolished, indicating that a phosphorylated compound was responsible for the stimulatory effect. The luciferin/luciferase test for ATP As long ago as 1947 McElroy [38] showed that when a crude extract of lampyrid beetles (Photinus pyralis) was exposed to ATP a light signal was generated. When the stimulated muscle solution was applied to luciferin extract a light signal was produced (see [18]). Procedures applied to perfused frog musculature As this work developed the question “does ATP appear extracellularly as a result of tissue damage?” was frequently asked. Dissection damage to the sartorius muscle was avoided by the use of the perfused hindlimb preparation of the frog, where instrument trauma was limited to the insertion of cannulae. The identification procedures for ATP were applied to the perfusate from stimulated (indirect) hindlimb musculature. Table 1 shows the relationship of ATP output to the frequency of motor nerve impulses [23]. When these amounts (0.24–1.8 nmol 100 g−1 min−1) are compared to the amounts of ATP released from isolated sartorius (50 nmol 100 g−1 min−1, [6]), it is seen that there is a great reduction detected in the perfusate sample, presumably as a result of exposure to greater ecto-ATPase activity that was not encountered in solutions bathing isolated muscles. Table 1Relationship of ATP output to frequency of stimulation of perfused frog limbaHzpmol volley−1 100 g−1pmol 100 g−1 min−114.024021.518051.4420103.01,800aData from Forrester and Hassan [23] Procedures applied to human muscle in vivo The appearance of ATP in the perfusate from active frog hindlimb musculature raised the possibility that ATP might be released from active muscle into the circulating blood in vivo. The identification procedures were applied to human plasma obtained from resting and exercising human forearm muscle [20, 21]. Figure 4 gives an example of the combined use of gel filtration chromatography, firefly luciferase, and apyrase for the identification and quantification of ATP in human plasma from exercising human forearm muscle [22]. Estimates of plasma ATP levels can only be approximate, since the accumulation in the blood is continuously offset by ATPase activity of both plasma and the surfaces of the erythrocyte and endothelial cells. Fig. 4Elution of ATP and plasma samples from exercising human forearm through a Sephadex column. Elution fractions were estimated with firefly extract. a ATP, 10−6 g.ml−1. b Post-occlusion human sample from a subject during forearm exercise. c Samples from another subject exercising forearm muscle, fraction at 9 ml was incubated with apyrase and retested on firefly extract. Fraction at 10 ml, incubation control. Reproduced from Forrester [22], with permission of the Physiological Society Sequelae The present-day knowledge and extensive classification of purine receptors (see review by [48]) has vindicated the concept of an extracellular system of ATP action influencing many tissues of the body. However, the mode of ATP release from cells remains a formidable challenge (see review by [56]). In the cases of skeletal muscle, cardiac muscle, and erythrocyte some specific circumstances leading to ATP release, e.g. membrane depolarization, hypoxia, and mechanical distortion [25], should be addressed further. Association of ATP release from depolarizing membrane In 1962 Abood et al. [1] assessed outflux of various phosphates during membrane depolarization of excitable tissues. Prior to that Holton [32] detected ATP release from sensory nerves stimulated antidromically. An elegant experiment by Israel et al. [33] showed discrete release of ATP closely related to membrane depolarization in the electric organ of Torpedo. In 1973 Ohta et al. [40] studied and compared the mode of action of several veratrum alkaloid compounds on axonal membranes (squid giant axon). Some of them exerted a highly specific action on the resting sodium permeability, with veratridine having the most potent effect. Two results led them to conclude that the depolarization evoked by veratridine was caused by specifically increasing the membrane permeability to sodium ions: (1) application of tetrodotoxin (TTX), which specifically blocks sodium channels, restored the resting membrane potential after depolarization by veratridine and (2) removal of sodium ions from both external and internal phases of the axon, thus eliminating the sodium term from the Goldman-Hodgkin-Katz constant field equation describing the membrane potential [31], had the result that veratridine had no depolarizing action after sodium removal. In 1977 White [63] demonstrated ATP release from depolarizing isolated synaptosomes prepared from rat brain. The synaptosomes were suspended in a solution containing firefly extract. Depolarization was achieved by (1) elevation of extracellular potassium and (2) with application of veratridine. A light signal was evoked in both circumstances. Prior exposure of the synaptosomes to TTX, thus blocking the sodium channels, also blocked the light signal evoked by veratridine. This suggested that the release of ATP in response to veratridine in the absence of TTX was due to membrane depolarization following the opening of sodium channels. A record of one such experiment by Tom White is shown in a previous publication ([25] Fig. 16b). Hypoxia The potency of hypoxia in evoking ATP release from heart cells and erythrocytes has been demonstrated [4, 24]. In the case of human erythrocyte, ATP release in response to hypoxia was very effectively blocked by the nucleoside transport blocker nitrobenzylthioinosine (NBTI). The nucleoside transporter has now been cloned [26] and is thought to be a member of a group of unclassified integral membrane proteins (see review by [8]). The connection between blockade of a nucleoside transporter and inhibition of nucleotide release (ATP) in response to hypoxia is unclear.Some headway has been made toward unraveling oxygen-sensing mechanisms in certain oxygen-sensitive cells. In 1988 Lopez-Barneo et al. [36], studying chemoreceptive carotid body cells, demonstrated that lowered oxygen tension inhibited potassium channels, resulting in cell depolarization. The presence of oxygen-sensing potassium channels has been reported in many different cell types, for example, in pulmonary vascular myocytes, which constrict due to membrane depolarization following hypoxia [44] and in rat pheochromocytoma cells [12]. Does ATP release associated with membrane depolarization, as reported above [63], occur in these cells? Perhaps release of ATP from heart cells and erythrocytes in response to hypoxia involves the presence of oxygen-sensitive potassium channels, the sequence of events being hypoxia, leading to inhibition of oxygen-sensitive channels, followed by membrane depolarization and ATP release. To date, these potassium channels have not been identified in heart cells or erythrocytes. Mechanical distortion A broad spectrum of mechanical distortion exists, ranging from gross distortion of whole cells to the impact of suction electrodes in the patch-clamp techniques for recording of currents passing through single ion channels in the membrane. There is little doubt that mechanical distortion of the whole cell membrane can lead to the release of ATP into the immediate environment. Sprague et al. [59] showed a close relationship between the extent of erythrocyte deformation and ATP release. Increases in shear stress applied to endothelial cells have been shown to liberate “vasoactive” substances from vascular endothelial cells [10, 45, 47, 54]. ATP release into the pulmonary vascular bed occurs within seconds after an increase in flow rate [30]. Grygoczyk and Hanrahan [27] clearly showed that minimum mechanical disturbance, such as changing the solution bathing the preparation, enhanced ATP release within seconds. Schwiebert [55] emphasized that the degree of suction pressure in patch-clamp techniques may determine the rate of ATP appearance. A further factor complicating physical distortion of the cell membrane is the possible involvement of the underlying cytoskeletal system. Frequency of membrane channel opening is influenced by exposure to cytochalasin B, which disrupts the cytoskeleton [62]. A comprehensive review by Morris [39] outlining the widespread occurrence of mechanosensitive ion channels in a variety of tissues is highly recommended. Perspectives Towards the search for an ATP channel in the plasma membrane Braunstein et al. [7], investigating the mechanism of cell volume regulation, showed that the cystic fibrosis transmembrane conductance regulator (CFTR) restored cell volume (following hypotonic challenge) by a mechanism involving ATP release. This release could be blocked by DIDS and gadolinium, suggesting that ATP might be released through a channel separate from the CFTR. Their working hypothesis was that cell volume is sensed and transduced by an ABC transporter (ATP binding cassette) enhancing ATP release through activation of a separate but associated ATP channel. The earlier observation [4] that hypoxic-induced ATP release from erythrocyte was associated with “band 3” membrane prompted the speculation that band 3 could be equated with the CFTR protein moiety. However, these proteins have been cloned and sequenced and are quite dissimilar in character [34, 52].ATP passage across the inner mitochondrial membrane has long been known. Thinnes et al. [60] have identified a voltage–dependent anion channel (VDAC, “porin”) which acts as a conduit for newly synthesized ATP in and out of the mitochondrion. VDAC is present in a multitude of tissue types, including human skeletal muscle [51] (Table 1). Exercise and circulating purines An early example of the appearance of purines in the circulating blood in response to graduated whole body exercise was obtained by Peter Parkinson [43]. It is known that a large proportion of circulating ATP is degraded in passage through the lung [15], perhaps, along with widespread ecto-ATPase activity, minimizing the effect of widespread vasodilatation and offsetting unwanted hypotension and collapse during exercise! The products of purine degradation are probably distributed generally, especially to tissues that cannot synthesize the purine ring moiety (see “Discussion,” [11]). This phenomenon could appropriately be included in the concept of “ATP expansion” proposed by Abraham et al. [2]. This begs the question: is one of the beneficial effects of an exercise regime due to the intermittent “internal transfusion” of purines, including ATP, from exercising muscle to sustain “ATP expansion” in the tissues? Use of ATP in cancer therapy Encouraging results have been obtained in the field of cancer therapy using extracellular ATP. A direct effect of ATP on tumor cells was demonstrated by Rapaport and coworkers [49, 50]. It was shown in various human cell lines that an increase in cellular ATP pools resulted in inhibition of DNA replication in the synthesis phase of the cell cycle, with subsequent arrest of cell growth. Inhibition of cancer growth in Ehrlich tumor cells by extracellular ATP has also been found [14, 35]. The dramatic weight loss seen in some forms of advanced cancer (cachexia) can be slowed with a regime of intravenous ATP [3]. As shown in Fig. 5 maintenance of muscle strength was also a benefit. Fig. 5Changes in body weight and muscle strength in advanced non-small-cell lung cancer after ATP administration. a Weight change. b Elbow flexor performance (top); knee extensor performance (bottom). Reproduced from Agteresch et al. [3], with permission Protective effect of ATP against ischemia Improved recovery of ischemic liver, kidney, and myocardial tissues after perfusion of ATP-MgCl has been shown [9, 37, 41, 58]. Protective effect of ATP in radiation therapy Impairment of wound healing following radiation therapy has been shown to be ameliorated with use of ATP-MgCl perfusion regimes [57].Although little understood, these protective effects may be associated with general elevation of tissue ATP levels, restoring the total “energy pool” of the body (see “ATP expansion” in review by [2]). Aging and extracellular ATP levels The release of ATP from vascular endothelial cells in rat caudal artery is reduced with advancing age. Both spontaneous ATP release and release induced by methoxamine, an α-adrenoceptor agonist, were reduced in aged vascular endothelium [28]. It is interesting that enhanced release of adenyl purines from these cells was invoked by hypotension induced by exercise [29]. A clear example of the effect of age on the ATP content of human erythrocytes is shown in Table 1 [46].Aging has also been shown to diminish phosphorus metabolites in left ventricular hypertrophy as measured by 31P magnetic resonance spectroscopy [42].If aging is associated with lower tissue levels of ATP, and applied exogenous ATP can replenish these levels (“ATP expansion” see above), could aging reversal be brought about by increasing the exogenous ATP levels? At this juncture it should be noted that extracellular ATP can act as a powerful trigger of programmed cell death (apoptosis) in endothelial cells [13, 53]. Perhaps the phenomenon of apoptosis is designed to prevent the Malthusian nightmare of universal immortality! Epilogos And so what began as a project for the assessment of quantal release of acetylcholine from motor nerve terminals1 resulted in the finding that ATP was released from active skeletal muscle, leading to the establishment of a significant role for ATP in the hyperemia of exercising skeletal and heart muscle, as well as an appreciation of the widespread effects of extracellular ATP. Intrinsic to the concept of serendipity in science is the element of surprise, which, although conferring excitement to investigators, only reveals our frailty of prediction in scientific matters. Inspirata accident magis saepe quam quae speres—What you do not expect happens more frequently than what you do expect (Plautus).

Biogerontology-4-1-2174522

Energy metabolism, altered proteins, sirtuins and ageing: converging mechanisms?

The predominant molecular symptom of ageing is the accumulation of altered gene products. Nutritional studies show that ageing in animals can be significantly influenced by dietary restriction. Genetics has revealed that ageing may be controlled by changes in intracellular NAD/NADH ratio regulating sirtuin activity. Physiological and other approaches indicate that mitochondria may also regulate ageing. A mechanism is proposed which links diet, exercise and mitochondria-dependent changes in NAD/NADH ratio to intracellular generation of altered proteins. It is suggested that ad libitum feeding conditions decrease NAD availability which also decreases metabolism of the triose phosphate glycolytic intermediates, glyceraldehyde-3-phosphate and dihydroxyacetone-phosphate, which can spontaneously decompose into methylglyoxal (MG). MG is a highly toxic glycating agent and a major source of protein advanced-glycosylation end-products (AGEs). MG and AGEs can induce mitochondrial dysfunction and formation of reactive oxygen species (ROS), as well as affect gene expression and intracellular signalling. In dietary restriction–induced fasting, NADH would be oxidised and NAD regenerated via mitochondrial action. This would not only activate sirtuins and extend lifespan but also suppress MG formation. This proposal can also explain the apparent paradox whereby increased aerobic activity suppresses formation of glycoxidized proteins and extends lifespan. Variation in mitochondrial DNA composition and consequent mutation rate, arising from dietary-controlled differences in DNA precursor ratios, could also contribute to tissue differences in age-related mitochondrial dysfunction. NAD and life-span Genetic studies using a range of organisms have indicated that enzymes called sirtuins are linked to the control of ageing and life-span (Longo and Kennedy 2006; Leibiger and Berggren 2006; Lin and Guarente 2003). Sirtuins catalyse NAD-dependent deacetylation of histones (and other proteins) with the concomitant release of nicotimanide and O-acetyl-ADP-ribose (Howitz et al. 2003). Other studies have suggested that metabolism of the redox couple NAD/NADH provides a link between sirtuin activity and the control of cell senescence and organism life-span (Denu 2003, 2007; Belenky et al. 2007; Bordone and Guarente 2005): NAD-dependent protein deacetylation helps maintain the juvenile phenotype, whereas inhibition of deacetylation activity by NADH or nicotinamide, or by NAD unavailability, promote the onset of cellular aging and decrease organism lifespan. Ageing, dietary restriction and NAD Ageing can be delayed in various organisms by dietary restriction (DR) induced by a permanent decrease in calorie intake (called caloric restriction—CR). Recent observations have shown that an intermittent feeding (IF) protocol, which need not involve any overall reduction in calorie intake, can also delay ageing (Martin et al. 2006; Masternak et al. 2005; Mattson and Wan 2005). The mechanisms by which DR delays ageing and increases life-span are far from completely understood (Sinclair 2005). It is likely, however, that both CR and IF promote similar effects on the frequency of glycolysis and subsequent fasting periods (Hipkiss 2006a and 2007), i.e. glycolysis would be discontinuous, only operating post-prandialy. In contrast, glycolysis would be almost continuous under ad libitum (AL) feeding conditions. It is suggested that during the periods of fasting (induced by either CR or IF) the NAD/NADH ratio would differ from that prevailing in the AL case where fasting would be unlikely or negligible. In the AL condition, continuous glycolytic throughput would tend to provoke an accumulation of NADH and lower NAD availability, whereas the CR- and IF-induced fasting would decrease glycolytic NAD demand and increase NADH oxidation and NAD regeneration. Ageing and accumulation of altered proteins At the biochemical level ageing is characterized by the accumulation of altered protein molecules. The changes in protein structure result from intrinsic polypeptide instability as well as the actions of deleterious endogenous and exogenous agents (see Hipkiss 2006b; Schoneich 2006 and refs. therein). As yet it is unclear how changes in NAD metabolism might induce generation of altered proteins which characterise the aged phenotype. Protein glycation and ageing Formation of protein advanced glycation end-products (AGEs) is an important consequence of ageing and is increased particularly under conditions of uncontrolled glucose metabolism (e.g. hyperglycaemia) (see Ahmed and Thornally 2007; Thornalley 2007 and refs. therein for recent reviews). Protein AGEs can themselves induce inflammatory conditions and provoke production of reactive oxygen species (ROS) which can further compromise cell function. Indeed recent studies have shown that decreasing dietary AGE intake preserves defence functions against oxidative stress and decreases tissue damage in humans, and extends lifespan in mice, while increasing dietary AGE intake is correspondingly deleterious and accelerates ageing and decreases life-span (Cai et al. 2007; Uribarri et al. 2007a, b). Hence it is at least conceivable that decreasing metabolically-generated protein AGEs could help decrease the overall AGE load and could have beneficial effects by suppressing ageing and extending lifespan. NAD and accumulation of methylglyoxal, an endogenous glycating agent NAD is essential for the metabolism of the glycolytic intermediate glyceraldehyde-3-phosphate (G3P) via the action of glyceraldehyde-3-phosphate dehydrogenase (GAPDH), generating 1,3-diphosphoglycerate (1,3DPG) and NADH as products. It is argued above that in the AL condition, glycolysis would be continuous, which would tend to lower NAD levels and raise NADH levels. This would occur especially should mitochondrial-mediated NADH re-oxidation to NAD be correspondingly lowered to compensate for the extra ATP synthesised via glycolysis, assuming cellular ATP demand remains unchanged. Limitation of NAD availability would lower GAPDH activity and promote an accumulation of G3P. The immediate precursor of G3P is dihydroxyacetone phosphate (DHAP); both of these trioses can glycate proteins. More importantly, however, is the fact that both G3P and DHAP can spontaneously decompose into methylglyoxal (MG), a highly toxic and very reactive glycating agent. It is likely therefore that changes in NAD availability could strongly influence MG production. It has previously been proposed that differences in glycolytic frequency could help explain why dietary restriction delays cellular and organism aging, possibly due to decreased MG generation during DR conditions (Hipkiss 2006a). MG is a highly active glycating agent which is thought to be responsible for the increased protein/lipid glycation detected during hyperglycaemic conditions and for much of the protein/lipid glycation associated with diabetic complications (Ahmed and Thornalley 2007; Thornalley 2007). Although MG is a normal cellular constituent, its excessive production is deleterious (plasma MG levels are raised to around 800 nmol/l in young diabetics compared to about 400 nmol/l in young non-diabetic subjects, (Han et al. 2007)). Importantly, MG can induce many of the deleterious physiological and biochemical changes characteristic of the aged phenotype, including increased ROS generation, mitochondrial dysfunction, apoptosis and inhibition of cell division (see Hipkiss 2006a and refs therein). A number of studies, some very recent, reinforce the notion that changes in cellular MG content are important determinants of the formation of altered protein that characterise senescence (Gomes et al. 2006). Even at non-toxic concentrations, MG can influence cell proliferation by forming adducts with growth factor receptors (Cantero et al. 2007). MG can also inhibit the activity of GAPDH (Lee et al. 2005), causing triose phosphate accumulation and thereby increasing MG generation, and so inducing a highly deleterious cycle. MG can induce apoptosis (Nicolay et al. 2006) and also affect gene expression and signal transduction, at least in cultured cells (Du et al. 2003; Yao et al. 2006; Ramasamy et al. 2006). Two studies in Drosophila have shown that mutation in triosephosphate isomerase (the enzyme which converts DHAP into G3P, preceding GAPDH in the glycolytic pathway, and which is known to undergo age-related post-synthetic modification in vivo (Gracy et al. 1990)) is highly deleterious, causing paralysis, neurodegeneration and decreases life-span (Gnerer et al. 2006; Celotto et al. 2006), possibly because of MG accumulation. Human studies have shown that a deficiency in triosephosphate isomerase activity, causes increases in the levels of both DHAP (up to 20-fold (Schneider 2000)) and MG (Ahmed et al. 2003), and induces neuromuscular degeneration and early death (Schneider et al. 1965; Valentine 1966). Other studies have shown that MG induces apoptosis in neutrophils (Gawlowski et al. 2007), inhibits extracellular matrix remodelling (Chong et al. 2007) and can interfere with the stress response (Oya-Ito et al. 2006) by suppressing NF-kappaB-responsive gene activation (Laga et al. 2007). Hence it is reasonable to suggest that any increased MG generated in AL fed animals, compared to animals subjected to CR or IF, could make a significant contribution to cellular dysfunction. During the fasting periods in DR animals, NADH generated during glycolysis would be oxidized mitochondrially for ATP production, and NAD would be regenerated thereby allowing continued G3P metabolism, and preventing triose phosphate accumulation and consequently suppressing MG generation. This condition would decrease MG-induced macromolecular glycoxidation, mitochondrial damage, dysfunctional signalling and gene expression, as described above. Such a scenario is consistent with the findings that raising NAD levels, or lowering NADH levels by increasing its oxidation, also promote sirtuin activation, with concomitant beneficial effects on cell survival etc. Table 1 illustrates the interrelationship and overlap between sirtuin regulation, generation of altered proteins and mitochondrial activity, exerted by metabolic effects on NAD and NADH levels. Table 1Predicted effects of aerobic exercise and fasting induced by caloric restriction or intermittent feeding, on NAD and NADH levels, methylglyoxal (MG) levels, mitochondrial (mito) activity and sirtuin activityConditionsNADNADH MGMito activitySirtuin activityAd libitum fedLowHighHigh LoweredLoweredFastingHighLowLowIncreasedIncreasedAerobic exercise HighLowLowIncreasedIncreasedIncreased MG levels are partly responsible for the increased generation of altered proteins that accompanies ageing Any situation such as fasting which maintains NAD levels, either via regeneration from NADH, or by synthesis de novo or via a scavenging pathway, would facilitate metabolism of the MG precursors G3P and DHAP, and so decrease the incidence of MG-induced macromolecular damage. The increase in free-radical-mediated damage which occurs during AL feeding, compared to the CR and IF conditions, might occur as a result of not only MG-induced generation of ROS following its reaction with proteins etc., but also via plasma membrane-mediated NAD(P)H-oxidase activity. Furthermore, because less ATP is required from mitochondrial function due to continuous ATP synthesis via glycolysis in the AL-fed state, the decreased supply of electrons (as acetyl-CoA or from NADH) to the electron transport chain would tend to produce more incompletely reduced oxygen moieties i.e. oxygen free-radicals. Any increased intra-mitochondrial ROS production could also increase the probability of mitochondrial dysfunction. Protection against MG is afforded by the glyoxalase system which consists of two enzymes; glyoxalase I (GLX I), which uses glutathione to convert MG to a D-lactoyl-glutathione, and glyoxalase II (GLX II), which completes the detoxification by generating D-lactate and reduced glutathione. Over-expression of GLX I can inhibit formation of hyperglycaemia-induced AGEs (Shinohara et al. 1998), while a deficiency in GLX I in humans is associated with increased protein glycation (Miyata et al. 2001). GLX II activity may be rate-limiting in MG detoxification; GLX II over-expression is protective against MG-induced cell death, whilst its deficiency promotes MG-induced cell death (Xu and Chen 2006). It is also interesting that tumour necrosis factor can induce phosphorylation of GLX I which also results in substantial increase in cellular MG (van Herreweghe et al. 2002). Tissue differences in ageing susceptibility Tissues appear to age at different rates as shown by the varied incidence of dysfunctional mitochondria between tissues in the same organism. Variation in tissue susceptibility to MG may reside partly in differing levels of the glyoxalase system together with those molecules (glutathione, polyamines, carnosine, creatine, pyridoxamine) which normally exert protective carbonyl scavenging activity towards glycating agents such as MG (Hipkiss 2005; de Arriba et al. 2006). Dietary-induced effects on metabolism could conceivably also affect mitochondrial DNA composition and hence mitochondrial protein structure and function. It has been found that mitochondrial DNA mutation rate may vary up to three-fold according to the relative concentrations of the four deoxyribonucleoside triphosphates in the nucleotide pool (Song et al. 2005; Mathews and Song 2007). It is possible that dietary changes could affect the composition of the nucleotide pool and thereby affect mitochondrial DNA composition during its synthesis. Pool composition could vary between tissues, and any consequent differences in mitochondrial DNA mutation rate would contribute to tissue-specific age-related mitochondrial change. Thus mitochondrial dysfunction could be either a cause or a consequence of ageing (Hipkiss 2003), depending on the prevailing circumstances. The beneficial effects of functional mitochondria on NAD regeneration The recent observations (i) by Belenky et al. (2007) showing that life-span extension in yeast is dependent upon NAD synthesis, (ii) that efficient mitochondrial function was necessary for maximal longevity in yeast (Piper et al. 2006), and (iii) that mitochondrial uncoupling, which increases NADH oxidation, decreases telomere damage and delays senescence in cultured human fibroblasts (Passos et al. 2007), are observations entirely consistent with the above proposal. The proposed beneficial effects of NADH oxidation to regenerate NAD via mitochondrial function would also help explain how aerobic exercise may delay development of the aged phenotype including production of altered proteins, as well as resolve the apparent paradox that increased oxygen utilization suppresses age-related change. The efficient regeneration of NAD via effective mitochondrial function is also consistent with mitochondrial ageing theories which postulate that mitochondrial dysfunction is key to the onset of ageing. Also consistent with the present proposal are the very recent findings of Smith et al. (2007) who concluded that, in the yeast Saccharomyces cerevisiae at least, elevated respiration is an important determinant of chronological longevity. They observed that growth on non-fermentable carbon sources, which forced the cells to employ respiration exclusively, extended lifespan, but which caloric restriction did not further enhance. This again illustrates, simplistically perhaps, the potential anti-ageing functions of aerobic respiration and the deleterious effects of glycolysis, both possibly mediated via changes NAD and NADH levels, which in turn regulate MG generation. Controversially, however, these authors also found that caloric restriction-mediated lifespan extension occurred independently of sirtuin activity in Saccharomyces cerevisiae. Other functions induced by DR Ageing is a complex phenomenon. It is likely that the rate-limiting event which increases cellular and hence organism vulnerability to death may vary according to circumstances. For example anti-oxidant functions may not be limiting in conditions where oxidative stress is not involved. There are an increasing number of findings suggesting that proteolytic dysfunction involving either proteasomes or autophagy cause altered protein to accumulate and compromise cell survival and which can be affected by dietary restriction (Bergamini et al. 2003). Conversely activation of autophagy by inhibiting the target of rapamycin (TOR) signalling pathway can increase lifespan, at least in yeast (Bonawitz et al. 2007) and a nematode worm (Henderson et al. 2006). The recent observation that the sirtuin-like activity, histone deacetylase 6 (HDAC6), may provide a mechanistic link between the autophagic and ubiquitin-proteasome proteolytic systems in Drosophila (Pandey et al. 2007), and the observation that up-regulation of neuronal sirtuin1 activity elevates the activity of the α-protease and prevents accumulation of the amyloid peptide (Qin et al. 2006), support the idea that both formation and degradation of aberrant proteins are important for control of ageing and related disorders. Conclusion It is proposed that dietary-induced changes in NAD and NADH levels, as revealed by their regulation of sirtuin activity, may also control the concentration of deleterious glycolytic intermediates G3P and DHAP, and thereby also control formation of MG and generation of protein AGEs. The accumulation of MG and protein AGEs may compromise tissue function including mitochondrial activity and thereby contribute to organism ageing. Conversely, conditions that stimulate mitochondrial function will help regenerate NAD, maintain sirtuin activity and decrease formation of protein AGEs, intra- and extra-mitochondrial ROS can thereby delay ageing onset.

Behav_Genet-4-1-2257998

Power Calculations Using Exact Data Simulation: A Useful Tool for Genetic Study Designs

Statistical power calculations constitute an essential first step in the planning of scientific studies. If sufficient summary statistics are available, power calculations are in principle straightforward and computationally light. In designs, which comprise distinct groups (e.g., MZ & DZ twins), sufficient statistics can be calculated within each group, and analyzed in a multi-group model. However, when the number of possible groups is prohibitively large (say, in the hundreds), power calculations on the basis of the summary statistics become impractical. In that case, researchers may resort to Monte Carlo based power studies, which involve the simulation of hundreds or thousands of replicate samples for each specified set of population parameters. Here we present exact data simulation as a third method of power calculation. Exact data simulation involves a transformation of raw data so that the data fit the hypothesized model exactly. As in power calculation with summary statistics, exact data simulation is computationally light, while the number of groups in the analysis has little bearing on the practicality of the method. The method is applied to three genetic designs for illustrative purposes. Introduction The importance of statistical power in (behavior) genetic analyses is evident in the number of articles devoted to power calculations. Power has been studied in virtually all research designs, ranging from the classical twin design (Martin et al. 1978; Neale et al. 1994), to extended family designs (e.g., Heath et al. 1985; Heath and Eaves 1985; Posthuma and Boomsma 2000), to sibpair and family linkage and association designs, either in- or excluding gene by environment interaction (Abecasis et al. 2000a, b; Boomsma and Dolan 1998; Dolan et al. 1999; Fulker and Cherny 1996; Purcell 2002; Purcell and Sham 2002; Sham et al. 2000; Sham and Hewitt 1999; Sham et al. 2002; Van den Oord 1999). For a wide range of genetic designs, the Genetic Power Calculator1 (Purcell et al. 2003) can be used to calculate power. However, for customized designs and specific research questions, researchers may have to resort to their own procedure to carry out power calculations. Power calculation based on the likelihood with the general Pearson–Nyman statistical decision theory takes two forms. First, the non-centrality parameter λ of the non-null χ2-distribution can be calculated in the analysis of exact sufficient statistics (e.g., Dolan et al. 1999). If the distribution of the data is multivariate normal, the expected variance covariance matrix Σ and the means vector μ are sufficient statistics, as they define the likelihood of the data up to an arbitrary constant (Azzelini 1996). Second, when Σ and μ are not sufficient statistics, the non-centrality parameter λ of the non-null χ2-distribution can be estimated on the basis of the analysis of simulated data using Monte Carlo simulation methods (e.g., Fulker and Cherny 1996; Abecasis et al. 2002a, b; Purcell 2002; van den Oord 1999). The latter is computationally intensive, but does not require the presence of sufficient summary statistics, whereas the former is computationally light, but does require sufficient summary statistics. The aim of the present note is to discuss a third method of power calculation, which we refer to as exact data simulation. This method is suitable when data are multivariate normal, and sufficient summary statistics are in principle available, but the number of possible groups is prohibitively large (say, in the hundreds). The large number of distinct groups renders power calculations on the basis of the summary statistics impractical. Usually, researchers resort to Monte Carlo based power studies under such circumstances. However, exact data simulation, in combination with the definition variable facilities in packages like Mplus (fourth edition, Muthén and Muthén 1998–1997) or the freely available Mx program2 (Neale et al. 2003), is also applicable, and is more efficient than raw data simulation. Exact data simulation was used by Dolan et al. (2005) to evaluate the effects of missing data on the power in structural equation modeling, and by Van der Sluis et al. (Under revision) to evaluate the power to detect gene by environment interaction in sib-pair association studies. Although the technique of exact simulation is in itself not new (Bollen and Stine provided the basics in 1993, and the technique of exact data simulation has recently been integrated as a distinct function in the freely available R-program3), we wish to bring it to the attention of geneticists since this method of calculating power has general value in the field of genetic modeling. Below, we shortly recapitulate the basics of power analysis, and then outline the procedure of exact simulation, which may be implemented readily (we use the freely available R program). The method is illustrated in three genetic designs. Although we confine ourselves in this paper to illustrations in the context of genetic designs, we stress that this form of simulation can be used for power calculations in a wide range of other designs such as random effects models, growth curve and simplex models, and structural models. Power calculation The concept of power is closely related to the two types of statistical errors: the Type I error (i.e., the probability of rejecting a true hypothesis, α), and the Type II error (i.e., the probability of accepting a false hypothesis, β). Power is defined as 1 − β, i.e., the probability of rejecting a false hypothesis, or the probability of not making a Type II error. The basic aim of a power study is to determine the sample size N, which is required to achieve adequate power, given chosen α and a particular effect size. For example, suppose that we want to fit a classical univariate ACE twin-model (see Fig. 1), and we expect that additive genetic effects (a) account for 40% of the phenotypic variance, besides effects of shared (c) and unique (e) environment. We denote this model 1, or hypothesis 1, H1. Under model 1, four parameters are estimated: the path coefficient for the additive genetic effects A (a), the path coefficient for the shared environment C (c), the path coefficient for the unique environment E (e, which includes measurement error), and the means of the twins (μ), which are usually set to be equal within twin pairs and across MZ and DZ twins. The parameter vector θ for this model with df1 degrees of freedom, is θ1 = {a, c, e, μ}, from which the covariance matrices Σ1mz and Σ1dz and mean vector μ1 can be derived. Now consider a second model, which we denote model 0, or H0, in which the additive genetic effects are assumed to account for only 5% of the phenotypic variance. This alternative model will be characterized by parameter vector θ0 = {c, e, μ}. Note that in model 0, parameter a is not estimated (and thus not part of θ) but fixed at a value that corresponds exactly to 5% of the phenotypic variation being explained by additive genetic effects. This model has df0 degrees of freedom, covariance matrices Σ0mz & Σ0dz, and mean vector μ0. This alternative model H0 is nested in the null-model H1, because the parameters in θ0 represent a subset of the parameters in θ1 (e.g., Bollen 1989; Satorra and Saris 1985). The study of power is subsequently concerned with the probability of rejecting the false model H0 in favor of the true model H1, given α, the true value of parameter a (the effect size), and sample size N. Fig. 1Classical univariate ACE-twin model To calculate power, we adopt the method of Satorra and Saris (1985, see also Saris and Satorra 1993), which is based on the normal theory log-likelihood ratio test statistic T. In a single group, T is calculated as follows: where N is the sample size, p denotes the number of variables in the analysis, and Σ and μ, and S and m represent the theoretical and observed variance covariance matrix and the means vector, respectively. Given that the assumptions of normal theory maximum likelihood are met (e.g., multivariate normality, and a large sample of independently and identically distributed cases), and under the assumption that Σ and μ represent the true model (Σ1 and μ1), the test statistic T follows a χ2 distribution with df1 degrees of freedom, i.e., T ∼ χ2(df1) (Azzeline 1996; Bollen 1989). If Σ and μ do not represent the true model but the alternative model (Σ0 and μ0), and given regularity conditions are satisfied (practically amounting to multivariate normality, limited misfit and large sample size N), the test statistic T follows a non-central χ2 distribution with df0 degrees of freedom and non-centrality parameter λ, i.e., T ∼ χ2(df0, λ), where λ > 0. Given the significance level of the test α, and the difference in degrees of freedom between the true model and the alternative model, df1 − df0, the criterion level cα can be obtained from a χ2 table. If the test statistic T exceeds this criterion level, i.e., T > cα, then the alternative model is rejected in favor of the true model (i.e., the fit of the alternative model to the data is significantly worse than the fit of the true model). The aim of power studies is to determine the probability of observing T > cα, i.e., P(χ2(df0, λ) > cα), given Σ0, μ0, Σ1, μ1, N, and α, i.e., the probability of rejecting the alternative model in favor of the true model. The non-centrality parameter λ can be obtained by fitting the alternative model to the true Σ1 and μ1, whereby λ equals the difference in the χ2 fit statistic of the model H1 and the χ2 fit statistic of the alternative model H0. That is (again in a single group), and the non-null distribution of this test statistic is χ2(df1 − df0,λ). A variety of programs can subsequently be used to integrate the non-null distribution to obtain the power (e.g., R, Mx; see also Hewitt and Heath 1988). Note that some packages, such as the Mx program, also compute the total sample size that would be required (given the reported proportion of subjects in each group) to reject the hypothesis at various power levels. As stated in the introduction, power calculations usually take on one of two forms. First, one may be in the position that all information present in the raw data can be summarized in the covariance matrix Σ and the means vector μ, in which case Σ and μ are sufficient statistics, because they define the likelihood of the data up to an arbitrary constant (Azzelini 1996). In that case, one can derive the expected population statistics Σ and μ for every group in the study design under H1 and H0, and base the power calculations on these summary statistics. The second method of power calculation is applied when sufficient summary statistics are not available. For example, the population statistics Σ and μ do not summarize all information present in the raw data when the continuous data are a mixture (i.e., a convex combination of different distributions; McLachlan and Peel 2002), or when data are missing at random (MAR; Shafer and Graham 2002). In family studies, gene by environment interaction may render the summary statistics insufficient. For instance, Purcell (2002) showed how environmental moderation on the means and variances can be modeled (see Fig. 2). For both MZ and DZ twins, the variance of twin i is calculated as: while the for MZ twins, covariance between twin i and twin j is calculated as: and for DZ twins as: The expected values of means for all twins are calculated as: Fig. 2Univariate ACE-twin model including moderation on the variances and the means If the environmental moderator is categorical or ordinal (e.g., gender or affection status), the sufficient statistics (Σ and μ) are available, assuming that the data are normally distributed conditional on the levels of the moderator. In that instance, the twins in a pair may be concordant with respect to the moderator (i.e., both twins score 0, or both twins score 1), or the twins may be discordant with respect to the moderator (i.e., scoring 0 and 1, respectively). For both MZ and DZ twins, Σ and μ can be formulated for all possible combinations, such that all information present in the raw data is summarized with 6 different variance covariance matrixes Σ (3 for the MZ twins, and 3 for the DZ twins) and 3 different means vectors μ (assuming no relation between zygosity and mean), as such distinguishing 6 different groups. By comparing the fit of the model including moderator effects on both the means and the variances, to a model in which the moderator only affects the means (say), one can obtain an estimate of non-centrality parameter λ, from which the power to detect the effect of the moderator can be derived. However, if the moderator is continuously distributed, sufficient statistics cannot be calculated. In the absence of sufficient statistics, power calculation may be conducted by means of Monte Carlo simulation. This design implies determination of the values of the parameters of interest (e.g., based on previous studies or corresponding to realistic effect sizes), and subsequent (quasi-) random data generation according to the true model H1, with realistic sample size N. By fitting the false model H0 to the data simulated according to the true model H1, an indication of the power is obtained. However, in contrast to the situation in which sufficient summary statistics are available, parameter values are not recovered exactly when the H1 model is fitted, as the random data are the outcome of a stochastic sample process. Therefore, the difference in the χ2 statistic of the model H0 and the model H1 cannot be taken as an exact estimate of the non-centrality parameter λ. To solve this, a large number of datasets are usually generated, and λ is estimated as the mean of the difference in χ2 obtained in these data sets minus the number of degrees of freedom (df1 − df0). Since power studies often concern multiple parameters with multiple values, such Monte Carlo simulation studies can be prohibitively intensive. As an alternative, simulated sample sizes may be chosen very large to induce asymptotic behavior of the χ2 statistic. However, how large a sample size should be chosen depends on the study design in question, and very large sample sizes also render the analyses computationally intensive. Exact data simulation Power calculations based on sufficient summary statistics are computationally relatively efficient to carry out. However, the actual feasibility of this type of power calculation depends on the number of distinct groups. If the number of groups is large (i.e., >100), it may be more convenient to carry out Monte Carlo based power calculations. We now introduce the concept of exact data simulation, which shares the virtues of the power studies based on summary statistics, but is more practicable given a large number of distinct groups. The idea of exact data simulation is that data, which are randomly generated to begin with, can subsequently be transformed to fit the null-model H0 exactly. That is, first a data file is generated using a normal distribution quasi-random number generator. These data are then transformed, using a transformation proposed by Bollen and Stine (1993), so that the variance covariance matrix and means are exactly as specified under the model H0. Assume a total sample size of N, and k distinct groups with known probability pk. Let Y denote the Nk × q data matrix for group k, where Nk is N*pk (possibly rounded to the nearest integer) and q is the number of variables. Let m denote the q × 1 vector of observed means and S = YtY/(N − 1)−mmt be the observed covariance matrix, Σ the expected covariance matrix implied by model H1, and μ the expected means vector implied by the model H1. Let S1/2 and Σ1/2 then denote the square root factorization of the positive definite matrices S and Σ such as given by a Cholesky factorization. It can then be shown that the covariance matrix and mean vector of data matrix Z, which is obtained through the following transformation of Yequals Σ and μ, exactly (Bollen and Stine 1993). In Eq. 7, J is a unit vector of length q, and ⊗ denotes the Kronecker product. This transformation allows one to create raw data for numerous groups that fit the null-model exactly. Consequently, when the null-model used for the generation of the data is fitted to these transformed data, all parameter values used for the simulation are recovered exactly. Let Log L0 and Log LA denote the maximum values of log-likelihood functions. The difference 2Log LA − 2Log L0 equals the non-centrality parameter λ. Compared to multi-group power calculation with summary statistics, which becomes unpractical when the number of groups is large, the practicality of exact data simulation is unaffected by the number of groups in the analysis. However, one issue does require attention. When the number of groups is large (e.g., 256, see Illustration 1 below), the probability pk that a subject belongs to a group k may be relatively small. In order to calculate the Cholesky decomposition of the observed variance covariance matrix S, the number of observations in a group Nk should equal at least q + 1 (i.e., to ensure that S is positive definite and the Cholesky decomposition is possible). There are two ways to handle this problem. First, one can choose N to be sufficiently large that all groups, even those with small probabilities, by choosing a very large overall sample size N for the simulation. Power analyses based on this very large sample size produce non-centrality parameters which can subsequently be used to calculate power for other, more realistic sample sizes. Second, one can choose a smaller overall sample size, and accept that not all possible groups will be represented in the power calculations. This choice is usually justified since very small groups (e.g., including 2 subjects out of a possible 10,000) do not contribute much to the power. However, power calculations are more precise when all groups are represented in the simulation, i.e., overall N is large. Furthermore, it is in principle possible that the presence of all groups is required for model identification. Note that packages like Mx and R will estimate the number of data vectors required for a power of e.g. 80%, given the proportion of subjects in each group. So while Nk ≥ q + 1 is required for exact simulation, Mx will return an overall sample size N, in which many groups may represented by fewer than q + 1 observations, which is in line with what one would expect to observe in research practice. Using the exact data simulation script subsequently to simulate data with the sample size advertised by Mx would not result in a power of 80% since the groups with Nk < q observations are not represented in the simulated data and thus do not contribute to the power. Having discussed the concept of power, and the exact data simulation procedure, we will now illustrate the virtues of exact data simulation with three behavior genetics examples. We chose Mx to analyze the simulated data because of the program’s inbuilt option to calculate the sample size required for different power levels given the non-centrality parameter. However, the non-centrality parameter can also be obtained through other software (e.g., LISREL, Mplus). The calculation of sample sizes required for different power levels can then be done using other programs like R. The R-scripts used to simulate the data, and the Mx scripts used to analyze the data are available in the Mx scripts library.4 A small R-script for power computations based on non-centrality parameters can be downloaded from the library as well. Illustration 1: multivariate ACE-model with data MCAR Let us consider a four-variate ACE-model with data obtained from MZ and DZ twins (no additional family members). We assume a model with one common genetic factor, one common shared environmental factor, and specifics for A, C and E for all four traits. The model is illustrated in Fig. 3. Parameter values are chosen such that additive genetic influences, shared environmental influences, and non-shared environmental influences explain 50%, 30% and 20% of the total variance, respectively. Of the additive genetic variance, 60% is attributable to the common genetic factor (i.e., 30% of the total variance), and 40% to the specifics of A (i.e., 20% of the total variance). Of the shared environmental variance, 50% is attributable to the common shared environmental factor (i.e., 15% of the total variance), and 50% to the specifics of C (i.e., 15% of the total variance). Table 1 contains the correlation matrices for the MZ and DZ twins in the case that data are not missing. Means for all traits are equal to zero. We consider the power to reject the alternative hypothesis that the genetic specifics for all four traits explain not 20% but only 5% of the total variance,5 i.e., one common genetic factor is (almost) sufficient to explain all genetic variance and covariance in the four traits. If there are no missing data (situation S1), then one could simply use summary statistics to obtain power information, as this is a 2-group analysis. However, suppose we want to study the influence of data missing completely at random (MCAR) on the power to reject the hypothesis that all genetic specifics are zero. Here we consider two scenarios. First, we study the case that the probability of data being MCAR is 20% for all variables (situation S2). Given that we have q = 2 × 4 observations per family, this kind of missingness could yield 28 − 1 = 255 possible data patterns, i.e., 255 different groups (we discard the group in which all data are missing). In that case, power calculations using summary statistics are impractical, whereas exact data simulation is feasible. Note that some data patterns are rather unlikely, e.g., the probability of observing a valid observation for the first trait of the first twin only, while all other observations in the family are missing, is .8*(.27) = 1.024−05. Remember that the Cholesky decomposition cannot be calculated if the number of observations in a group Nk does not equal at least q + 1, so the simulated N needs to be very large if one wants all data patterns to be present in the simulated data set (about (q + 1)/1.024−05 ≈ 900,000). Yet, as very rare observations will hardly contribute to the power (5 of the 900,000 cases in the present example), one can just as well adopt a smaller sample size, and accept the fact that some groups (i.e., patterns of observations) will not be represented, and calculate power given the most likely patterns of observations. Fig. 3Four-variate ACE-model with common factors for additive genetic and shared environmental effects, and specifics for A, C and ETable 1Illustration 1: Four-variate cross-trait-cross-twin MZ correlations (below diagonal) and DZ correlation (above diagonal) for data without missingnessTwin 1Twin 2Trait1Trait2Trait3Trait4Trait1Trait2Trait3Trait4Twin 1Trait11.00.45.45.45.55.30.30.30Trait2.451.00.45.45.30.55.30.30Trait3.45.451.00.45.30.30.55.30Trait4.45.45.451.00.30.30.30.55Twin 2Trait1.80.45.45.451.00.45.45.45Trait2.45.80.45.45.451.00.45.45Trait3.45.45.80.45.45.451.00.45Trait4.45.45.45.80.45.45.451.00Note: Sample size N is not reported; as the simulations are exact, this correlation matrix should result independent of the sample size chosen for the simulations when data are not missing Second, we study the case where the probability of data being MCAR is 40% for the first two variables, and zero for the second two (situation S3). This could be a realistic scenario in practice, for example, when a questionnaire study that measures two traits is extended during the data collection to include two additional traits, or when data from different studies are combined (one study in which all four traits were measured, while another study only included measurements of two traits for, e.g., economical reasons). Given that we have q = 2 × 4 observations per family of which only four variables show missingness, this would yield 24 = 16 possible data patterns, i.e., 16 different groups. With only 16 groups, power calculations using summary statistics would be feasible. However, one efficiently setup exact data simulation script can handle both this simple pattern of missingness, and more complex. For the simulations we chose an overall sample size of 50,000 families (1/3 MZ, 2/3 DZ), which means that for situations S1 and S3, all groups (data patterns) are represented (50,000 and 49,999 cases simulated, respectively), while for situation S2, only the 163 most likely groups of the possible 255 are represented (49,999 cases simulated). The three simulated data sets were subsequently analyzed in Mx. In the Mx-script, we specify different groups for the MZ and DZ twins. Because we use full information maximum likelihood to accommodate the missingness, we do not need to specify different groups for all possible missing data patterns. The Mx command ‘option power’ (α = .05, df = 4) was used to obtain an estimation of the total sample size that would be required for a power of 80%, given the current proportions of subjects in each group. We find that for situation S1 (no missingness), 302 families are required for 80% power to reject the alternative hypothesis that the genetic specifics for all four traits explain 5% rather than 20% of the variance each, while for situations S2 (20% missingness for all variables) and S3 (40% missingness for only variable 1 and 2) the number of families required to obtain 80% power is estimated at 494 and 474, respectively. We hasten to note that these results are not informative for the case that data are missing at random (MAR), rather than MCAR (see Schafer and Graham 2002, for a comprehensive review on missingness and statistical procedures for handling missing data). These power calculations took about 2 min for each situation S. Within the Monte Carlo framework, acquisition of similar power results would take at least T times as long for each situation S (where T is the number of replications one chooses to do). Given that the time required to write the data simulation script is equal for Monte Carlo simulation and exact simulation, it is clear that exact simulation saves a lot of time. Illustration 2: gene by environment interaction with latent G and measured, categorical E Gene by Environment (G × E) interaction is an important issue. From the perspective of the power study, a problem with the presence of G × E when the E is continuously distributed is that it renders single summary statistics insufficient; in the presence of G × E, (co)variances and means depend on the level of the environmental moderator, as we have seen in Eqs. 3–6. Purcell (2002) showed how G × E on the means and variances can be modeled if G is latent, and E is measured. In power calculations in the G × E context, one can adopt a multi-group design, if the environmental moderator is categorical. For example, consider a classical ACE-twin design. If the environmental moderator is dichotomous (e.g., males versus females, young versus old, smoking versus non-smoking), the sample consisting of MZ and DZ twin pairs can be split up into twin pairs who are concordant with respect to the moderator (e.g., both twins do, or do not, smoke), and twin pairs who are discordant with respect to the moderator (only one of the twins in a pair smokes). With a dichotomous moderator and only two subjects per family, power calculations using summary statistics are feasible as there are only three distinguishable groups (not accounting for the distinction between MZ and DZ twins). However, suppose that you have measured an environmental moderator with 4 levels (coded 0, 1, 2, 3) in twin-pairs and their parents. With four persons per family and four possible moderator levels, there are 44 = 256 possible family configurations. In that case, multi-group analysis with summary statistics is impractical, and exact data simulation may be used instead. For this illustration, parameters a, c, and e were all set to 1, such that the total variance equaled 3 (excluding moderating and main effects). The moderator, which was assumed independent of genotype in this illustration, was coded 0 to 3, such that the group with 0 on the moderator can be considered the baseline condition. The probability for all moderation levels was set to .25, and moderation levels were modeled as independent across family members (i.e., the probability for each family member’s moderation level was independent of the moderation levels of the other family members). Moderation in C, E and the means was fixed to 0, but the regression weight of the moderator was set to .2 for the additive genetic effects, such that the moderator explained 20% of the variance in the total population (i.e., 0%, 13%, 24%, and 34% of the variance, respectively, depending on the level of the moderator). The model is illustrated in Fig. 4. For the simulation, we chose an overall sample size N of 10,000 (1/3 MZ and 2/3 DZ twins). Fig. 4Univariate ACE-model for parents and twin-offspring, including moderation on the variances and the means Given these simulated data, we want to estimate the power to reject the alternative hypothesis that all moderator effects on the variances are zero (i.e., no G × E, or C × E, or E × E). In practice, one would fix all regression weights concerning the moderating effects on the variances (βa, βc, and βe) to zero at once, resulting in a test with 3 degrees of freedom. The data were analyzed in Mx: different groups were specified for the MZ and DZ twins, and the moderator featured as a so-called definition variable. The ‘option power’ command (α = .05, df = 3) was again used to obtain an estimation of the total sample size that would be required for a power of 80%, given the current proportions of subjects in each group. With the probability for all moderation levels fixed to .25, all 256 groups were represented in both the MZ and the DZ twins in the simulated data file (9,984 cases simulated), and for the chosen values, the analysis shows that we would need 165 families for a power of 80%. If we were to change the moderator level probabilities from .25 for every level to .4, .3, .2, and .1 for levels 0, 1, 2 and 3 respectively, then 179 and 225 groups would be represented in MZ and DZ twins, respectively (9,705 cases simulated). In that case, 206 families would be required for a power of 80% even though the moderator effect (βa) is unchanged. These power calculations took at most 1 minute in total. Again, acquisition of similar power results would take at least T times as long within the Monte Carlo framework (where T is the number of replications). Assembling the data simulation script takes equally long for both types of simulation, so overall, exact simulation saves time. Illustration 3: association for a tri-allelic locus with different allele frequencies The aim of association studies is to determine whether genetic variation is associated with the risk for disease or the expression of a continuously distributed trait. Association studies may produce false positives, i.e., significant association in the absence of any true genetic effects. Population stratification is one source of false positives, i.e., the mixture of two populations with different allele frequencies and different phenotypic means. Fulker et al. (1999) showed that this type of spurious association can be avoided in a family-based study design. In this illustration, we focus on the situation in which data are available for pairs of siblings. Although this design allows for the simultaneous modeling of linkage and association, we limit the analysis to the association, but note that linkage information (i.e., IBD sharing estimation) could be included in exact data simulation scripts. For the present illustration, however, we assume that the locus under study is the QTL itself and not a marker in linkage disequilibrium with the QTL. If a locus is diallelic, 22 = 4 genotypes can be distinguished: AA, AB, BA and BB (of course, in practice, there are only 2 + 1 = 3 distinguishable groups as AB and BA are the same, but when simulating the data exactly, it is convenient simply to treat them as different groups). These 22 genotypes give rise to (22)2 = 16 possible combinations of siblings (not accounting order), i.e., a 16 group analysis. Note that this is the simplest case: with 3 alleles, the number of possible sib-pairs is already (32)2 = 81, and when the locus under study is a polymorphic marker, with, say, 15 possible alleles, the number of distinguishable sib-pairs is (152)2 = 50,625. Clearly, multi-group analyses with sufficient summary statistics quickly become impractical as the number of alleles––or loci––increases. We illustrate the use of exact data simulation in the context of the sib-pair association design, for a tri-allelic locus with alleles A, B, and C, with frequencies p, q, and r, respectively. The aim of this particular power calculation is to determine the influence of the allele frequencies on the power to detect a QTL. The biometrical model for a tri-allelic locus is summarized in Table 2. As with the more familiar diallelic case, the expected genotypic value is assumed zero, so that everything is scaled in terms of deviations. In the case of three alleles, 2 genotypic values are distinguished, which were both fixed to .206, so that AA was associated with an increase of .206, BB with an increase of .206, and CC with a decrease of−.206 to .206 = −.412. Dominance was assumed to be absent, so the genotypic effect for the heterozygous genotypes AB, AC and BC was calculated as the mean of the effects of the homozygous groups. In the case of equal allele frequencies (p = q = r = 1/3), this QTL explains 2.5% of the variance (as determined using regular regression with the phenotype as dependent variable and genotype as predictor). Note that the variance explained by the QTL depends on the allele frequencies, so even though the genotypic values remain the same across all simulations, varying the allele frequencies affects the effect size of the QTL effect. For all simulations, background variance was decomposed such that additive genetic effects explained 30%, and unique environmental influences (E) explained 70% of the variance that remained after the QTL-effects was taken into account. Overall sample size N was fixed to 10,000 (note that because of rounding, the actual N modeled will not be equal to the overall sample size N of 10,000; see Table 3).Table 2Expectations for a tri-allelic locus following the standard biometric model when dominance is assumed absentGenotypeAAABBBACBCCCGenotype frequency fijp22pqq22pr2qrr2Genotypic value gijx(x + y)/2yx + z/2 = −y/2y + z/2 = −x/2zNote: p, q, and r denote the frequencies of alleles A, B and C, respectively; x is the genotypic value associated with genotype AA, y the genotypic value associated with genotype BB. As the genotypic value for genotype CC is z = −x − y (i.e., x + y + z = 0). μqtl denotes the expected contribution of the QTL to the population mean, and σ2qtl denotes the expected contribution of the QTL to the population variance (adapted from Falconer and Mackay 1996)Table 3Results for illustration 3: Power calculations for sib-pair association with a tri-allelic locus with fixed genotypic valuesFrequencies alleles A, B, CEffect size (%)Actual NNr of groups represented χ2(6)Observed power N required for power of 80%.33/.33/.332.59,63981837.8241157.25/.5/.251.79,98581734.3571185.45/.45/.1.69,99381352.9631386.1/.45/.452.89,99280966.6981141Note: Effect size is defined as % of variance explained by QTL; Actual N refers to actual number of sib-pairs in the analysis; Nr of groups represented refers to the number of groups, of the possible 81, that were represented in the analysis; χ2(6): the χ2-value of the test for association when the genotypic effects for all 6 distinguishable genotypes are fixed to zero; Observed power refers to the power observed for the modeled sample size N Note that between and within effects were exactly equal (i.e., B = W) as we did not model population stratification; all between and within parameters can thus be fixed to be equal without loss of fit. The overall test for genetic association then involves fixing the genotypic effects for all 6 distinguishable genotypes (AA, AB, AC, BB, BC, and CC) to zero, i.e., 6 degrees of freedom. The simulated sib-pair data were analyzed in Mx, using the ‘option power’ command (α = .05,df = 6) to obtain the sample size required for a power of 80%, given the current proportions of subjects in each group. The results presented in Table 3 show that the power to detect a QTL with certain genotypic values depends on the allele frequencies. As expected, the power is greatest when the frequency for the allele with the largest genotypic value (allele C) is highest. With exact data simulation, these power calculations took about 2 min for each choice of allele frequencies. Again, it would take at least T times as long to obtain similar power results within the Monte Carlo framework (where T is the number of replications), while the time required to write the data simulation script takes equally long for both types of simulation. Conclusion In this paper we discussed a third method of power calculation, which can be useful when sufficient summary statistics are available in principle, but the number of possible groups is so large to render a multi-group analysis impractical. The illustrations presented in this paper represent only a few of the possible (behavior genetics) designs in which exact data simulation may prove useful. Other models for which exact data simulation can be used include random-effects models, latent growth curve models, simplex models, and (hierarchical) structural models, either or not in the context of genetics, just to name a few. Exact data simulation does not require more programming skills, or programming time, than Monte Carlo simulation, but one may save a lot of time analyzing the simulated data and calculating power, especially when one wishes to construct graphs of power vs. effect size. In this paper, we used the Mx program to analyze the simulated data because of its inbuilt power calculation function. Another useful option of Mx in this context is the possibility to output individual likelihood statistics for each raw data group. This information can be used to identify the groups that contribute most to the power to detect the effects of interest. Of course, various other statistical software packages (e.g., QTDT, LISREL, MPlus, R) can also be used in combination with exact data simulation to obtain the non-centrality parameters required for power calculations. We emphasize that the power results obtained through exact data simulation are exactly similar to power results obtained through the analysis of summary statistics, and, just like power calculation using summary statistics, asymptotically similar to results obtained through Monte Carlo simulation (depending on the number of runs used in Monte Carlo). Differences between those two customary method of power calculation and exact data simulation only occur when subgroups have very low probabilities and the simulated overall sample size is not large enough to include all possible groups to sufficient extent; these small groups may then not be represented in the exact power simulation, while they may be (more or less) represented in other methods. However, as stated previously, the ensuing differences with respect to the power results, are very small as such small groups hardly contribute to the power anyway. Even so, to avoid the exclusion of small groups, one should choose a sufficiently large overall sample size in exact simulation, such that all groups are represented. This is perfectly doable, and does not alter the practicability of the method as it still involves analyzing a single (yet larger) dataset. The non-centrality parameter obtained in the analysis of the large simulated data set can subsequently be used to calculate the power for smaller, more realistic sample sizes. Alternatively, one may decide to accept the absence of certain groups, and the implied slight underestimation of power. Happily, the discrepancy between the intended N and the realized N is simple to calculate (as demonstrated in the R script available in the Mx scripts library), so that one can readily obtain an impression of the implications of this decision. Throughout the paper, we have assumed that the data, conditional on group, are normally distributed, so that sufficient statistics are in principle available. With respect to situations that preclude sufficient statistics, the present method may still have some use. For instance, a continuous moderator in a G × E model, as discussed by Purcell (2002), might be approximated by a 5 point or 7 point Likert scale, which would render exact simulation possible in principle (see illustration 2). Finally we note that the extension of this method to discrete data would obviously be very useful, and does seem feasible.

Photosynth_Res-3-1-1779625

A re-investigation of the path of carbon in photosynthesis utilizing GC/MS methodology. Unequivocal verification of the participation of octulose phosphates in the pathway

A GC/EIMS/SIM methodology has been developed to re-examine the path of carbon in photosynthesis. Exposing isolated spinach chloroplasts to 13CO2 on a solid support for a defined period followed by quenching and work-up provided a mixture of labelled sugar phosphates. After enzymatic dephosphorylation and derivatization, the Mox-TMS sugars were analysed using the above method. The purpose of the study was to try to calculate the atom% enrichment of 13C in as many of the individual carbons in each of the derivatized sugars as was practical using diagnostic fragment ions. In the event, only one 45 s experiment provided sufficient data to enable a range of enrichment values to be calculated. This confirmed that D-glycero-D-altro-octulose phosphate was present in the chloroplasts and was heavily labelled in the C4, C5 and C6 positions, in keeping with the hypothesis that it had an inclusive role and a labelling pattern consistent with a new modified pathway of carbon in photosynthesis. Introduction In two previous papers, we described a GC/MS methodology, which is capable of providing a rapid, specific and quantitative means of analysis of 13C incorporation from 13CO2 into those sugar phosphates involved in the path of carbon in photosynthesis (Irvine et al. 1992; MacLeod et al. 2001). This provided a framework for a re-investigation of the photosynthetic carbon reduction (PCR) pathway in plants, commonly known as the Calvin Cycle (Calvin 1956). Although the PCR pathway as presented by Calvin (Fig. 1) is widely accepted, its re-examination is warranted for the following reasons: (i) the evidence upon which Calvin based the pathway was incomplete, (ii) more recent investigations support the inclusion of other sugar phosphates, notably octulose phosphates, in the pathway (Williams and MacLeod preceding paper), and (iii) the substantial advances in analytical technologies since Calvin’s data was collected in the 1950s, using 14C-labelling and paper chromatography, have the potential to provide more definitive and comprehensive results than were possible for Calvin to obtain. Fig. 1The path of carbon in photosynthesis (Calvin 1956). Each asterisk (*) represents a carbon atom labelled during one turn of the cycle. For explanation of abbreviations, see preceding paper This paper describes the measurement, using GC/EIMS/SIM, of 13C incorporation into individual carbon atoms of C4 to C8 sugar phosphates following short-term exposure of isolated chloroplasts to 13CO2 during PS. Materials and methods Chemicals Sorbitol was purchased from Koch-Light, UK, sodium pyrophosphate from Mallinckrodt, USA and Percoll from Pharmacia, Sweden. Enzymes and cofactors were obtained from either Sigma, USA or Boehringer-Mannheim, Germany. Ion exchange resins were obtained from BioRad, USA. Analytical grade chemicals were obtained from Ajax, Australia, BDH, UK or Merck, Germany. All solvents were distilled before use while pyridine was pre-dried and distilled from calcium hydride under argon. Water was obtained from a Millipore Milli-Q system. The 13CO2 gas (99.9% atom enriched) was purchased from Cambridge Isotopes, UK and prepared to 1% in nitrogen in a gas cylinder. A second cylinder was made up with 1% unenriched CO2 in nitrogen. Gas analysis of the cylinders was carried out on a Varian 6000 gas chromatograph using a 1.83 m × 0.32 cm glass column packed with Porapak N 100–200 mesh stationary phase. Plant material Spinach (Spinacia oleracea, Yates Hybrid 102) seeds were obtained from Henderson Seed Co., Pty. Ltd., Lower Templestowe, Vic., Australia. Growth of plant material; chloroplast isolation; chlorophyll assay; measurement of chloroplast intactness; polarographic measurement of chloroplast activity in suspension The above methods have been described in detail in the accompanying paper (Williams and MacLeod preceding paper). Measurement of chloroplast activity on filter membranes The activity of the isolated chloroplasts mounted on a filter membrane was measured polarographically using a Clark oxygen electrode adapted for measurement of the oxygen evolution from leaf-discs (Delieu and Walker 1981). Methods for the preparation of chloroplasts for use in the leaf-disc oxygen electrode were based on those previously described (Cerovic et al. 1987). Chloroplasts were mounted on a filter membrane by filtering a suspension of chloroplasts through a cellulose nitrate membrane (Sartorius, 50 mm diameter, pore size 8 μm). This was accomplished using a Millipore solvent clarification apparatus fitted with a 100 ml funnel. A suspension of chloroplasts equivalent to 250 μg chlorophyll in 40 ml resuspension medium at pH 7.6 containing 330 mM sorbitol, 50 mM HEPES-KOH, 2 mM EDTA, 1 mM MgCl2 and 1 mM MnCl2 was allowed to filter through the membrane under gravity or with a very light suction so that the chloroplasts became trapped on the surface of the membrane in a thin film of resuspension medium. The chloroplasts were washed with a further 10 ml of the resuspension medium and then given a final wash with 3 ml of assay medium containing 6 mM inorganic phosphate, 22.5 U ml−1 alkaline phosphatase (Sigma, human placenta) and 1,000 U ml−1 catalase in resuspension medium. A disc of 1,018 mm2 (representing 94% of the chloroplasts) was cut from the membrane and mounted in a leaf-disc oxygen electrode (Hansatech). Once assembled, the leaf-disc oxygen electrode was charged with an atmosphere of 1% CO2 in N2 at atmospheric pressure and sealed. The chloroplasts were exposed to incandescent light from a pair of bifurcated fibre optic light sources at an intensity of 1,200 μEm−2 s−1 measured at the underside of the leaf-disc electrode light window. The temperature of the water bath supplying the water jacket, which regulated the oxygen electrode temperature was maintained at 23°C. After a lag period of approximately 1.5 min, the evolution of O2 could be monitored as a digital voltage on the oxygen electrode control box and was also recorded on a Goerz Metrawatt SE 120 chart recorder. The rate of O2 evolution reached a linear and maximum rate at approximately 4 min after switching on the lights. The rate was calculated from the linear portion of the trace. Experiments in the leaf-disc oxygen electrode Chloroplasts were prepared for experiments in the leaf disc oxygen electrode (as discussed for the assay of chloroplast activity on filter membranes in that equipment) and a linear rate of O2 evolution was established. At this time, a volume (60 ml) of 1% 13CO2 in N2 was flushed through the 6 ml chamber of the oxygen electrode that contained the membrane-mounted chloroplasts and after resealing the chamber, the chloroplasts were allowed to photosynthesize in the 13CO2-enriched atmosphere for a prescribed period of time before the reactions were terminated using the quenching methods described below. Experiments were conducted for 0, 30 and 45 s exposure to 1% 13CO2 in N2 and the “control” experiment was conducted with chloroplasts exposed to 1% 13CO2 in N2 for 45 s in the dark. In addition, a “blank” experiment was conducted in which the full extraction procedure was followed in the absence of chloroplasts to measure the extent of any contaminants that may interfere with the analysis. Five repetitions of each experiment were carried out using the same chloroplast preparation. Reactions were terminated by dissembling the apparatus as quickly as possible and plunging the chloroplast-containing membrane into a beaker of liquid nitrogen. It was estimated that this procedure took no more than 2.5 s. The beaker containing the membrane was then stored in a freezer at −20°C where the liquid nitrogen was allowed to evaporate. After overnight storage at −20°C the enzymes were denatured and the metabolites were extracted by removing the beaker containing the membranes from the freezer and immediately adding boiling 80% ethanol (75 ml) and boiling for a further 5 min. The membranes were then removed from the extract and thoroughly rinsed with water, adding the washings to the extract. After cooling, the pooled extracts were then evaporated to dryness at 37°C under a stream of dry nitrogen. Design of a specific photosynthesis apparatus Because of the limitations of the oxygen electrode for the type of experiments that were required, it was decided to design and construct an apparatus better suited for the purpose. The basic requirements in the design and construction of the photosynthetic apparatus were (1) that the chloroplasts be mounted on a filter membrane inside a small chamber with a transparent window for the admission of light, (2) the filter membrane should be large enough to hold sufficient chloroplasts to provide enough extracted material for GC/MS analysis, (3) the atmosphere above the membrane must allow the establishment of steady state photosynthesis with unenriched CO2 and the rapid replacement of this with the same concentration of 13CO2 whilst causing minimum disturbance to the steady state, (4) the volume of the chamber above and below the membrane should be small in order to minimize the time taken to flush out the 12CO2 with 13CO2 using a reasonable gas flow rate. Also the volume of the chamber below the membrane must be small in order to minimize the “dead volume” of residual 12CO2, and (5) there must be provision for an appliance to inject a liquid to quench the photosynthetic reactions after a given time and for the convenient removal of the chloroplast extract after the quench. Filter holders were available (Sartorius SM 165 08B) which, with minimal modification, provided a suitable basis on which to construct a prototype photosynthesis apparatus. The rapid replacement of unenriched CO2 with 13CO2 was provided for by the inclusion of two gas inlet ports each supplying an internal gas manifold around the outer perimeter of the chloroplast-supporting membrane. The manifolds directed 13CO2 across the entire membrane and drove excess gas out through a series of 6 gas outlet holes drilled in the housing above the centre of the membrane, which vented the expelled gases to the atmosphere. An electro-mechanical timer unit was constructed to permit the automatic and rapid switching of the gas supply from a gas cylinder containing 1% unenriched CO2 in nitrogen to another cylinder containing 1% 13CO2 in nitrogen via a pair of electrically operated gas valves, causing minimal disruption to the steady state of photosynthesis. The inlet and outlet ports for the filter holder provided suitable ports for the entry and withdrawal of the quenching solution. A 10 ml syringe containing the quenching solution was attached to the filter inlet port via a tap, which could be opened to allow the injection of the quenching solution at the completion of the photosynthesis experiment. By applying a gentle suction via a tap at the filter outlet port, the extract could then be withdrawn into a flask containing boiling 80% ethanol located below the apparatus. The chloroplasts and apparatus were protected from the heat produced by the hotplate by the provision of a reflective heat shield. Light was provided via fibre optic cables, which greatly reduced the transmission of heat from the light source and reduced the need for cooling of the apparatus. The concentrations reported for chloroplast metabolites are variable but have usually been found to be in the range from 5 to 50 nmol per mg chlorophyll (Gerhardt et al. 1987; Giersch et al. 1980; Giersch 1979). A filter membrane loaded with chloroplasts equivalent to 250 μg of chlorophyll should therefore contain 1.25–12.5 nmol of most metabolites of interest. Assuming minimal losses during processing for GC/MS analysis, this was considered to be adequate chloroplast material for each experiment. Experiments in the photosynthesis apparatus Chloroplasts were prepared on a filter membrane as described for the experiments in the oxygen electrode. The loaded membrane was mounted intact in the photosynthesis apparatus. A gas mixture consisting of 1% unenriched CO2 in nitrogen was applied to the two gas inlet ports via an electrically operated valve connected to a specially made electro-mechanical timer unit. The gas stream (at a flow rate of 200 cm3 min−1) was humidified by bubbling it through a tube of water before entering the valve. The chloroplasts were illuminated through the transparent window in the apparatus by a pair of 150 W tungsten lamps, each directed through a bifurcated fibre optic light guide. The gas mixture was supplied for 4 min before the timer operated the valves supplying the gas and switched to the source supplying 1% 13CO2 in N2 for the prescribed period. Experiments were conducted for 0, 6, and 30 s exposure to 1% 13CO2 in N2 and the “control” experiment was conducted with chloroplasts exposed to 1% 13CO2 in N2 in the dark for 30 s. The reactions were terminated and the metabolites were extracted by injecting hot 80% ethanol (5 ml) into the apparatus and allowing the chloroplasts to stand in the solution for 10 s. The ethanolic chloroplast extract was then drawn through the membrane using gentle suction into a flask that contained boiling 80% ethanol (30 ml). The extraction process was repeated with a further volume of boiling 20% ethanol (5 ml) followed by boiling water (5 ml). The combined extract was then boiled for 5 min and allowed to cool. After cooling, the extract was evaporated to dryness at 37°C using a stream of dry N2 gas. Preparation of samples for GC/MS analysis The dried samples from the photosynthesis experiments described above were redissolved in water (2 ml) and then processed for GC/MS analysis according to the following methods. The chloroplast extract to be analyzed by GC/MS was applied to an Elut Bond strong anion exchange (SAX) solid phase extraction column (500 mg, Varian Associates, USA) and washed through with deionized water (30 ml). The eluate was discarded and the fraction containing both mono- and bisphosphates of the sugars was eluted with 0.5 M NH4HCO3 (5 ml) and collected in a polypropylene centrifuge tube (50 ml). Most of the buffer was removed by blowing the sample dry with a stream of dry nitrogen gas, whilst warming the sample in a water bath at 37°C. The samples were then dephosphorylated using human prostatic acid phosphatase, which was chosen for its ability to catalyze dephosphorylation quantitatively without causing the transformations produced by other phosphatases (Irvine et al. 1992). Dephosphorylation of up to 0.5 μmoles of sugar phosphates was carried out by the addition of 0.5 units of prostatic acid phosphatase to the sugar phosphate solution in 0.1 M ammonium acetate–acetic acid buffer pH 4.6 (typically 200 μl) and incubated for 24 h at 30°C. The reaction was terminated by heating at 100°C for 2 min. Following dephosphorylation, the solutions containing the free sugars were deionized by stirring for 30 min with 0.5 g of mixed bed resin prepared using equal weights of anion (Bio-Rad AG 1 × 8, 200–400 mesh, HCO3− form) and cation (Bio-Rad AG 50 W × 8, 200–400 mesh, H+ form) exchange resins together with deionized water. The resin was removed by filtration through a small column fashioned from a pipette tip (5 ml) plugged with cotton wool and containing a further 0.25 g of the mixed bed resin. The sample was washed through with water (15 ml), collected in a polypropylene centrifuge tube (50 ml) and lyophilized. The resulting residue was washed with deionized water (1 ml) into an Eppendorf tube (1.5 ml) and again lyophilized. The residue was redissolved in water (120 μl) and the entire sample was transferred to a Reacti-Vial (200 μl) and dried over P2O5. The sugars were subsequently dried under high vacuum for 1 h. Derivatization of dephosphorylated sugar phosphates Sugars were prepared for GC/MS analysis as the methoxime-TMS derivatives using methods similar to those first described by Sweeley (Sweeley et al. 1963) for 1 mg scale derivatizations. In our investigation, smaller-scale derivatizations were required. Up to 350 nmol (approx. 50 μg) of the dried sugars in Reacti-Vials (0.2 ml) were derivatized by the addition, under dry nitrogen, of 1% methoxylamine in pyridine (40 μl) and refluxing at 80°C for 2 h in a Reacti-therm heater. The mixture was cooled to room temperature then silylation was accomplished by the addition under dry nitrogen, of 10 μl of Regisil (Pierce) followed by heating for 0.5 h at 80°C. The mixture was cooled to room temperature then diluted as appropriate with dry pyridine for GC/MS analysis. Injections of 1 μl ideally contained no more than 125 pmol of the sugar under analysis in order to avoid saturation of the mass selective detector, as determined by sensitivity runs with derivatized glucose. GC/MS analysis of derivatized sugars Derivatized sugars were analyzed by capillary GC/MS on a Hewlett Packard 5890 gas chromatograph interfaced with a Hewlett Packard HP5970B MSD and controlled by a Hewlett Packard HP59970C ChemStation. Gas chromatography was performed on an HP-1 capillary column (12.5 m × 0.2 mm i.d., 0.33 μm cross-linked methyl silicone stationary phase) in the splitless mode with helium carrier gas (flow rate 1 ml min−1). The column temperature was held at 100°C for 2 min with the filament off for the passage of the solvent and excess reagent through the column. The temperature was then increased to 250°C at 10°C min−1, and held at 250°C for 3 min. Spectra were obtained by electron impact (EI) ionization at 70 eV. Full scan spectra of the derivatized sugars were usually obtained by scanning from m/z 100 to 660 at a rate of 0.77 scans s−1 and represent an average of scans taken across the top of each relevant chromatographic peak. Selected ion monitoring (SIM) experiments on clusters of ions of interest in the spectra of the derivatized sugars were carried out in groups of 8–12 ions per peak with dwell times of 30 ms per ion. All results represent the mean of at least three separate injections unless otherwise stated. Corrections for natural abundance isotopes of C, H, N, O and Si and contributions from adjacent ions one mass unit lower of comparable intensity were made in calculating 13C enrichments in individual ions. Only fragment ions which had been found to have essentially a single origin as the result of the GC/MS analysis of specifically 13C-labelled sugar standards were used in this study (Irvine et al. 1992; MacLeod et al. 2001). Identification of sugar phosphates in chloroplast extracts The individual unlabelled sugar phosphates present in the chloroplast extracts were identified by the retention times and full scan mass spectra of their dephosphorylated Mox-TMS derivatives on the above GC/MS system when compared with those of authentic sugars run under identical conditions (Irvine et al. 1992; MacLeod et al. 2001). Quantification by GC/MS of sugar phosphates in chloroplast extracts Quantification of analytes is best achieved by the inclusion in the sample mixture of a known amount of an internal standard having similar chemical and physical characteristics to those of the analytes. Ideally, for GC/MS analysis the internal standard should be a stable isotope analogue of the analyte under investigation and when there are multiple analytes, a stable isotope of each analyte is preferred. In this investigation however, the analytes (derivatized sugars) are themselves expected to become multiply-labelled with the 13C isotope during photosynthesis and the stable isotope dilution method therefore could not be used. The alternative approach was used of adding a compound that is similar to the sugars under investigation but is unlikely to be encountered in chloroplasts and would be well separated from the analytes on GC/MS. For SIM, the mass spectrum of the derivatized standard chosen for this GC/MS investigation needed to contain an ion which was common to all of the sugars being measured. In addition, this ion must not become labelled in a 13CO2 experiment. The sugar alcohols seemed to be an obvious choice. The mass spectra of the per-TMS derivative of these compounds show a prominent m/z 147 ion representing the pentamethyldisiloxonium ion ([(CH3)3SiOSi(CH3)2]+), as do the mass spectra of all of the sugar methoxime TMS derivatives. Since this ion does not contain any carbons derived from the sugar, it cannot become labelled during a 13C isotope incorporation experiment. Thus, SIM of the m/z 147 ion was the method of choice for quantitation studies. Erythritol was chosen as the internal standard because (1) the m/z 147 ion of its per-TMS derivative gave a linear and reproducible response when calibrated against the same ion in the mass spectra of Mox-TMS sugars at levels up to 120 pmol and (2) it was well separated on the HP-1 GC column from the sugars under consideration. Sugars which were available in the free form with a high degree of purity were dried for 3 days over phosphorous pentoxide under a low vacuum. Approximately 0.02 g of the dried powder or syrup (ketopentoses) was weighed to five decimal places to prepare standard solutions of each sugar. Appropriate dilutions of these were taken to prepare a set of 0.5 mM standard solutions. Some sugars were only available as the phosphate esters and frequently in the presence of various amounts of buffer salts. In these cases, after weighing the dried powders and preparing standard solutions, they were assayed enzymically to establish their concentrations prior to use. Sedoheptulose, D-glycero D-ido- and D-glycero D-altro-octuloses were assayed as their bisphosphates using an aldolase-based assay (Bergmeyer and Bernt 1974). Erythrose was assayed as the monophosphate ester using a transketolase-based assay (Paoletti et al. 1979). Sugar phosphates (3 μmol) were then evaporated to dryness in a rotary film evaporator and dephosphorylated in a solution (0.5 ml) which contained citrate buffer 50 μmol, pH 5.4) and human prostatic acid phosphatase (1.25 units). After allowing the dephosphorylation to proceed for 4 h at 37°C, the reaction was terminated by heating at 100°C for 2 min. The samples were then treated with mixed bed resin and lyophilized. Recovery from the dephosphorylation step was assessed by carrying out parallel dephosphorylations of glucose 6-phosphate, fructose 6-phosphate and fructose 1,6-bisphosphate, all of which were assayed enzymically as the free sugars. The recovery of the monophosphates was 88% and of the bisphosphate was 78%. These figures provided a useful estimate of the likely recoveries of those sugars (above), which could not be assayed in the non-phosphorylated form. Appropriate dilutions of the lyophilized sugars were then prepared to produce a set of 0.5 mM solutions for use as standards. Groups of sugar mixtures were prepared, each containing 500 μl of the 0.5 mM standard solution. Each sugar was thus present in a volume of 2 ml and at a concentration of 0.125 mM. The first group contained erythrose, fructose and D-glycero D-ido-octulose. The second group contained glucose, sedoheptulose and D-glycero-D-altro-octulose. Pentoses were calibrated separately. A series of dilutions of the standard mixtures containing 0.025, 0.125, 1, 2, 3, 4 or 5 nmol of each sugar together with 5 nmol of the erythritol standard were added to separate ReactiVials. The mixtures were then first dried over P2O5 then under high vacuum and derivatized (Mox-TMS). This gave solutions which contained 0.5, 2.5, 20, 40, 60, 80, or 100 pmol of each sugar together with 100 pmol of the erythritol standard in each μl of the derivatized solution. GC/MS analysis was carried out by the measurement of the peak areas obtained for the m/z 147 ion of each sugar using single ion monitoring. Three injections were made at each sugar level and the average ratio of sugar to erythritol peak area was determined. In those cases where the Mox-TMS derivative of the sugar showed peaks for both syn- and anti-forms, the area of the leading peak was used. A graph plotting the measured ratio of each sugar m/z 147 ion area to the erythritol m/z 147 ion area against the actual ratio of the sugar to erythritol for the series of sugar concentrations was then plotted. The slopes of these graphs then represent the calibration factors relating the sugars to the internal standard, erythritol, and were determined as the slope of the least squares line of regression of measured area ratio on actual ratio (Table 1). Table 1Sugar calibration factorsaSugarInterceptSlopeCoefficient of correlation (r)Erythrose−0.0171.0680.999Ribose/Arabinose0.0010.6250.998Xylulose/Ribulose−0.0101.0770.999Fructose−0.0080.3050.995Glucose−0.0110.5240.998Sedoheptulose−0.0130.6780.998D-g D-a-Octulose−0.0090.3180.996D-g D-i-Octulose−0.0050.1770.992aErythritol used as internal standard Results and discussion Initial experiments were carried out using the leaf-disc oxygen electrode during the time that the special photosynthesis apparatus was being constructed in-house. Obvious disadvantages inherent in this approach were (1) the difficulty in achieving complete, instantaneous displacement of 12CO2 with 13CO2 and (2) the time taken to dissemble the apparatus and quench the photosynthesis reaction after exposure of the chloroplasts to 13CO2, leaving open the possibility of partial back exchange of the 13C-labelled sugars. One advantage was that the photosynthetic activity of the chloroplast preparations could be measured in situ prior to carrying out the 13CO2 exchange experiments. There were several pitfalls that had to be overcome before PS exchange experiments with 13CO2 could be performed. Foremost of these was the low concentrations of sugars of the PS pathway observed in the chloroplast extracts in initial experiments using unlabelled CO2, together with the presence of other sugar-like components in the GC/MS analysis . The latter was eventually traced to adsorption and decomposition of the small quantities (pmol) of sugar phosphates present in the chloroplasts on the surfaces of glass vessels used during the workup procedure. To minimise decomposition, these were replaced where possible with non-glass vessels. Losses due to adsorption could not be completely avoided and necessitated combining extracts from five consecutive experiments (using the same chloroplast preparation) in order to try to obtain useful data on those sugars present in lowest concentrations. Contaminants introduced during the workup presented another problem in the GC/MS analysis, which required extensive investigation of their sources. The only contaminant that could not be completely eliminated was sorbitol but this did not interfere with the sugar analysis as the retention time of its TMS derivative differed from those of the Mox-TMS sugars under investigation. The experiment carried out in the oxygen electrode at zero time exposure of the chloroplast preparation to 1% 13CO2 in nitrogen showed, after work-up, no incorporation of label in the sugars. As well, the “control” experiment carried out with chloroplasts exposed to 1% 13CO2 in the dark for 45 s produced no 13C enrichment in any of the sugars in the chloroplast extract. After analysis of the results of four experiments carried out in the oxygen electrode, involving exposure of the chloroplasts to 13CO2 for 30 s (twice) and for 45 s (twice), only one of the 45 s experiments produced a reasonably comprehensive set of results (Table 2). The main factor contributing to the inadequacy of the remaining experiments was the lack of sufficient material to enable GC/MS/SIM runs to be performed on all of relevant ion clusters as set out in a previous paper (see Table 2, MacLeod et al. 2001). Table 2Enrichment resultsa for 45 s experiment No. 2 in the leaf disc oxygen electrodeSugarm/zCarbons13C013C113C213C313C413C5Xylose160C1-267.224.28.6205C4-580.319.7262C1-351.026.813.48.9307C3-561.221.512.54.7452C1-547.121.717.913.3Lyxose/Arabinoseb160C1-266.724.88.5262C1-355.530.713.7307C3-568.518.912.6Xylulose205dC4-590.35.64.1263C1-354.626.513.55.4364C1-450.023.815.58.12.5452C1-546.522.116.99.93.90.8Ribose/Ribulosec160C1-267.024.78.3205dC4-581.911.76.4262/3eC1-359.123.112.75.1307C3-566.117.611.05.3319C2-550.424.114.77.92.8Fructose205C5-689.510.5263C1-365.917.811.64.8307C4-668.216.910.14.8319C3-658.620.311.86.42.9364C1-449.622.316.28.73.2Glucose160C1-273.619.27.2205dC5-678.014.67.4307fC4-660.921.511.66.1319C3-636.523.122.212.95.3364C1-433.227.322.212.54.9Unidentified heptulose307C5-757.422.612.97.1319C4-726.327.025.615.06.1421C3-724.324.524.116.38.02.8Sedoheptulose205dC6-772.419.48.3262C1-341.429.617.39.02.6307fC5-752.025.115.07.9319C4-725.526.025.616.26.7364C1-431.727.422.513.25.3466C1-517.021.526.220.010.74.5Octulose205dC7-878.614.17.4307fC6-858.222.912.76.3319C5-841.825.219.210.23.6331C4-831.221.521.314.57.83.8421C4-833.621.521.014.46.92.6466C1-534.622.321.513.36.02.3aAfter correction for natural abundance of 2H, 13C, 15N, 18O, 29, 30Si and contributions from some neighbouring ions. Results represent an average of at least three consecutive GC/MS runs. Standard deviations for most sugars were <1% and for the less abundant sugars (xylose, arabinose/lyxose, octulose and unidentified heptulose), standard deviations were <2%. When significantly different enrichments were measured for the syn and anti-methoxime isomers, it was assumed that some interference was present causing the higher result and the one showing the lower enrichment was used; otherwise both syn and anti-isomers where present were used in the calculationsbLyxose/Arabinose refers to either one or both sugars which could not be separated on the capillary GC column used in these analysescRibose/Ribulose refers to both sugars, which, although they could not be resolved were both shown to be present from their characteristic mass spectra (MacLeod et al. 2001)dThe intensity of the m/z 207 ion is enhanced due to column bleed, despite background correctioneThe MS of Ribose contains a small m/z 262 ion while that of Ribulose has a large m/z 263 ionfThe m/z 305 ion contributes to m/z 307 and 308. No correction has been made for this Experiments were then carried out using the custom-made photosynthesis apparatus, involving exposure of isolated chloroplasts to 1% 13CO2 in N2 for 30 s in the dark and for 0, 6 and 30 s under controlled light conditions. After work-up, both the “dark” experiment and zero time exposure showed no 13C incorporation in any of the sugars analysed. As was the case for all but one of the oxygen electrode experiments, the 6 s and 30 s experiments after work-up did not contain a sufficient concentration of the sugar Mox-TMS derivatives, even after combining five experiments, to enable selective ion profiles to be obtained on all of the relevant ions in all of the sugars. This was particularly disappointing as the specially designed apparatus performed well otherwise. In all of the above experiments, the concentrations of the sugar phosphates present in the chloroplast extracts were determined by SIM of the m/z 147 ion present in the mass spectra of the dephosphorylated Mox-TMS derivative, using erythritol as internal standard (see Materials and methods). A correction was made for losses incurred during the chloroplast work-up and dephosphorylation stages by adding 14C-labelled glucose 6-phosphate to a single membrane-mounted chloroplast preparation immediately prior to quenching in the special apparatus. The measured recovery figure of 73% of 14C-labeled glucose represents a minimum recovery value since the method eventually used involved the pooling of the extracts from five membranes, which would be expected to lead to reduced losses. Table 3 lists the values obtained. As expected, the two “blank” runs showed only traces of glucose, fructose and pentoses while the 45 s “dark” experiment showed lower than normal levels of all sugars except glucose. The other six experiments, five using the oxygen electrode and one using the special apparatus, showed sugar concentrations which, although varying from experiment to experiment, were of the same order of magnitude to those previously reported (Lilley et al. 1977). Table 3Chloroplast metabolite concentrations (nmol (mg chl)−1)Equipment for experimentApparatusLeaf disc oxygen electrodeExposure to 13CO2 (sec:)03004503045CompoundBlankaBlankaControlbNo 1No 2No 1No 2Erythrosec0.000.000.000.000.280.140.550.79 0.35Xylose0.000.390.000.000.000.000.861.100.61Lyxose/Arabinose0.000.630.000.000.000.001.050.680.81Ribose/Ribulose/Xylulose0.205.740.211.864.702.549.7914.876.61Fructose0.2933.710.312.6729.6623.4842.06117.658.43Glucose0.3310.550.356.718.213.6015.6339.3816.18Unidentified heptulose0.000.000.000.000.290.000.851.040.84Sedoheptulose0.019.040.011.0612.971.1721.9538.9016.26D-g D-a-Octulose0.0000.170.000.280.400.001.533.53daExtract containing chloroplast medium without chloroplastsb45 s exposure to 13CO2 in the darkcTentatively identifieddm/z 147 ion not monitored No C3 sugars could be analysed under the GC conditions used since they eluted with the residual derivatizing agents. A tetrose sugar was present which had the same mass spectrum and elution time as erythrose (Table 3) but subsequent analysis of label incorporation into selected ions showed no 13C enrichment. The unidentified heptulose listed in Table 3 did show very significant 13C incorporation in the few ion clusters measured (Table 2), comparable to that observed in sedoheptulose. D-g-D-a-Octulose was identified by the retention times of its syn- and anti-isomers but the earlier eluting D-g-D-i-octulose was not detectable. The enrichment results obtained in Table 2 were sufficient to calculate atom% excess in some but not all of the individual carbon atoms of the sugars listed (see Table 2 in MacLeod et al. 2001). For example, the enrichment in C-2 of ribose (23.8%) was obtained by subtraction of m/z 307 (C3-5) from m/z 319 (C2-5). Similarly, subtracting the 13C enrichments in the ions m/z 307 and 205 (C4-5) gave the atom% excess in C3. C-1 (12.1%) was calculated by subtracting the above value for C2 from the enrichment in the m/z 160 ion (C1-2). In some cases where it was not possible to obtain enrichment data for individual carbon atoms, values for consecutive two carbon units were calculated. While it is obvious that the results obtained fall short of the objective, for reasons given above, they nevertheless allow some pertinent observations to be made. The presence of D-g-D-a-octulose in isolated chloroplasts at concentrations between 0.2 and 3.5 nmol (mg chl)−1 is unequivocally confirmed (Table 3). In the 45 s experiment, the atom% 13C excess in carbons 4, 5 and 6 of the octulose is between 25% and 30%, comparable with the values calculated for C-2 and C-3 of ribose, C-3 and C-4 of fructose and glucose, and C-4 and C-5 of sedoheptulose (Table 4). These sugars and their above carbons are implicated in CO2 fixation in Calvin’s photosynthetic pathway (Fig. 1) and it is therefore not unreasonable to speculate that D-g-D-a-octulose also has a role in this pathway, in support of the conclusions drawn in the preceding paper (Williams and MacLeod 2006). The pattern of 13C labelling of carbons 4, 5 and 6 of D-g-D-a-octulose presented here is precisely the same as that shown in the octulose 1,8-bisphosphate by NMR spectroscopy of a spinach leaf extract after photosynthesis in 13CO2 (Bartlett et al. 1989). Table 4Atom% excess 13C in individual carbons and 13C enrichments in C2 moietiesaCarbonXyloseLyxoseXyluloseRiboseFructoseGlucoseSedoheptuloseUnk. HeptuloseOctuloseC-1{24.2,{24.8,–12.1–{19.2,–––C-28.68.5–23.8–7.2–––C-323.7/24.1b16.7–24.5d14.140.1c–7.6–C-4{19.7d,–8.4{12.5d,23.8/24.7b27.523.6/51b,c54.225.0C-50–7.10{10.5,{16d,46.3/34.1b,c,d–28.2cC-600{21.1d–31.2dC-70{15.4,dC-80aA dash implies that the value could not be measured as a one- or two-carbon unitbCalculated from two separate sets of data, e.g. C1-3 minus C1-2 and C3-5 minus C4-5 both give a value for C3 of XylosecNot corrected for contribution of m/z 305 to the m/z 307 clusterdm/z 207 from residual column bleed taken as zero in m/z 205 cluster A further interesting observation was the presence of a second heptulose sugar, albeit in much lower concentrations than that of sedoheptulose (Table 3). It had a substantial atom% excess at C-4 of 54.2%, comparable with that observed for the same carbon in sedoheptulose. A GC/MS comparison of the retention time of its Mox-TMS derivative with that of mannoheptulose eliminated this as a possibility and so it remains unidentified. One of the anomalies of the Calvin Pathway (Fig. 1) has been the failure to identify erythrose phosphate in chloroplasts, even though it is proposed to play an integral role in CO2 fixation. In this study, although we were able to detect a four carbon sugar with the same retention time and mass spectrum as Mox-TMS-erythrose (Table 3), GC/MS/SIM analysis showed no incorporation of 13C in any of its fragment ions. It could therefore be assumed that the level of erythrose phosphate participating in the pathway is below that detectable by GC/MS/SIM. However, the identification and evidence for the presence of Ery 4-P in the Calvin and Pentose pathways has always been weak and contentious (Williams et al. 1980). Unfortunately, circumstances have not allowed us to repeat the above experiments on a scale which could provide sufficient amounts of sugar phosphates to allow more complete data on the time-course of 13C incorporation from 13CO2 during photosynthesis to be obtained. In support of an earlier study of gluconeogenesis in liver cells (Desage et al. 1989) it has, however, been shown here that GC/MS/SIM offers a viable alternative methodology for the study of positional isotope analysis of intermediates formed from stable-isotope labelled precursors in complex biological systems. In summary, the results reported here and in the preceding paper have established the formation and specific predictive labelling of octulose phosphate(s) in spinach chloroplasts by 13CO2 and 14CO2 during photosynthesis. The data do not unequivocally prove that octulose phosphates are direct and exclusive intermediates of a new set of partial reactions for the path of carbon that were inadvertently missed by Calvin and his colleagues. However we suggest that octulose phosphate formation may fulfil a role as mono- and bisphosphates in a shunt (as defined by Newsholme and Leech 1983) that is attached to the Calvin reaction scheme (Fig. 1). Figure 2 is a structural display of the new reaction scheme presented in the accompanying paper. The shunt shows that the octulose phosphate reactions intersect with and may in time and space compartmentalize Seh 1,7-P2 from SBPase and thereby limit its important role in the regulation of flux in the carbon path of PS. Fig. 2An extension of the Calvin pathway by the inclusion of an octulose phosphate shunt (vide infra). Each asterisk (*) represents a carbon atom labelled during one turn of the cycle. Sugar phosphates in red represent those participating in the octulose shunt. For explanation of abbreviations, see preceding paper

Clin_Oral_Investig-4-1-2238781

Is diagnosing exposed dentine a suitable tool for grading erosive loss?

Quantifying tooth wear in general and erosion in particular mostly is made by distinguishing between lesions restricted to enamel and lesions reaching the underlying dentine. Various scores for grading have been used, but in all systems, higher scores are given in cases of exposed dentine, thus, indicating a more severe stage of the condition. Clinical diagnosis of exposed dentine is made by assessing changes in colour or optical properties of the hard tissues. This paper aims to review the literature and discuss critically problems arising form this approach. It appears that classifying the severity of erosion by the area or depth of exposed dentine is difficult and poorly reproducible, and taking into account the variation of enamel thickness, the amount of tissue lost often is not related simply to the area of exposed dentine. There has still been very little longitudinal investigation of the significance of exposed dentine as a prognostic indicator. Further work and discussion is needed to reevaluate the explanative power of current grading procedures. Introduction This review illustrates the problems that clinicians and researchers have experienced in quantifying tooth erosion. Assessing the degree of erosion in an individual may be difficult [32], but is important, as it reflects net exposure to the erosive forces and the opposing forces of protection, for example from saliva. Grading erosion present plays a part in the assessment of the problem, the need for treatment and, potentially, may indicate means for prevention of further tooth wear. If a reliable method for assessing the loss in vivo of tooth substance were available, then it might be possible to determine if erosion was continuing or had been halted, for example by preventive actions. Whilst some assessment of enamel erosion might be possible by assessing the progressive loss of anatomical contour, this is more difficult when erosion has reached the dentine. The diagnosis of exposed dentine is made by assessing changes in colour [43] or changes in the optical properties of the hard tissues, the validity of which, however, has not yet been investigated. This paper will consider the following main points: Does dentine exposure correlate with the severity of erosion?Can dentine exposure be reliably and reproducibly diagnosed?What is the interplay of factors causing tooth wear and how can other forms of tooth wear be distinguished from erosion when the wear has reached dentine?Are the parameters used for grading tooth erosion applicable to both the primary and secondary dentitions?Could symptoms of dentine hypersensitivity assist in the grading of erosive loss? Does dentine exposure correlate with the severity of erosion? For grading, the overwhelming majority of tooth wear and erosion indices use the differentiation between lesions restricted to enamel and lesions reaching the underlying dentine. The rationale for linking severity of erosion to extension into dentine probably stems from experience with dental caries and restorative procedures, but has not yet been critically discussed. It is an interesting question whether the progression of erosion from enamel into dentine has any significance in terms of disease, although it may well have a bearing in terms of treatment and, sometimes, the presence of symptoms of hypersensitivity felt by the patient. One of the most widely used indices for assessing tooth erosion is that of Lussi et al. [30], albeit sometimes modified [6]. This index does indeed attempt to grade the severity of erosion by assessing the degree of dentine exposure in a simple and, presumably reproducible, way as shown in Table 1. As can be seen, grades 2 and 3 record increased severity of erosion on facial surfaces by judging the area of exposed dentine, rather than loss of dentine. On other surfaces, grade 2 requires the involvement of dentine. Table 1Use of the Lussi Index for grading the severity of dentine loss (in bold text)Index ScoreObservationFacial surfaces0No erosion, glazed appearance, absence of developmental ridges possible1Loss of surface enamel, dentine not involved2Erosion into dentine <50% of affected surface3Erosion into dentine >50% of affected surfaceOther surfaces0No erosion, glazed appearance, absence of developmental ridges possible1Slight erosion, rounded cusps, restorations stand proud of enamel, no dentine erosion2Severe erosion, more pronounced than score 1, dentine involved Other indices use thirds of the affected surface with exposed dentine as thresholds [13, 35, 41] or score the severity of lesions using a more detailed grading with respect to depth of the lesion, as enamel loss only as code 1 or 2, loss of enamel with exposure of dentine with the enamel dentine junction (EDJ) visible as code 3, loss of enamel and dentine beyond EDJ as code 4 and loss of enamel and dentine with exposure of pulp as code 5 [36]. Despite variations in the definition of scores, all indices give higher scores when dentine is exposed, thus, indicating a more severe stage of the condition. This suggests that the amount of tissue loss must be higher when dentine is exposed than when enamel coverage is still present. When considering the thickness of enamel, however, it is obvious that this is not always the case. Enamel gradually thickens from the cervical region occlusally. At the coronal third of the crown, the enamel thickness is about 1.5 mm, in the cervical third, about 0.5 mm and becomes increasingly thinner towards the enamel cementum junction [26]. Thus, when the entire smooth surface of a tooth is exposed to acid, the dentine will be first exposed in the cervical region. At the occlusal area, enamel is thickest at the functional cusps, in upper molars. In lower molars, however, the enamel is thinnest at the mesiobuccal cusp, and is progressively thicker in order from the mesiolingual and distolingual to the distobuccal cusp tips [26]. This would explain the finding that lower first molars, particularly the mesio-buccal cusp tip, is the region where exposed dentine occurs most often. Taking into account the gross variation in enamel thickness, it is not surprising that the correlation between exposed dentine and the severity of wear is weak. A comparison of the amount of tissue loss assessed visually and histological findings revealed that the percentage of teeth with exposed dentine was high even in cases with minor substance loss and, in contrast, that enamel can be present in teeth with moderate to severe substance loss. Furthermore, dentine was exposed in all cases of cupping or grooving even if only minor substance loss occurred [17]. Similar results have been found in a study investigating primary teeth [5]. This showed that in 31 teeth with visually diagnosed wear, only three had enamel present histologically, and even in one of ten teeth in which no wear was diagnosed visually, the histological examination revealed exposed dentine. Once dentine is exposed, it is more prone than enamel to abrasion [25, 21] since its microhardness is much lower than that of enamel [33]. When the severity of lesions is defined by the degree of exposure of dentine, the severity score could assume that tooth wear will progress more rapidly in the future. However, this has not been proven in incidence studies. The predictive value of a diagnosis of exposed dentine to future dental erosion or other forms of tooth wear is far from clear, and longitudinal studies are urgently needed to determine the progression, particularly of erosion, at different sites in the dentition. Can dentine exposure be reliably and reproducibly diagnosed? Cupping of the cusp tips on molar teeth is one of the most common presentations of erosion, for example in teenagers (Fig. 1). This has been found in a recent epidemiological study of erosion in Iceland [6] and was also reported in other studies [16, 30, 42]. Fig. 1Cupping of the cusp tips on the molar teeth is an early sign of erosion where extension of erosion into dentine is difficult to determine One major problem in assessing the degree of erosion in cupping is whether or not the dentine has been exposed at the base of the lesion. Because labial and palatal enamel surfaces have a greater surface area, the observer better appreciates their surface anatomy than with a narrow cup-like lesion formed into a tooth cusp tip. Thus, grading the severity of cuppings is more difficult than perhaps it is for labial and palatal surfaces. Ganss et al. [17] demonstrated that when compared with histological examination, only 65% of areas with exposed dentine, 88% of areas with enamel present and 67% of all areas examined had been diagnosed correctly. Clinicians demonstrated relatively poor reliability and reproducibility in their diagnosis of dentine exposure regardless of their professional experience. Al Malik et al. [5] have also shown that while grading erosion from visual and photographic records gave comparable scores, the degree of erosion was underestimated compared with the degree of tooth wear determined histologically. One way to increase the reliability of indices is to keep the grading scheme simple because minor differences in severity are difficult to assess, at least with the naked eye. For example, in a study from van Rijkom et al. [43], two scores for enamel and dentine tissue loss each were used scoring slight enamel wear as 1, deep enamel wear as 2, wear into dentine as 3 and wear into dentine for more than one half of smooth surfaces as 4, and the Kappa values achieved were low. Larsen et al. [28], in contrast, attempted a prevalence study of tooth erosion using seven examiners and only three clear criteria defining grade 1 as erosion limited to the enamel, erosion exposing the dentine less than 50% of the area of the surface as 2 and erosion exposing dentine in more than 50% of the surface as 3. Despite a careful calibration, including a thorough discussion of the criteria and the experience gained during the study, substantial inter-examiner differences in diagnosis were found. On the other hand, a good strength of agreement was achieved with an index proposed by Larsen et al. [27] where five scores for incisal/occlusal surfaces were used. Further investigations specifically aimed at assessing the reliability and reproducibility of scoring the involvement of dentine in the grading of erosion is necessary, should this criterion be included a clinical index of erosion. One future perspective could be the development of instrument-supported approaches. A recent report of optical techniques involving nitrogen-laser-induced fluorescence [42] appeared to offer the possibility of differentiating accurately between erosion into enamel and into dentine. This method may, according to the authors, even be applicable in vivo, although the reported studies were in vitro. What is the interplay of factors causing tooth wear and how can other forms of tooth wear be distinguished from erosion when the wear has reached dentine? Factors leading to tooth wear, including erosive agents, such as dietary and gastric acid interact with protective factors from the diet, oral hygiene procedures and, particularly, saliva. Interestingly, one study of the relationship between erosion, soft drink consumption and reflux disease found the consumption of Coca-Cola® three times per week to be the dominant factor in the development of molar erosion [24], but the interplay of factors is often complex. Although the net outcome of erosive and reparative factors does not necessarily have to be tooth wear, it appears that once tooth wear has started, the interaction of these factors becomes ever more complex, not least by the time tooth wear has reached the dentine. Thus, the interplay of physical and chemical factors and the relevance of this to dentine wear will be considered together. Relevant physical forces are attrition and abrasion either from mastication or from oral hygiene procedures. Frequent exposure to dietary or gastric acid will lead to a softening of the dental hard tissues, making it easier for forces of attrition and abrasion to contribute significantly to the overall wear on the teeth. Whilst the effects of acids on enamel are relatively simple, leading to the dissolution of mineral and surface softening, the results on dentine are more complex. Chronic exposure to acids not only leads to an increasing loss of mineral but also to a progressive exposure of the organic dentine matrix, the effects of which, however, are not well known. The most investigated interplay between erosion and physical forces is the abrasive effect of toothbrushing. An increased wear of eroded enamel and dentine was demonstrated in a number of studies [7–9, 14, 21, 38, 45], even if one study did not prove this effect for dentine [18]. Azzopardi et al. [11] have also shown that a combination of erosion and abrasion causes more tooth wear than either tooth wear force on its own. Results of epidemiological studies are ambiguous in demonstrating only weak [2, 29] or no [3, 37, 39] association between the occurrence of erosion and oral hygiene habits. With respect to abrasion from mastication, the occlusal surfaces are at particular risk. When dentine is exposed, abrasive foods and oral hygiene products tend to hollow out softer surface regions [31]. This can take place when an abrasive diet is consumed or in cases of acid-weakened dental tissues where a less abrasive bolus could also have similar effects. Clinically, the shape of occlusal lesions can be remarkably similar, making the diagnosis of the predominant aetiological factor difficult to determine. Perhaps commonest among these non-erosive factors contributing to tooth wear is attrition [12]. There is little information about the relative prevalence of facets in relation to different aetiological factors for wear and their progression when dentine is reached. It appears, however, that attrition tends to be superimposed on abrasion in cases of abrasive diet, as facets seldom occur in ancient remains and are not more prevalent in subjects with acidic diet when compared to subjects with an average western diet [15]. The problems of measuring tooth erosion in an individual presenting with tooth wear as shown in Figs. 2, 3 and 4 are considerable, as not only is it difficult to envisage how much dentine and enamel has been lost but it is also increasingly difficult to attribute the loss of tissue to one particular aetiological factor or another as the tooth wear progresses. Even in clear cases of acid erosion, there may be a combination of intrinsic and extrinsic acid contributing to the clinical appearance. Fig. 2Clinical case to illustrate the difficulties of attributing diagnosis of tooth wear. This example of severe tooth wear was thought to be attrition, superimposed on erosion, of the occlusal surfaces of posterior teeth and palatal surfaces of maxillary teethFig. 3Erosion of the palatal cusps of maxillary molar and premolar teeth in a patient with gastro-oesophageal reflux disease. Note that dentine is visible on the mesio-palatal cusp of the first molar toothFig. 4Typical appearance of erosion in a patient with gastro-oesophageal reflux disease showing palatal erosion of the maxillary tooth cusps and buccal erosion of the mandibular tooth cusps. The degree of tooth wear and the presence of restorations make grading difficult Are the parameters used for grading tooth erosion applicable to both the primary and secondary dentitions? One particularly difficult problem is assessing tooth wear in the deciduous dentition where the tooth structure withstands the forces of tooth wear somewhat less than the permanent dentition. An example is shown in Fig. 5 of a case where the aetiology of considerable erosion was never clear even when it continued into the permanent dentition. Indices used for the permanent dentition have also been used in the primary dentition [3, 35], but also, a special index for children has also been developed [36]. Fig. 5Clinical appearance of severe erosion, probably combined with other forms of tooth wear, in the deciduous dentition of a patient in whom no pathological reflux disease was recorded. The reported diet was relatively normal but erosion later appeared in the first permanent molar teeth The prevalence of erosive wear in the deciduous dentition varies between 6 and 100%, and the percentage of children with erosion into dentine varies between 1 and 50% [4, 10, 20, 23, 46]. The high prevalence of erosive wear in the deciduous dentition is remarkable when the short period in function is considered, but could be explained by the softer nature of the primary tooth substance leading to increased wear from physical impacts. The susceptibility to acids, in contrast, is probably not greater than for primary than for permanent dental tissues [22]. The enamel of deciduous teeth, however, is much thinner than that of permanent teeth [19], which will lead to the rapid exposure of dentine. These factors will tend to make scores greater for deciduous teeth, and the links with known aetiological factors seem less clear. Studies using the same criteria for the primary and permanent dentition revealed higher prevalence data for the primary than for the permanent dentition when lesions involving dentine are regarded [3, 35]. It is unclear if these data really reflect more severe tissue loss occurring in children or if the thin deciduous enamel leads to early exposure of dentine also in cases of minor wear, which would mean an overestimation of the condition. Perhaps the more relevant question is whether tooth erosion in the deciduous dentition is a prognostic indicator of likely erosion in the permanent dentition, as was the case in the subject illustrated in Fig. 5. Meurman and ten Cate [34] have pointed out that there were no longitudinal clinical studies of progression of erosion in the deciduous teeth nor does the continuing of erosion into the permanent dentition appear to have received wide attention. In the last decade, only one study has demonstrated an increased risk of erosion in the permanent dentition when erosion had been present in the primary dentition. As study models were used, the relevance of exposed dentine was, however, not assessed [16]. Identifying the cause of erosion in the deciduous dentition may enable preventive measures, or treatment in the case of reflux disease, that could eliminate the risk of erosion in the permanent dentition. The UK National Guidelines in Paediatric Dentistry [40] notes the importance of recording the probable aetiological factors but notes specifically the unsuitability of the Smith and Knight Index [41] for measuring small longitudinal changes in tooth erosion and the problems of the complex interaction of causative and protective factors that make tooth erosion a more complex diagnostic and prognostic problem than would perhaps be thought initially to be the case. Could symptoms of dentine hypersensitivity assist in the grading of erosive loss? Once tooth wear has extended into dentine, there is a possibility that the patient will experience symptoms of hypersensitivity [44]. There are numerous causes of dentine hypersensitivity and, therefore, great care is needed in diagnosing hypersensitivity as a result of tooth erosion reaching dentine. The mechanisms underlying hypersensitivity [44] strongly suggest that exposure of dentine tubules to the mouth will cause hypersensitivity in some, but not all, subjects. Consequently, the presence of symptoms of dentine hypersensitivity in a patient with signs of tooth erosion is strongly suggestive of extension of the erosion into dentine. While acidic drinks may expose the dentine tubules and remove early plaque formed on the dentine, thus leading to the symptoms of hypersensitivity, other factors such as toothbrushing may also contribute to this [1]. In conclusion, distinctions are commonly made between erosion into enamel and erosion into dentine. Methods of scoring are derived from parameters associated more with dental caries and restorative procedures for that disease rather than specifically for tooth wear. The interplay of tooth wear factors is complex and requires further research. Classifying the severity of erosion by the area or depth of exposed dentine is difficult and poorly reproducible and, particularly with respect to the variation of enamel thickness, the amount of tissue lost often is not related to dentine exposure. There has still been very little longitudinal investigation of the significance of exposed dentine as a prognostic indicator. Further work is needed to reevaluate the explanative power of current grading procedures.

J_Abnorm_Child_Psychol-4-1-2268722

Hot and Cool Forms of Inhibitory Control and Externalizing Behavior in Children of Mothers who Smoked during Pregnancy: An Exploratory Study

This study examined whether children exposed to prenatal smoking show deficits in “hot” and/or “cool” executive functioning (EF). Hot EF is involved in regulation of affect and motivation, whereas cool EF is involved in handling abstract, decontextualized problems. Forty 7 to 9-year-old children (15 exposed to prenatal smoking, 25 non-exposed) performed two computerized tasks. The Sustained Attention Dots (SA-Dots) Task (as a measure of “cool” inhibitory control) requires 400 non-dominant hand and 200 dominant hand responses. Inhibitory control of the prepotent response is required for dominant hand responses. The Delay Frustration Task (DeFT) (as a measure of “hot” inhibitory control) consists of 55 simple maths exercises. On a number of trials delays are introduced before the next question appears on the screen. The extent of response-button pressing during delays indicates frustration-induced inhibitory control. Prenatally exposed children showed poorer inhibitory control in the DeFT than non-exposed children. A dose–response relationship was also observed. In addition, prenatally exposed children had significantly higher (dose-dependent) conduct problem- and hyperactivity-inattention scores. There were no significant group differences in inhibitory control scores from the SA-Dots. These results indicate that children exposed to prenatal smoking are at higher risk of hot but not cool executive function deficits. Introduction Prenatal tobacco exposure has consistently been associated with children’s externalizing behavior (e.g. Button et al. 2005; Huijbregts et al. 2007, in press; Kotimaa et al. 2003; Mick et al. 2002; Wakschlag et al. 2006). In contrast, studies into associations between prenatal tobacco exposure and children’s cognitive functioning show very mixed results (e.g. Batstra et al. 2003; Fried et al. 1998, 2003 versus Breslau et al. 2005; Huijbregts et al. 2006; Trasti et al. 1999). This is surprising because externalizing behavior problems often co-occur with cognitive difficulties. Cognition in ADHD, ODD/CD, and ADHD+ODD/CD Cognition has been widely studied in ADHD (for reviews, see Barkley 1997; Castellanos et al. 2006; Sergeant et al. 2003; Wilcutt et al. 2005). This research has shown, among others, problems with inhibitory control, working memory, vigilance, reward sensitivity (delay aversion), and time processing and currently focuses on different cognitive endophenotypes of ADHD (i.e. biologically based phenotypes that carry genetic loadings and index liability to disease; see Castellanos and Tannock 2002). Oppositional Defiant Disorder (ODD), Conduct Disorder (CD) and high levels of physical aggression have also been associated with cognitive difficulties such as language-based verbal skills and executive function (EF) (Henry and Moffitt 1997; Hill 2002; Séguin et al. 1999). Controversy still exists about whether verbal and, particularly, EF deficits are present in ODD/CD without comorbid ADHD (Pennington and Ozonoff 1996). Since Pennington and Ozonoff’s review on EF in different forms of developmental psychopathology, an increasing number of studies have taken into account the high comorbidity of ODD/CD and ADHD. This has resulted in reports of both qualitative and quantitative differences in EF-dysfunction between disorders. For example, Van Goozen and colleagues (2004) reported the absence of problems with Working Memory (WM) and planning in a group of children with Oppositional Defiant Disorder (ODD) and a group with combined ODD/ADHD compared to healthy controls. Both are executive functions that have frequently been reported to be deficient in ADHD. Oosterlaan et al. (2005) compared ADHD, ODD/CD, and ADHD+ODD/CD groups and found that WM and planning deficits were specific to ADHD (and explained by ADHD in the combined disorder). Distinguishing ADHD and ODD/CD is more difficult when it concerns inhibitory control (Oosterlaan et al. 1998). Inhibition has been defined and operationalized in a number of different ways, e.g. as stopping an ongoing response, interference control and inhibition of prepotent responses (Barkley 1997), and studies have generally not included tasks measuring such different forms of inhibition. Furthermore, inhibition has been listed as one of the executive functions (e.g. Pennington and Ozonoff 1996), but has also been described as an important cognitive ability underlying many different executive functions (Zelazo et al. 1997). Lack of inhibitory control has been proposed as the central deficit in both ADHD (Barkley 1997) and ODD/CD/physical aggression (e.g. Lau et al. 1995), but is not necessarily similarly dysfunctional in both types of disorders. Nigg (2003) has suggested that ADHD is predominantly associated with dysfunctional executive inhibition, whereas conduct problems are predominantly associated with dysfunctional motivational inhibition. Nonetheless, a ‘motivational’ pathway (or endophenotype) to cognitive deficits (e.g. altered reward sensitivity) in ADHD has also been identified (Solanto et al. 2001; Sonuga-Barke 2002). ODD/CD and ADHD may be distinguished more easily on other aspects of emotional or social information processing. As a starting point it may therefore be good to use a theoretical framework that, in addition to motivation, accommodates more different aspects of emotional or social information processing. Such a model has been proposed by Zelazo and Müller (2002). Hot and Cool Aspects of Executive Functioning Zelazo and Müller (2002) made a distinction between the ‘cool’ aspects of EF more associated with dorsolateral regions of the prefrontal cortex (PFC) and the ‘hot’ aspects more associated with ventral and medial regions. Cool EF is elicited by abstract, decontextualized problems, whereas hot EF is elicited by problems that involve the regulation of affect and motivation (i.e., the regulation of basic limbic system functions). Neuro-anatomical evidence for a distinction between hot and cool executive functions stems from research into risky (‘hot’) decision making versus ambiguous (‘cool’) decision making (Krain et al. 2006). The authors performed a meta-analysis of studies measuring brain activity during risky decision making (involving low probabilities of high rewards and high probabilities of low rewards) and/or ambiguous decision making (where the probability of a specific outcome is unknown or close to chance and choices do not differ in reward value), and found that there were significant differences between risky and ambiguous decision-making in regions of the orbitofrontal cortex (OFC: more activity in risky than in ambiguous decision-making), dorsolateral prefrontal cortex (DLPFC: more activity in ambiguous than in risky decision making) and anterior cingulate cortex (ACC: more activity in ambiguous decision-making than in risky decision-making in caudal regions, and more activity in risky than in ambiguous decision-making in rostral regions). Zelazo and Müller use a problem-solving framework to discuss EF. Each stage of problem-solving (problem representation, planning, execution, and evaluation) is subserved by EF, which can take both hot and cool forms. They provide examples of each for every problem solving stage. For example, for the problem representation stage cool EF could be measured by asking participants to re-represent test items in different ways (but not act upon the re-representation). Hot measures of problem representation would include most tests of Theory of Mind, where participants are required to represent something (e.g. feelings, intentions) from multiple points of view. For the planning stage, tower tasks that require children to describe their plans without actually executing them are put forward as an example of ‘cool’ EF, whilst strategic deception, where participants are shown two pre-formulated plans, one deceptive and one not, and are asked to select the one that would trick the other person, was offered as an example of ‘hot’ EF. In the execution stage, card sorting (i.e. select according to two different rules/principles) was given as an example of cool EF, and delay of gratification as an example of hot EF (e.g. choosing between a small reward immediately or a larger reward later (prudence) or between a reward for self now or a reward for self and other later (altruism)). Finally, in the evaluation phase error detection and error correction in the context of monitoring progress toward a goal are provided as examples of cool EF and error detection and correction in the context of extinction as examples of hot EF (i.e. after a certain amount of responding, it can be shown that the reward supply has been exhausted, a motivational input that might lead to response extinction). In the problem-solving framework inhibitory control is considered to underlie EF in all information processing stages and consequently has its hot and cool variants as well. Thus, when inhibitory control involves emotion, affect, and/or motivation, it may be classified as a hot EF, whereas inhibitory control could be classified as cool EF when it is required in an “abstract” or “neutral” context. Neurobiology of Prenatal Tobacco Exposure and Externalizing Behavior There is increasing evidence that maternal prenatal smoking is more strongly associated with ODD/CD and with the co-occurrence of ODD/CD and ADHD than with ADHD alone (Huijbregts et al. 2007; Nigg and Breslau 2007; Wakschlag et al. 2006). This evidence alone might be enough to suggest that children prenatally exposed to tobacco might specifically show cognitive dysfunction when emotion is involved, specifically when tasks involve inhibition. Further indirect support for such a hypothesis comes from neurobiological evidence showing similar or closely related effects on the central and peripheral nervous system associated with prenatal smoking and with antisocial/disruptive behavior. Antisocial and disruptive behaviors have convincingly been associated with changes in the functioning of the autonomic nervous system (ANS) and the hypothalamic–pituitary–adrenal (HPA) axis. Both have very important roles in stress regulation. ANS-functioning is generally measured by (changes in) heart rate (HR), which reflects sympathetic and parasympathetic nervous system activity, and (changes in) skin conductance (SC), which reflects only sympathetic nervous system activity. HPA-functioning is generally measured by (changes in) the adrenal stress hormone cortisol. Resting HR, basal SC and basal cortisol levels were all shown to be lower in children with disruptive behavior disorders. In addition, reduced SC reactivity and blunted or absent cortisol responses to stress have consistently been reported (for reviews, see Ortiz and Raine 2004; Van Goozen et al. 2007). These findings have lead to the fearlessness theory of antisocial behavior (Raine 1993), which claims that low levels of arousal are markers of low levels of fear and the stimulation-seeking theory (Zuckerman 1979), which claims that low arousal represents an aversive physiological state and that individuals with tonically low arousal seek out stimulation to raise their arousal to optimal or normal level. Antisocial/disruptive behavior has also been associated with neurotransmitter abnormalities, for example with impairments in the serotonergic (5-HT) system. The 5-HT system interacts with the HPA-system at several levels. For example, activation of (postsynaptic) 5-HT receptors in the hypothalamus stimulates HPA axis activity and causes elevation in cortisol, while a disruption of 5-HT neurotransmission by tryptophan depletion lowers HPA axis functioning (blunts cortisol-elevating effect of stressor). Glucocorticoids also modulate 5-HT receptor binding/sensitivity (see Van Goozen et al. 2007). Importantly, the ANS/HPA abnormalities in children with conduct problems have not been replicated in children with ADHD only (Herpertz et al. 2001; Snoek et al. 2004; Waschbusch 2002). Animal studies investigating the consequences of prenatal nicotine exposure have shown an upregulation of nicotinic cholinergic receptor binding sites as well as persistent deficits in cerebrocortical choline acetyltransferase activity and hemicholium-3 binding to the presynaptic choline transporter, which is indicative of cholinergic hypoactivity (Seidler et al. 1992; Slotkin et al. 2006). These studies also show changes in noradrenergic responsiveness following prenatal nicotine exposure. The most important neurotransmitters of the ANS are acetylocholine (parasympathetic) and norepinephrine (sympathetic). In line with the chronic underarousal theories of antisocial/disruptive behavior, it may be hypothesized that, without continued nicotine exposure, the increased number of cholinergic receptor binding sites will be understimulated, possibly resulting in stimulation-seeking and fearlessness. Furthermore, abnormalities in serotonergic neuromodulation following prenatal nicotine exposure have been indicated (suppression of the 5HT1A receptor subtype and upregulation of 5HT2 receptor subtype) (Seidler et al. 1992; Slotkin et al. 2006). We had already stressed the interactions between 5HT and HPA-functioning. The commonalities in the neurobiology of antisocial/disruptive behavior and the neurobiology of prenatal smoking are striking. Moreover, factors affecting the same neurobiological system are more likely to reinforce each other’s impact (Moffitt et al. 2005). A number of studies have already reported abnormalities in ANS/HPA-functioning and levels of arousal in children of mothers who smoked during pregnancy (Browne et al. 2000; Horne et al. 2004; McDonald et al. 2006; Ramsay et al. 1996). The Present Study Taken together, the strong associations between prenatal smoking and offspring conduct problems, the specific affect-mediated (‘hot’) cognitive deficits supposedly associated with conduct problems, and the common neurobiology of prenatal nicotine exposure and conduct problems that particularly affects behavior and cognition under stressful conditions (i.e. behavior and cognition that is affect-mediated or ‘hot’), suggest that it would be useful to investigate aspects of hot cognition in children exposed to prenatal smoking (see also Raine 2002). Studies into associations between prenatal tobacco exposure and cognition have generally focused on academic achievement or rather broad cognitive constructs such as intelligence, verbal ability and executive functions (e.g. Batstra et al. 2003; Breslau et al. 2005; Huijbregts et al. 2006; Trasti et al. 1999), and have not yet used a distinction between hot and cool aspects of cognition. There are many possible causes for the inconsistent results which will be discussed later, but in the present study we will contrast hot and cool aspects of cognition, more specifically hot and cool forms of inhibitory control because a lack thereof has been hypothesized to be central to the behavior problems displayed by children of mothers who smoked during pregnancy. It is expected that children of mothers who smoked will particularly display problems with hot inhibitory control. Since a number of shared neural structures and neuromodulatory mechanisms are involved in cool and hot EF it cannot be ruled out that children prenatally exposed to nicotine will also have problems with cool inhibitory control but it may be hypothesized that these, if present, will be of a less serious nature. Method Participants Forty children (16 boys, 24 girls; all of Caucasian ethnicity) attending two primary schools in Southampton (UK) participated in the study: 25 children from mothers who had not smoked during pregnancy (9 boys, 16 girls), 15 from mothers who had smoked during pregnancy (7 boys, 8 girls). Of these mothers, 6 smoked 1–9 cigarettes/day and 9 smoked ≥ 10 cigarettes/day (prenatal smoking was established through self-report by the mothers; although there is a risk for a social desirability bias, several studies have indicated a relatively strong association between retrospective self-report and blood/urine cotinine-levels (i.e. the main nicotine metabolite), e.g. Law et al. 2003; Pickett et al. 2005). Mean age of the participants was 7.98 (range 7.0–9.0, SD = 0.68). Informed consent was obtained from all children and their parents/caretakers and from the school boards. The study was approved by the University of Southampton—School of Psychology Ethics Committee. Tasks and Measures Sustained Attention Dots Task The Sustained Attention Dots task (SA-Dots, De Sonneville 1999) is a computerized task consisting of 600 presentations of 3-, 4-, and 5-dot patterns (random configurations). The 3, 4, or 5 dots are each presented 200 times. Participants use both hands for responding. They must press the mouse key of the non-dominant hand (the ‘no-key’) when 3 or 5 dots appear on the screen and the mouse key of the dominant hand (the ‘yes-key’) when 4 dots appear. A response has to be given between 250 and 6,000 ms after a signal. The fixed post-response interval is 250 ms. The program provides auditory feedback after an error. Since a response with the non-dominant hand is required twice as often as a response with the dominant hand, a response with the non-dominant hand becomes the prepotent response. Thus, inhibitory control is required when four dots appear. The less participants are capable of inhibiting biased response tendencies, the more misses (i.e. pressing the ‘no-key’ when four dots appear) they will produce relative to false alarms (i.e. pressing the ‘yes-key’ when three or five dots appear). The Bias score is the difference between number of misses and the mean of the number of false alarms when three dots appeared and when five dots appeared (see Huijbregts et al. 2002a). A second measure used to assess the quality of inhibitory control is post-error slowing. Post-error slowing is usually considered a strategic adjustment based on performance monitoring (Klein et al. 2007; Wiersema et al. 2007). In order to be able to slow down after an error inhibitory control is required. In order to control for general response speed, the difference between mean RT of correct responses after errors and overall mean RT of correct responses was selected as measure for post-error slowing (RTae). Delay Frustration Task The Delay Frustration Task (DeFT, Bitsakou et al. 2006) is a task where participants are presented with a series of simple maths questions (only additions in the children’s version) on a computer screen. Four possible solutions are displayed on the screen together with each question. Participants are asked to select the correct answer by pressing one of four buttons on a response box. The next question is presented as soon as a response has been recorded. However, on 16 out of 55 trials access to the next trial is delayed. On 8 out of the 16 trails in the post-response delay condition, a short delay (2–10 s) occurs, whilst on the remaining eight trials a delay of 20 s occurs. During the 20 s delay periods, the number (NP) and duration (TP) of responses made on a response key are recorded. These measures and their product (NP*TP) index delay frustration. In the instruction, participants are told that the computer shows signs of malfunctioning and that if the computer appears to not register their response, they should just wait until the next trial appears and do not have to respond again. It is considered to be indicative of inhibitory control if participants refrain from constantly pressing response box buttons during the delay periods. Practice sessions preceded both the SA-dots- and the DeFT test sessions, which were held in separate quiet rooms in both schools. Total test time per child was approximately 30 min (approximately 15 min for each task). Strengths and Difficulties Questionnaire (SDQ) The parent–teacher rated SDQ (Goodman 1997) is a short behavioral screening questionnaire of children aged 4 to 16 years, which can be completed within approximately 5 min. For the present study, teacher ratings were obtained, which correlate highly with parent ratings (Goodman 1997). The SDQ consists of 25 items comprising five different subscales (five items each): emotional problems, conduct problems, hyperactivity/inattention, peer relationship problems, and prosocial behavior. The validity of this five-factor structure has been confirmed in large population samples (e.g. Goodman 2001). Each item is scored on a three-point scale: ‘not true’, ‘somewhat true’, or ‘certainly true’. ‘Somewhat true’ is always scored as 1, with the scoring of ‘not true’ and ‘certainly true’ varying with each item, being equal to either 0 or 2. A higher score indicates greater problems except for prosocial behavior, where a higher score indicates more positive behavior. A total difficulties score (range 0–40) is obtained by summing the scores of the emotional problems-, conduct problems-, hyperactivity/inattention-, and peer relationship problems subscales. Of specific interest for the present study are the Conduct Problems (CP) scale, which has the following items: (1) often has temper tantrums or hot tempers; (2) generally obedient, usually does what adults request; (3) often fights with other children or bullies them; (4) often lies or cheats; and (5) steals from home, school, or elsewhere, and the Hyperactivity/Inattention (HI) scale, containing the following items: (1) restless, overactive, cannot stay still for long; (2) constantly fidgeting or squirming; (3) easily distracted, concentration wanders; (4) thinks things out before acting; and (5) sees tasks through to the end, good attention span. Examples of items from the other scales are: “Picked on or bullied by other children” (peer relationship problems); “Many worries, often seems worried” (emotional problems); and “Helpful if someone is hurt, upset or feeling ill” (prosocial behavior). The SDQ has good reliability, whether it concerns internal consistency of the items (mean Cronbach’s α = 0.73), cross-informant correlations or test–retest reliability, and good criterion validity, whether it concerns its predictiveness of diagnosed psychiatric disorders, (mental health) service contact/use, or the association of its scores with those on other behavioral screening questionnaires (Bourdon et al. 2005; Goodman 2001; Goodman and Scott 1999). Statistical Analysis Main analyses Because the number of children participating in this study was not large and the scores on the SA-Dots Task, the Delay Frustration Task, and the behavior scales of the Strengths and Difficulties Questionnaire were not normally distributed, we used nonparametric tests. Spearman rank order correlations were used to test whether scores on the Delay Frustration Task were related to scores on the SA-Dots Task. Mann–Whitney U tests were employed to compare children of non-smokers and children of mothers who smoked during pregnancy on cognitive and behavioral (SDQ) measures. In order to test for dose-dependency children of smoking mothers were further subdivided into those whose mothers smoked between one and nine cigarettes/day and those whose mothers smoked ≥10 cigarettes/day. These two groups were compared to each other and to children of mothers who did not smoke using Kruskal–Wallis H-tests. Exploratory analyses Our final analyses were hierarchical linear regression analyses predicting those variables for which significant group differences were observed in Mann–Whitney U tests from prenatal smoking and the sum of SDQ Hyperactivity-Inattention (HI) and Conduct Problems (CP) scores. Maternal prenatal smoking was introduced in a first block followed by CP+HI in a second block. These analyses should be considered exploratory because the sample is too small to reliably distinguish children with HI only or CP only. In order to determine whether HI or CP is more strongly associated with task performance than a combination of both of them, it would be necessary to have a HI only and CP only group as well (see Huijbregts et al. 2007; in press). Because the regression analyses are parametric tests, dependent variable scores were square root transformed in order to approach normal distributions. These exploratory analyses were performed because the relationship between prenatal smoking, externalizing behavior, and task performance can only be fully explored when all three variables are present in one analysis. Results Main analyses Table 1 shows means (+SDs) of the smoking and non-smoking groups for the subscales of the SDQ and for the outcome measures of the SA-Dots and the DeFT. Scores on the SA-Dots Task and the Delay Frustration Task did not correlate significantly: Bias × NP: r = 0.27, p = 0.143; Bias × TP: r = 0.27, p = 0.139; Bias × NP*TP: r = 0.26, p = 0.150; RTae × NP: r = 0.16, p = 0.355; RTae × TP: r = 0.13, p = 0.453; RTae × NP*TP: r = 0.20, p = 0.267. Despite the fact that scores on all different subscales of the SDQ were related to each other, only the scores on the hyperactivity-inattention and conduct problems scales could be predicted by maternal prenatal smoking, as shown by the outcomes of a series of Mann–Whitney tests. Thus, prenatal smoking did not predict the score on the emotional problems scale: Mann–Whitney U = 183.5, z = −0.46, p = 0.647, the peer relationships problem scale: Mann–Whitney U = 185.0, z = −0.44, p = 0.661, or the prosocial behavior scale: Mann–Whitney U = 194, z = −0.17 p = 0.865. In contrast, children of mothers who had smoked during pregnancy had more conduct problems (Mann–Whitney U = 134.5, z = −2.1, p = 0.038) and higher hyperactivity/inattention (Mann–Whitney U = 124.5, z = −2.0, p = 0.041) than children of mothers who had not smoked (mean ranks for conduct problems: 25.7 for exposed children and 18.0 for non-exposed children; for hyperactivity-inattention: 25.1 for exposed children and 18.4 for non-exposed children) (see also Fig. 1 for standardized CP and HI-scores for non-exposed versus exposed children). Dose–response relationships were observed when behavior of non-exposed, moderately exposed and heavily exposed offspring was compared with Kruskal–Wallis tests (conduct problems: mean ranks: 17.8, 22.4, and 26.8; χ2(2) = 5.7, p = 0.058; hyperactivity/inattention: mean ranks 17.8, 19.1, and 29.1; χ2(2) = 6.5, p = 0.040) (Fig. 2). Table 1Mean scores (+SDs) for children with and without prenatal tobacco exposure on the strengths and difficulties questionnaire, the SA-dots task and the delay frustration taskGroup Number (N)SDQSA-DotsDelay frustration taskCPHIRTaeBiasNPTPNP*TPOverall400.68 (1.2)2.8 (2.5)477.2 (507.1)9.0 (8.9)0.40 (.44)162.0 (145.5)123.2 (185.9)Not-Exposed250.48 (1.2)2.2 (2.2)473.6 (594.1)9.0 (10.3)0.29 (.40)116.9 (123.3)78.0 (139.5)Exposed151.0 (1.2)3.9 (2.6)483.6 (323.1)9.1 (6.5)0.57 (.46)234.1 (153.0)195.5 (229.6)SDQ: Strengths and Difficulties Questionnaire, CP: Conduct Problems, HI: Hyperactivity-Inattention, SA-Dots: Sustained Attention-Dots; Bias and RTae are the inhibitory control measures from the SA-Dots task: RTae: RT correct after error – RT correct general; Bias: Number of errors on 4-dot presentations – mean number of errors on 3 and 5 dot presentations. NP, TP and NP*TP are the inhibitory control measures from the Delay Frustration Task: NP: mean number of response button presses during 20-s delays. TP: time response buttons were pressed during 20-s delays.Fig. 1Mean standardized scores on the SA-dots task, the Delay Frustration Task, and the Strengths and Difficulties Questionnaire for non-exposed vs exposed offspring. Bias and RTae are the inhibitory control measures from the SA-Dots task. DF-np (mean number of presses during 20 s intervals), DF-tp (mean duration of presses during 20 s intervals), and DF-np*tp (product of DF-np and DF-tp) are the inhibitory control measures from the Delay Frustration Task. SDQ-CP: Strengths and Difficulties Questionnaire—Conduct Problems, SDQ-HI: Strengths and Difficulties Questionnaire—Hyperactivity/InattentionFig. 2Dose-dependency in associations between maternal prenatal smoking and scores on the Delay Frustration Task and the Strengths and Difficulties Questionnaire. DF-np (mean number of presses during 20 s intervals), DF-tp (duration of presses during 20 s intervals), and DF-np*tp (product of DF-NP and DF-TP) are the inhibitory control measures from the delay frustration task. SDQ-CP: Strengths and Difficulties Questionnaire—Conduct Problems, SDQ-HI: Strengths and Difficulties Questionnaire—Hyperactivity/Inattention With respect to task performance, children of mothers who had smoked during pregnancy (mean rank = 25.5) showed a higher response frequency (number of button presses: NP) during intervals in the Delay Frustration Task than children of non-smokers (mean rank = 16.5) [Mann–Whitney U = 97, z = −2.4, p = 0.016]. They also showed a longer response duration during intervals (time of button pressing: TP) (mean rank = 26.4) than children of non-smokers (mean rank = 16.0) [Mann–Whitney U = 84, z = −2.8, p = 0.005] (Table 1; Fig. 1). A dose response relation was shown by Kruskal–Wallis tests: NP: 0 cigs/day: mean rank = 16.5; 1–9 cigs/day: mean rank = 23.0; ≥10 cigs/day: mean rank = 27.2; χ2(2) = 6.2, p = 0.044; TP: 0 cigs/day: mean rank = 16.0; 1–9 cigs/day: mean rank = 23.2; =10 cigs/day: mean rank = 28.6, χ2(2) = 8.5, p = 0.014 (Fig. 2). For frequency and duration combined (NP*TP), a Mann–Whitney test also showed that children of smokers (mean rank = 26.0) showed less restraint during delays than children of non–smokers (mean rank = 16.3): U = 90, z = −2.6, p = 0.009. A dose–response relationship, indicating less inhibitory control during delays with an increase in prenatal smoking, was also observed (mean ranks for children of non-smokers, moderate smokers, and heavy smokers were 16.3, 22.8, and 28.1, respectively): χ2(2) = 7.5, p = 0.023 (Fig. 2). There were no significant differences between exposed and non-exposed offspring in inhibitory control scores from the Sustained Attention Task: Bias: U = 130.5, z = −1.1, p = 0.255; RTae: z = −0.03, p = 0.988; Dose response: Bias: χ2(2) = 3.1, p = 0.209; RTae: χ2(2) = 1.4, p = 0.502. Exploratory analyses An important question that remains is whether the levels of CP and HI explain associations of prenatal smoking with DeFT-performance. As expected HI and CP were significantly related (Spearman’s rho = 0.57, p < 0.001). Because not enough children with CP only or HI only could be distinguished in our sample, the sum of the SDQ CP- and HI-scores was used in subsequent regression analyses. Hierarchical linear regression analyses introducing prenatal smoking in a first block and CP+HI in a second block to predict DeFT-scores NP, TP and NP*TP showed significant regression models for each of the DVs: NP: F(2,32) = 3.9, p = 0.029; TP: F(2,32) = 7.5, p = 0.002; and NP*TP: F(2,32) = 3.8, p = 0.039. In none of these models prenatal smoking remained a significant predictor when CP+HI was introduced: NP: β = 0.268 (p = 0.122), a reduction from β = 0.361 (p = 0.033) in the model without CP+HI; TP: β = 0.294 (p = 0.067), a reduction from β = 0.427 (p = 0.011); and NP*TP: β = 0.264, a reduction from β = 0.355 (p = 0.036). In models with maternal prenatal smoking, CP+HI most strongly predicted time/duration of button pressing (TP): β = 0.393 (p = 0.016). Discussion Children of mothers who smoked during pregnancy exhibited problems with ‘hot’ inhibitory control, but not with ‘cool’ inhibitory control. Furthermore, children of mothers who smoked during pregnancy showed more conduct problems and hyperactivity-inattention than non-exposed children. The reliability of these effects was supported by the fact that dose–response relationships were also observed. The finding that problems particularly appear when cognitive tasks involve the regulation of affect is in line with the well-replicated association between maternal prenatal smoking and behavior regulation problems. In addition, CP+HI could account for the effects of prenatal smoking on DeFT-performance. These findings further support the contention that children of mothers who smoked during pregnancy specifically have problems when cognitive task performance involves regulation of affect. Our results indicate the importance of employing a theoretical framework supported by neuro-anatomical and neurobiological data when studying associations between maternal prenatal smoking and children’s cognitive and behavioral development. Although the pattern of results confirmed expectations, it was somewhat surprising that there were no group differences at all for the cool inhibitory control measures, nor were there significant correlations between SA-Dots inhibitory control scores and DeFT inhibitory control scores. Some associations were expected because hot and cool types of EF share common neural structures and neuromodulatory mechanisms, i.e. they are presumed to be part of an ‘interactive functional system’ (cf. Hongwanishkul et al. 2005), which is in line with the theory stating that cortical processes involving perception and cognition provide representations of the physical and social world to the sub-cortical motivational systems, whilst, in turn, these representations are influenced by appetitive and defensive needs (Derryberry and Rothbart 1997). We did not find evidence supporting such an overlap between hot and cool executive functions, but we cannot rule out that it would become apparent when other measures of cool inhibitory control are used. Because cognition in children of mothers who smoked during pregnancy has always been studied without a clear theoretical framework, the employment of Zelazo and Müller’s model is only a first step towards elucidating the cognitive profile of this population. Further explorations are warranted based on other cognitive models for which clear neuroanatomical evidence exists. It is possible that abnormalities in neuromodulation or neuroanatomical structures supporting hot EF only affect the most demanding cool EF-functions. The SA-Dots task in the present study had clear inhibitory control demands, but only minimal working memory or flexibility demands, which, particularly in combination with inhibitory control, might represent the most demanding forms of cool EF (e.g. Davidson et al. 2006; Huijbregts et al. 2002b). Although the present study was the first to use the hot versus cool distinction for examining EF in children whose mothers smoked during pregnancy, studies into ADHD and studies into differences in EF between children with ODD/CD and ADHD have employed the same or similar distinctions earlier. In ADHD research significant empirical support has been found for a distinction between delay aversion (which can be considered a form of hot EF) and executive inhibitory control (which can be considered a form of cool EF) (e.g. Luman et al. 2005; Solanto et al. 2001, but see Geurts et al. 2006). Both characterize ADHD, but not necessarily simultaneously. Castellanos et al. (2006) formulated the interesting hypothesis that inattention symptoms may be associated with deficits in cool EF, whereas hyperactivity/impulsivity symptoms reflect hot EF deficits. ADHD-research may however also have limited the conception of hot versus cool EF because hot EF has exclusively been studied in decision making paradigms, mostly (variants of) the Iowa Gambling Task (Bechara et al. 1994). Hot decision-making paradigms are indeed those that involve risks and rewards (Kerr and Zelazo 2004), but the definition of hot EF suggests that the distinction is applicable to most if not all executive functions. This is important because, within the concept of hot EF, further distinctions may be justified based on which aspect of affect/emotion is involved in performance of a task. For example, frustration and motivation (as measured by for example decision making or varying reward paradigms) are very different emotions with potentially different effects on task performance and different neurobiological correlates. Frustration is more closely related to stress, which, like nicotine, affects ANS/HPA-functioning (stress: Dickerson and Kemeny 2004; Van Goozen et al. 2000; nicotine: Browne et al. 2000; Horne et al. 2004; McDonald et al. 2006; Ramsay et al. 1996). Despite some neuro-anatomical overlap between brain areas active during stress and during motivationally driven situations (e.g. orbitofrontal (OFC-) involvement: Hynes et al. 2006; Rolls 2004), there is no clear-cut evidence for strong ANS/HPA-involvement during decision making tasks (i.e. tasks tapping into motivation). Thus, children of mothers who smoked during pregnancy might particularly demonstrate hot EF-deficits when stress or frustration are involved, but no hot EF deficits when motivation (or reward sensitivity) is involved. Limitations and recommendations for further studies Like prenatal smoking, deficient ANS/HPA-functioning is also particularly associated with conduct problems and co-occurring conduct problems and hyperactivity (Herpertz et al. 2001; Waschbusch 2002). Unfortunately the number of participants in the present study was too small to reliably distinguish children with solely hyperactivity, solely conduct problems, and with co-occurring hyperactivity and conduct problems. Larger samples would also be required to investigate whether children with conduct problems present with different types of hot EF-deficits than children with ADHD. The main problem of a small sample size is that it limits the strength of confidence one can have in the generalizability of findings to other samples. That is why this study should be considered exploratory. Furthermore, it would be desirable for future studies to incorporate a number of control variables in the analysis of hot versus cool executive functions in children of mothers who smoked during pregnancy. There are many different types of variables that could account for associations between maternal prenatal smoking and hot EF-functioning. One important candidate is parental education or IQ, which unlike potential confounders such as alcohol or drug use during pregnancy, quality of parenting, family functioning, family status, mother’s age when she had her first child and parental psychopathology, explained associations between maternal prenatal smoking and children’s cognitive outcome measures in a number of studies (Breslau et al. 2005; Huijbregts et al. 2006; Trasti et al. 1999). Moreover, associations between maternal prenatal smoking and offspring cognition remained significant in studies where parental education or IQ were not introduced as potential confounders (e.g. Batstra et al. 2003). Despite the fact that these studies investigated associations of prenatal smoking with general cognitive outcome measures such as IQ and academic achievement, which have been reported to correlate significantly with measures of cool but not hot EF (Hongwanishkul et al. 2005), and despite the fact that participants in the present study were all from the same area of Southampton which can be considered relatively uniform regarding socio-economic status (family income, parental education etc.), we have not included a direct control measure for parental IQ or education and can therefore not rule it out as a potentially significant confounder. Moreover, confounders of the association between maternal prenatal smoking and hot EF may be different from those that explain associations between maternal prenatal smoking and cool EF. Another limitation is that, even though our study revealed a clear difference between results on the two inhibitory control tasks, we did not collect children’s IQ—or academic achievement measures ourselves. The interpretation of our results would have been stronger if, like Hongwanishkul and colleagues, we would have been able to show that such measures were associated with our cool EF-measure, whilst they would not have had a significant relation with our hot EF-measure. A final limitation we would like to address concerns the validity of our claim that the Delay Frustration Task measures hot inhibitory control whereas the SA-Dots task measures cool inhibitory control. Whereas the SA-Dots task has been used regularly before to measure inhibition of prepotent responding in different clinical populations characterized by deficient inhibitory control (e.g. Huijbregts et al. 2002a; Konrad et al. 2004; Slaats-Willemse et al. 2003), the Delay Frustration Task was originally developed to measure delay aversion (Bitsakou et al. 2006). In the dual pathway model of ADHD, delay aversion is considered a largely autonomous construct (Sonuga-Barke 2002). The advantage of considering this task a measure of delay aversion is that one could then make a stronger case for it being a measure of “hot EF”, as such tasks have been used to measure this construct before. Still, the developers of this task recognize that it is likely that the Delay Frustration Task measures inhibitory control (i.e. participants have to stop a “prepotent” or “ongoing” response during the delays) (Bitsakou et al. 2006). Although based on Zelazo and Müller’s theory DeFT should be regarded as a hot EF task regardless of whether it is considered a delay aversion or an inhibition task, it might be best for future studies to select tasks which are completely similar apart from an ‘affect-’ (e.g. frustration) component in one version. The important message of this study is that two forms of EF can be distinguished which can be classified as hot and cool EF, and that in children of mothers who smoked during pregnancy a discrepancy can be observed between these two. Children whose mothers smoked during pregnancy are particularly at an increased risk of demonstrating (frustration-induced) hot EF-deficits. When these results are confirmed in larger samples that take into account the many potential confounders, intervention programs aimed at child factors could incorporate the training of specific emotion regulation skills in addition to the training of ‘cool’ EF-skills. Ideally, at risk families are identified before or early during pregnancy. Although prenatal smoking occurs in approximately 25% of pregnant women in Western countries (Breslau et al. 2005; Huijbregts et al. 2006), it has been shown that a family history of antisocial behavior in combination with prenatal smoking seriously compounds the risk of offspring disruptive behavior (Huijbregts et al. in press). Thus, like successful intervention programs aimed at increasing the knowledge about health (including smoking cessation) and care of mothers-to-be such as the Nurse Home Visitation Program (Olds et al. 1999), intervention programs aimed at improving child factors such as their emotion regulation skills should probably be provided to children with a combination of risk factors. When it is confirmed that ANS/HPA abnormalities mediate associations between prenatal smoking and children’s disruptive behavior and hot EF, clinical implications might also involve reinstating normal ANS/HPA axis functioning pharmaceutically (Van Goozen et al. 2007), for example by temporarily saturating glucocorticoid receptors. By restoring stress response systems, emotion regulation and related cognitive functioning could be improved, perhaps even more strongly in disruptive children whose mothers smoked during pregnancy than in disruptive children whose mothers did not smoke during pregnancy.

Eur_J_Health_Econ-_-_-1388085

The “Health Benefit Basket” in The Netherlands

This contribution describes the entitlements in Dutch health care and explores how these entitlements are determined and to whom they apply. The focus is on services of curative care. No comprehensive positive or negative list of individual services is included in formal laws. Instead, the legislation states only what general types of medical services are covered and generally the “usual care” criterion determines to which interventions patients are entitled. This criterion is not very restrictive and yields local variations in service provision, which are moderated by practice guidelines. It is conceivable, however, that the recent introduction of the DBC financing system will change the reimbursement and therefore benefit-setting policy. This contribution gives a detailed account of entitlements in the Dutch health care system. Furthermore, it explains who decides about the services covered and the legal context in which this is done. It also describes which parts of the health benefit basket are defined more implicitly and which more explicitly, outlining the criteria used for the definition of the basket. We present the main characteristics of the Dutch health care system, highlighting those that impact on the entitlements of the insured. We then present a detailed account of health care entitlements in curative care. Finally, we conclude by identifying the aspects that determine the health care package and require special attention in a European context. We specifically ask what future developments are likely to impact on this system for definition of entitlements. The main characteristics of the system The Dutch Constitution states that the government should provide services to improve and protect public health. Therefore about 80% of health expenditures are paid from public funds (mainly social insurance), the rest being financed from private insurance and out-of-pocket payments [1]. The health insurance system broadly consists of three compartments (see Table 1). First, every citizen is entitled to long-term care covered under the Exceptional Medical Expenses Act (Algemene Wet Bijzondere Ziektekosten, AWBZ). The AWBZ scheme pays for about 43% of Dutch health expenditure. Secondly, for acute care individuals are either publicly insured under the Sickness Fund Act (Ziekenfondswet, ZFW) or privately insured (37% and 15% of healthcare expenditure, respectively). The ZFW scheme covers among other things the costs of hospital admissions, treatment by physicians, pharmaceutical care, and medical aids.Table 1 The three “pillars” of the Dutch health care systemFirst pillarSecond pillaraThird pillarType of careLong-term careBasic curative servicesAdditional servicesPrimary regulation (laws)AWBZb,cPrivate insurancec, WTZc, ZFWbComplementary insuranceb,cSecondary regulation (decrees)Decree on Entitlement to Exceptional Medical Expenses InsuranceHealth Insurance Treatment and Services Decree–Tertiary regulation (discretion of Minister)Regulation on Subsidies in AWBZRegulations on Medical Specialist Care–Regulations on Entitlements to AWBZ-careRegulations Governing the Provision of Paramedical Assistancea To be replaced in 2006 by a statutory health insurance scheme that covers the entire populationb Administered by sickness fundsc Administered by insurance companies Persons who do not qualify for compulsory health insurance but cannot satisfy private insurers’ acceptance criteria (for example, if they represent a high risk) have access to a “standard insurance package” which is similar to that of the ZFW. This entitlement is specified in the Medical Insurance Access Act 1998 (Wet op de toegang tot ziektekostenverzekeringen, WTZ). The third compartment consists of complementary insurance policies that may be bought to extend coverage. The basic insurance package is defined mainly at the central government level, although only categories of services are identified. The AWBZ and ZFW (acts of Parliament) offer a list of the functional categories of health care services to which a patient is entitled (e.g., pharmaceutical care, hospital care, rehabilitation services). These formal laws indicate in general terms which areas of care are covered by the insurance scheme, but they do not specify the entitlements of the insured in detail. Usually further specification is provided in lower types of regulations by the government or the Ministry of Health, Welfare, and Sport and its advisory bodies. The lower forms of regulation specify the contents and the extent of the entitlements, specify conditions, and delegate responsibilities. Two important governmental decrees are associated with the AWBZ and the ZFW, respectively: the Decree on Entitlement to Exceptional Medical Expenses Insurance and the Health Insurance (Treatment and Services) Decree. The social health insurance system ensures that patients receive the care to which they are entitled. Management of the two-tiered public/private system, however, has proven difficult for the government, especially since the system lacks sufficient incentives for efficiency and has unfavorable equality characteristics. These conditions stimulated major reforms. Government involvement will be reduced significantly in 2006 when a new Health Insurance Act comes into force. This forms the final phase of the so-called Dekker plan (1987). The Dekker proposals set out conditions for the introduction of regulated competition in health care which should make the system more responsive to the preferences of consumers and less burdensome to manage. Health insurers have become main actors in promoting cost containment since they now hold full budget responsibility and may enter into contracting with providers. The government has promoted competition by enforcing open enrollment with health insurers and through establishing agencies that should regulate competition (e.g., an antitrust agency has been active in the health care sector since 1998). Nevertheless, the government will remain responsible for defining the health basket. A comprehensive description of the Dutch health insurance system is presented elsewhere [2]. The following section concentrates on the services made available in the health basket and the way in which these are decided. This description is limited to the curative sector. A complete account of health care services available in all health care sectors is offered by Stolk and Rutten [3]. Note that the integration of social and private insurance schemes into a national insurance implies that the coverage of the ZFW will be extended to the entire population. For this reason the present contribution describes only the ZFW benefit package. The benefit package for curative care The definition of entitlements for curative care is typical for the Dutch system: new interventions are usually implicitly introduced into a largely unspecified benefit package. A small negative list explicitly excludes services from reimbursement. This reflects the absence of a systematic procedure for the evaluation of curative services when defining the basket. Medical specialist care The ZFW specifies a general entitlement to specialist care by hospitals (both inpatient and outpatient care): patients are entitled to medical, surgery, and obstetric care. Entitlement to hospital care also exists under the AWBZ if the patient must stay in hospital longer than 365 days. The provided care is limited merely by professional norms following the “usual care” principle (appropriateness according to professional standards). Also, a general statement is made that health care should be provided efficiently. Therefore the definition of entitlements for hospital care is rather implicit and the “usual care” criterion is not very restrictive. However, the entitlements are conditional upon the referral by a general practitioner, by another specialist to whom a general practitioner had referred the insured, or, where obstetric care is concerned, by a midwife. A ministerial decree, the Regulations on Medical Specialist Care, further specifies entitlements by excluding specific forms of transplantation and plastic surgery. It also determines that only certain types of transplantations are covered, and states that entitlement to some services (e.g., plastic surgery) only exist if specific conditions are met. Finally, this regulation explicitly excludes services such as eyelid, ear or body sculpturing, in vitro fertilization (IVF), uvoloplasty for persons who snore, sterilization or undoing sterilization, and circumcision. It should be noted that IVF is a special case and its reimbursement policy changed several times. Today, health insurers are subsidized to fund IVF treatment for the insured after the first attempt has been paid out of pocket. Pharmaceuticals dispensed in hospitals are part of the entitlement to hospital care. They are generally financed out of the hospital budget, but the insurer may separately reimburse an additional percentage (up to 80%) of the cost for very expensive hospital drugs. The law remains rather vague about entitlements to hospital care, but a recently introduced pricing system has led to more explicit benefit definition. In 2005 a new system for hospital financing was introduced by law: the Diagnose Behandeling Combinaties (DBCs), a DRG-like system describing all products that are provided in hospitals. Medical experts were involved in the determination of these DBCs, which are defined as the whole set of activities (diagnostic and therapeutic interventions) of the hospital and medical specialist starting from the first consultation and diagnosis of the medical specialist in the hospital until discharge. Each DBC is characterized by a code combining information on diagnosis (International Classification of Diseases, 10th edition) and treatment. This system facilitates negotiations between health insurers and hospitals on prices (on a bilateral level), at the same time providing a catalogue of medical services. Table 2 shows how many different DBCs are defined within a medical specialty distinguishing between list A (prices fixed by the National Health Tariffs authority until further notice) and list B (prices negotiated by Sickness Funds and hospitals from January 2005 onwards). This catalogue of DBCs allows the specification of which DBCs are included in the basis package: the “red” list presents the number of DBCs not covered by compulsory insurance as determined by the Health Care Insurance Board on the basis of the “usual care” criterion and current explicit exclusions. Also an “orange” list exists with DBCs for which reimbursement is limited to certain groups of patients. Finally, the table lists the number of product groups that have been defined to cluster DBCs into homogeneous price groups, as negotiation on the price of each single DBC would not be feasible.Table 2 Overview of the DBC system (January 2005)SpecialtyNumber of DBCsProduct groupsaList AaList BaRed listbOrange listbOphthalmology1,7893181415Otorhinolaryngology6926–3514Surgery3,671182850553Plastic surgery13,7126320554433Orthopedics 3,0581412443Urology37,71795011617438Gynecology5362263434Neurosurgery1,25327151329Dermatology39610323311Internal medicine3,33519–61249Pediatrics3,4921091261Gastroenterology10,946190–34221Cardiology353––4439Respiratory medicine1,054145537143Rheumatology2,042129519617Allergology599––8Thorax surgery1,034–105027Psychiatry648–––3Neurology2,73918446143Geriatrics918–––16Radiotherapy468––19514Radiology13,216––4728Anesthesiology582–103819Clinical genetics100––3Total104,3501,3761,5113,649641a Published at the website of the National Health Tariffs authority (CTG/ZAio): http://www.ctg-zaio.nl/index.phpb Published at the website of the DBC maintenance organization: http://www.DBConderhoud.nl An exception to the way in which medical specialist services are regulated is mental health care, which is currently covered under the AWBZ. Also the first year of treatment is covered by the AWBZ and not by the ZFW, as is the case for other medical specialist services. The reason is that psychotherapy is often a long and intensive treatment. Ambulatory psychotherapy is reimbursed (maximum of 25 sessions, 50 in the case of personality disorders) if referral is by a general practitioner or psychiatrist. Patients’ entitlements to mental health care through the AWBZ include treatment (therapy sessions and medication), supportive guidance, and accommodation. Special regulations apply to mental health care in children and to treatment of addictions. The objective is to make mental health care a part of the basic benefit package under the new Health Insurance Act and therefore shift it to compartment 2 in a competitive environment. To permit this change DBCs in mental health care are now being defined. Primary care, dental care, and paramedical services Primary care, i.e., medical and surgical care provided by general practitioners, is covered under the ZFW, as specified in the Health Insurance (Treatment and Services) Decree. Care involves mainly consultations and visits, the prescription of pharmaceutical care, referral to medical specialists, and minor operations. Entitlements to these services are also defined implicitly: any type and quantity of care consistent with professional norms is covered by the scheme. Entitlements to dental care have been limited in the past and are defined explicitly. The ministerial regulation “Dental Care Health Insurance” stipulates that persons up to the age of 18 years are entitled to dental care under the ZFW. Dental care includes 14 types of services, among which are periodic checkups, fluoride application treatment, sealing, and periodontal care. Adults are entitled to dental care only under special circumstances: special dental conditions and physical or mental disabilities. Paramedical care—consisting of physiotherapy, Mensendieck or Cesar remedial therapy, speech therapy and ergotherapy—is partly covered. The first nine sessions are excluded from reimbursement, but thereafter some patients over the age of 18 years are entitled to physiotherapy, Mensendieck or Cesar remedial therapy for the treatment of chronic conditions that are defined in the Regulations Governing the Provision of Paramedical Assistance. Curative care at home Under exceptional conditions and with authorization by the sickness fund, specialized services of curative care may be provided at home. First, persons are entitled to kidney dialysis at home or in a dialysis center under ZFW. If the dialysis takes place at home, the costs of training a person to carry out dialysis or providing assistance during the procedure are also covered. The costs of inspecting and maintaining equipment, modifying the home, providing special sanitary fittings, and heating are also reimbursable. Second, the ZFW regulates the use of treatment for chronic intermittent ventilation. Those covered by the ZFW are entitled to treatment at a ventilation center. A center may lend the patient the equipment for use at home or in a location where several persons can use the equipment. Specialist and pharmaceutical care provided by or on the advice of the ventilation center are also included. Discussion A liberal benefit package? In The Netherlands decisions about the medical services offered to a patient are often left to the physician (and/or the health insurer). Negative lists are in certain categories used to exclude services (e.g., the new red and orange lists of DBCs). The extent to which new or old technologies (therapy, diagnostic services, and pharmaceuticals) are provided to the patient in the context of a DBC is left to the providers. However, some medical services were explicitly excluded for not being considered to belong to the domain of health care (e.g., plastic surgery). Additionally, limits were set to the provision of physiotherapy and psychotherapy for a lack of demonstrated effectiveness. Restrictive legislation only exists in the pharmaceutical sector where an itemized positive list of individual products is frequently updated [4]. For several other functional categories there are also positive lists, but these are not frequently updated (e.g., for medical devices and vaccines). Regarding pharmaceuticals it is interesting to observe that the reimbursement of extramural drugs (prescription filled by the community pharmacist) is centrally regulated, while for the use of drugs in the hospital the policy is liberal. The liberal policy has led to wide variations in the availability of certain expensive drugs across hospitals and to referral of “expensive” patients to specialized centers. The new hospital financing system will change this and will probably force decision makers to regulate the provision of expensive drugs within a DBC. Procedures and criteria for limiting the health basket The package can be limited by management of positive or negative lists, copayments, the definitions of conditions restricting eligibility, and promoting best practice, i.e., effective and efficient provision of care. Each instrument is applied within the Dutch system, although some sectors are more loosely managed than others. On the whole, emphasis is put on offering incentives for efficiency. Recently a no-claim system was introduced: if someone insured under ZFW has no expenses, he receives a bonus of €255 from his insurer or the difference if his expenses are less than €255. The liberal character of the basket is further illustrated by the fact that there is no systematic and central effort to use common criteria for defining the package as a whole. Nonetheless, decisions tend to be guided by a quite homogeneous set of considerations. The respective Ministers of Health and more specifically the Health Care Insurance Board have been using health technology assessment (and cost-effectiveness information) since the middle 1980s as a tool for deciding on large public programs (e.g., national cancer screening programs) and on the addition of expensive curative programs (e.g., transplant programs) to the package [5]. Especially in the case of pharmaceuticals dispensed outside the hospitals the role of cost-effectiveness as criterion for reimbursement is quite prominent. If the manufacturer can demonstrate greater effectiveness than existing drugs and wants a premium price, he must submit a pharmacoeconomic dossier and figures allowing the Health Care Insurance Board to base its policy on information on (cost-)effectiveness. Furthermore, many technology assessment studies have been financed from a special fund managed by the Health Care Insurance Board and later by the National Medical Research Council that have produced valuable information on the cost-effectiveness of new diagnostic and therapeutic procedures. Although not legally required, this information has been used to define the benefit package or for the development of practice guidelines. Further criteria have emerged as constant factors in decision making: severity of illness, individual vs. collective responsibility, affordability, and leakage (the probability of use outside a targeted patient group) [6]. Although initially consistent use of these criteria may have been doubted, gradually patterns emerged. Severity of illness, for example, played a role in deciding to finance the lung transplantation program although it was demonstrated to have a high cost-effectiveness ratio [7]. In relation to budget impact it also played a role in the decision not to fund PDE-5 inhibitors in the treatment of erectile dysfunction. Responsibility and affordability have played a role in deciding to restrict the reimbursement of pills for minor illnesses and dentistry. Leakage played a role in the decision to limit funding of some medicines (e.g., clopidogrel) to specific patient group. The increased understanding of the balance between the criteria is also recognized by policy makers. The role of practice guidelines An important way to influence the providing of service is through practice guidelines. There are many disciplinary and interdisciplinary practice guidelines for both primary and secondary care, some of which are also based on economic evidence. In The Netherlands a government program supported the development of practice guidelines in 1998–2002 which led to 31 consensus guidelines for medical practice addressing 23 disorders across seven disease groups as defined by the International Classification of Diseases. One example is the guideline for the use of cholesterol-lowering drugs in primary and secondary prevention which sets rules for the prescription of statins according to the patient’s personal risk profile. Future prospects An important development is the change in hospital financing, involving description of the benefit package in terms of DBCs which are or are not reimbursed. Currently it is being discussed how the definition of DBCs can be more detailed to include specific regulations on the technologies to be used within a certain treatment. A so-called “maintenance organization for DBCs” is set up to (re)define DBCs. The reforms of 2006 are intended to reduce government control in the health care sector and provide incentives to patients and insurers for cost consciousness. Health care should follow the preferences of consumers. It follows that entitlements are formulated in functional form rather than as specific facilities or services. It is still not clear whether local actors will assume a role in defining the basket, and whether the government will centralize decisions.

Pediatr_Nephrol-4-1-2214827

Hypertension in children with chronic kidney disease: pathophysiology and management

Arterial hypertension is very common in children with all stages of chronic kidney disease (CKD). While fluid overload and activation of the renin–angiotensin system have long been recognized as crucial pathophysiological pathways, sympathetic hyperactivation, endothelial dysfunction and chronic hyperparathyroidism have more recently been identified as important factors contributing to CKD-associated hypertension. Moreover, several drugs commonly administered in CKD, such as erythropoietin, glucocorticoids and cyclosporine A, independently raise blood pressure in a dose-dependent fashion. Because of the deleterious consequences of hypertension on the progression of renal disease and cardiovascular outcomes, an active screening approach should be adapted in patients with all stages of CKD. Before one starts antihypertensive treatment, non-pharmacological options should be explored. In hemodialysis patients a low salt diet, low dialysate sodium and stricter dialysis towards dry weight can often achieve adequate blood pressure control. Angiotensin-converting enzyme (ACE) inhibitors and angiotensin receptor blockers are first-line therapy for patients with proteinuria, due to their additional anti-proteinuric properties. Diuretics are a useful alternative for non-proteinuric patients or as an add-on to renin–angiotensin system blockade. Multiple drug therapy is often needed to maintain blood pressure below the 90th percentile target, but adequate blood pressure control is essential for better renal and cardiovascular long-term outcomes. Introduction Hypertension is one of the most common sequelae of chronic kidney disease (CKD) in children [1]. CKD-associated hypertension develops by a large variety of pathophysiological mechanisms. Blood pressure is one of the most critical determinants of the progression rate of renal failure in children [2, 3], and cardiovascular mortality in childhood onset renal failure [4]. Therefore, good antihypertensive management can substantially contribute to better renal and patient survival of adults with childhood-onset CKD. As in many other complications of chronic renal failure, patients often do not report symptoms of hypertension, so an active screening approach is needed to prevent end-organ damage. This article aims to give a short overview of the different pathophysiological pathways that lead to hypertension in CKD in order to explain how these can be targeted by different therapeutic approaches. Pathophysiology of hypertension in CKD Fundamentally, increased blood pressure is caused by an increase in cardiac output and/or of total peripheral resistance. Both can be altered by a plethora of different mechanisms in uremia and renal failure. Additionally, children with certain underlying diseases, e.g. glomerulopathies and polycystic kidney disease, are especially susceptible to hypertension [3, 5]. Figure 1 gives an overview of the most important pathways involved. Fig. 1Interplay of different factors in the generation of hypertension in chronic kidney disease (BP blood pressure, CO cardiac output, TPR total peripheral resistance, PTH parathyroid hormone, Na sodium) Activation of the renin–angiotensin–aldosterone system plays a pivotal role in renal hypertension. While plasma renin activity is typically found to be markedly elevated only in patients with renal artery stenosis, many patients with CKD have ‘inappropriately normal’ renin levels (i.e. lower levels would be expected, considering their degree of hypertension and fluid overload [6, 7]). Hyper-reninemia occurs probably due to renin secretion in poorly perfused areas such as cysts and scars or after microangiopathic damage or tubulo-interstitial inflammation [8, 9] and leads to angiotensin II-mediated vasoconstriction as well as aldosterone-mediated salt retention, thus increasing both total peripheral resistance and blood volume. Additional delayed effects of a high angiotensin II tone include inflammation, cardiac hypertrophy and endothelial cell damage, mesangial cell proliferation and fibrosis [10], which contribute further to hypertension and end-organ damage. Sodium retention and consequent fluid overload have long been recognized as causes of hypertension in CKD. Hypertensive children on dialysis have lower residual urine output than their normotensive peers have [5]. While inter-dialytic weight gain is correlated with the inter-dialytic increase in ambulatory blood pressure, the correlation is rather weak (in children r = 0.41 [11]). This may be due to delayed effects [12] but also points to important volume-independent factors regulating blood pressure (BP) in patients on hemodialysis. This is also illustrated by the fact that nephrectomy in children on dialysis lowers mean blood pressure, despite causing anuria [13]. It has been proposed that fluid overload leads to hypertension only in those patients in whom peripheral resistance fails simultaneously to fall, i.e. when additional factors interfere with vascular autoregulation [14]. A growing body of evidence suggests that increased activity of the sympathetic nervous system (SNS) is an important volume-independent cause of hypertension. Campese et al. demonstrated that renal denervation improves both hypertension and increased sympathetic activity caused by phenol injection into rat kidneys [15]. Muscle sympathetic nerve activity is also elevated in hypertensive patients with chronic renal failure [16]. The underlying mechanisms of this phenomenon are, as yet, unclear and may include afferent signals from the failing kidney as well as dopaminergic abnormalities and the accumulation of leptin in CKD [17, 18]. Interestingly, not only beta blockade but also angiotensin-converting enzyme (ACE) inhibition can reduce the sympathetic hyperactivation of CKD [16, 19]. However, as sympathetic hyperactivity is also a feature of renovascular hypertension [20], essential hypertension and hypertensive patients with polycystic kidney disease [13], it appears that sympathetic activation also occurs independently of renal function. The most established cause for sympathetic over-activation is renal ischemia caused by renal artery stenosis [20, 21], but renal cysts might also cause local renal ischemia. While children with end-stage renal disease (ESRD) usually have normal plasma noradrenaline and adrenalin concentrations, hemodialysis per se leads to substantial increase in both plasma renin activity and catecholamines, which can contribute to hypertension [22]. Recent experimental evidence suggests that renalase—an amine oxidase specifically expressed by the kidney—lowers blood pressure and heart rate. Its activity is markedly reduced in patients with ESRD [23]. However, whether the cardiovascular effects of this enzyme are really due to its catecholamine-metabolizing activity is still controversial [24]. There has been debate about the role of nitric oxide (NO) in mediating endothelial cell damage and hypertension in CKD. Newer studies have demonstrated that, in uremic patients, reduced NO stimulation leads to reduced agonist-induced endothelium-dependent vasodilatation, whereas other vasodilatory pathways are not affected. Renal failure leads to the accumulation of endogenous NO synthase inhibitors such as asymmetric dimethyl-L-arginine (ADMA), which appears to be due to increased generation and decreased metabolism rather than decreased clearance [25]. ADMA independently predicts overall mortality and cardiovascular events in patients with ESRD [26], as well as progression of CKD [27]. While ADMA is related to blood pressure in animal models of CKD, clinical studies have not found differences in blood pressure [25]. Endothelial NO synthase is also suppressed by hyperparathyroidism in rats with CKD [28]. In contrast to ADMA levels, those of serum parathyroid hormone (PTH) correlate highly with blood pressure in patients with CKD [29]. Whereas acute infusion of PTH has a hypotensive effect, chronic hyperparathyroidism leads to accumulation of calcium inside vascular smooth muscle cells, enhancing their sensitivity to calcium and norepinephrine [30, 31]. This effect can be blocked by calcium channel antagonists. A number of drugs commonly administered in CKD can cause iatrogenic hypertension. For example, erythropoietin (EPO) causes blood pressure elevation over several weeks. This may be via arterial wall remodeling, causing increased vascular resistance [32]. Vaziri et al. have proposed that EPO acts directly on voltage-independent calcium channels in smooth muscle cells, leading to a decreased sensitivity to the vasodilatory action of nitric oxide [33]. The resulting possibility of calcium channel antagonists as ‘specific’ therapy for EPO-induced hypertension has been successfully tested in rats [34]. Glucocorticoids lead to fluid retention by their mineralocorticoid effect. Cyclosporine A causes vasoconstriction of glomerular afferent arterioles and hyperplasia of the juxtaglomerular apparatus, with subsequent increased release of renin and angiotensin II [35]. Increased circulating catecholamines and endothelin-1 precursors, and increased renal sodium absorption via the Na-K-2Cl co-transporter in the loop of Henle [36], have also been demonstrated after cyclosporine A treatment. Tacrolimus appears to be less pro-hypertensive than cyclosporine in children after renal transplantation [37], but, for other reasons, it is less commonly used prior to transplantation [38]. Treatment with growth hormone leads to water and sodium retention by the distal nephron [39], mediated by increased intra-renal insulin-like growth factor (IGF)-1. However, growth hormone (GH) does not appear to increase blood pressure in children with CKD or Turner syndrome, despite both groups being prone to hypertension [40, 41]. Management Owing to the lack of acute symptoms and to the serious long-term consequences of hypertension, an active screening approach is necessary to detect elevated blood pressures early and to prevent end-organ damage [42]. In contrast to other complications of CKD, which become prevalent only in later stages of CKD, hypertension is already very common in CKD stage 1, with over 63% of children affected. In CKD stages 4 and 5 the incidence increases further to 80% [1]. More than 50% of children with ESRD have uncontrolled hypertension, despite widespread use of antihypertensive drugs [5, 43]. The frequency of blood pressure measurement during screening and during therapy should be appropriate to the patient’s risk of developing uncontrolled hypertension. A routine for this is suggested in the flow chart in Fig. 2. In general, we would consider all children with CKD to be at least at intermediate risk, and, therefore, we recommend 3-monthly clinic blood pressure (CBP) measurements; children with ESRD, multiple pro-hypertensive medications or confirmed hypertension should be considered as at high risk. In our experience 24 h ambulatory blood pressure measurement (ABPM) is a very valuable tool and should be performed annually in high-risk populations such as renal transplant recipients or those with rapidly progressive renal disease. ABPM is also useful to exclude white-coat hypertension, which is a problem even in children under long-term medical care [44]. While there are equivocal results about the long-term consequences of white-coat hypertension, end-organ damage could be clearly demonstrated in children with masked hypertension, i.e. elevated ABPM but normal CBP [45], underlining the usefulness of this method. Fig. 2Flow diagram for the choice of method to measure blood pressure during screening, follow up and treatment (CBP clinic blood pressure ABPM 24 h ambulatory blood pressure monitoring, HBP home blood pressure measurement BP blood pressure, BMI body mass index) Whichever method is used, techniques should be in accordance with international consensus statements, and the appropriate pediatric reference ranges are very important. These are now available for clinic [46, 47], ambulatory [48] and home blood pressure measurements [49]. The diagnosis of hypertension should be based on at least three clinic blood pressure measurements above the 95th percentile [46]. As white-coat hypertension is very common, and the effects of thorough investigation and treatment can be far-reaching, we feel ABPM is advisable in nearly all newly diagnosed hypertensive children. In countries where health care providers do not cover the cost of ABPM, home blood pressure measurements may be helpful in confirming the diagnosis, but they do not pick up the nocturnal blood pressure dynamics [44]. Once hypertension has been confirmed, good management should not focus only on pharmacological therapy, but also on detection and treatment of end-organ damage and, where appropriate, improvement of the dialysis regime and consideration of therapeutic life-style changes. Echocardiography and ophthalmological examination are the most important assessments of end-organ damage in hypertensive children. Left ventricular hypertrophy (LVH) is common in children with CKD, even during antihypertensive therapy [50]. LVH in patients with normal clinic blood pressure may indicate masked hypertension in untreated children [45] or insufficient efficacy of antihypertensive drug therapy in treated children (e.g. insufficient dose, non-compliance, or short duration of action of selected antihypertensive drugs). Echocardiography and ophthalmological evaluation should be repeated at regular intervals in children with initial signs of end-organ damage [46]. From personal experience we would also recommend follow-up examination in children with persistent hypertension or in those at high risk (e.g. ESRD). Additional assessments of end-organ damage include measurement of carotid intima media thickness and of pulse wave velocity, which reflect functional alterations of arterial wall properties caused by hypertension and other factors such as hyperparathyroidism [51]. These investigations are still mainly used for research purposes, but clinical use has been facilitated by the recent provision of pediatric reference ranges [52]. In hypertensive patients who are undergoing renal replacement therapy, improvement of the dialysis prescription should be the primary therapeutic approach to hypertension, before pharmacological treatment is started.Extracellular volume overload can be efficiently reduced by adequately long dialysis times. Even though the reduction of fluid overload may take several weeks to translate into normalized blood pressure, drug-free control of hypertension is possible in many, if not most, hemodialysis patients [12]. A number of studies have shown improved BP control with short daily, long intermittent or nocturnal hemodialysis [53]. While conventional dialysis schedules can also improve control of hypertension by targeting dry weight more aggressively [54], this tends to be tolerated less well, with more intra-dialytic hypotensive episodes. Lower concentrations of dialysate sodium also produce a moderate fall in blood pressure on a population level, with best results in patients with previously high blood pressure [55]. In addition to intensified dialysis, dietary sodium restriction is a useful adjunct to intensified dialysis in avoiding sodium overload, and it reduces intra-dialytic hypotensive episodes. Indeed, some authors find dietary sodium restriction plus low sodium dialysate to be equally effective as time-intensified dialysis, and they maintain that with good dietary advice a low salt diet need not be unpalatable [56]. However, restriction of fluid and salt intake requires considerable patient motivation, which is often a problem in the adolescent population. Therapeutic life-style changes can be effective in lowering blood pressure if obesity contributes to the patient’s hypertension. Obesity in children with CKD is uncommon and usually associated with steroid treatment; however, it is seen more commonly in populations with a higher background risk of obesity [3]. While weight loss can be effective in reducing blood pressure in overweight children with normal renal function [57], unfortunately, there are no controlled studies of life-style interventions in children with CKD. In our experience calorie-reduced diets are rarely effective in this population, probably due to the multifactorial etiology of hypertension in renal disease. Also, the recommended intake of fresh fruit and vegetables may be hard to achieve if a potassium- and phosphate-reduced diet is also necessary. Interestingly, weight loss improves salt-induced increases in blood pressure in obese children [58]; therefore, combined calorie- and sodium-reduced diets may be particularly effective in obese CKD children who are salt retainers. Pharmacological treatment remains the mainstay of antihypertensive management in all stages of CKD. In approximately 75% of children with CKD stages 2–4, blood pressure control below the 95th percentile can be achieved by antihypertensive monotherapy, but 50–60% of children need more than one drug if intensified BP control (<50th percentile) is targeted [Effect of Strict Blood Pressure Control and ACE Inhibition on Progression of CRF in Pediatric Patients (ESCAPE) trial, unpublished results]. In children with ESRD adequate control is much harder to achieve: a review of the North American Pediatric Renal Transplant Cooperative study (NAPRTCS) database also showed that over 50% of children on dialysis have uncontrolled hypertension [43], and a Polish nationwide survey found that in only 57% of children with ESRD was hypertension adequately controlled, despite the use of multiple drug therapy in 65% [5]. Even though multiple drug therapy is often required, it is advisable to start with a single drug at a low dose and to titrate upward until blood pressure is controlled [46]. Exceptions are hypertensive emergencies, such as hypertensive encephalopathy, when intravenous (i.v.) treatment should be started promptly. ACE inhibitors and angiotensin II type 1 receptor blockers (ARBs) are the most useful drugs, as they are not only antihypertensive but also slow down the progression of renal failure more efficiently than do other antihypertensive treatments [59]. Surveys among pediatric nephrologists from both sides of the Atlantic show that ACE inhibitors are increasingly popular and are now the most commonly used drugs for pediatric hypertension [60, 61]. The renoprotective effect of renin–angiotensin system (RAS) blockade is due to a combination of reduced proteinuria, lower intra-glomerular pressure through selective dilatation of the glomerular efferent arteriole, and anti-inflammatory and anti-fibrotic effects [10]. Additionally, RAS inhibition reduces the sympathetic hyperactivity seen in CKD [16]. There is convincing evidence from studies of adults that all proteinuric patients should receive renin–angiotensin system blockade, even if they do not have hypertension [42]. Major side effects are a moderate increase of potassium and creatinine; these should be monitored more closely with declining renal function. Sexually active adolescent girls must use contraception. ACE inhibitor-induced cough appears to be less common in children than in adults [62]. There is no clear evidence suggesting clinical superiority of ARBs over ACE inhibitors [63], and a number of drugs in both classes are labeled for children. Food and Drug Administration (FDA) approval for children under 6 years old is expected soon for a number of ARBs. Combination therapy with an ACE inhibitor and an ARB should be considered for patients who continue to show proteinuria while undergoing monotherapy, since it is effective in further reducing proteinuria (and progression of CKD) in adults [63, 64]. Experience in children is positive but very limited [65–68], and the increased risk of hyperkalemia and renal failure should be realized [69]. Therefore, for improved control of hypertension without proteinuria, the combination of single RAS blockade with a diuretic is preferable. The use of fixed-dose combinations of RAS antagonists with a thiazide diuretic may be an option in adolescents where compliance is an issue, but less so in younger children where frequent dose adjustments are required [46]. From trials in hypertensive patients without renal disease it appears that the blood pressure lowering effect, per se, is relevant for cardiovascular risk protection, without any class-specific benefits. Therefore, in essential hypertension, the choice of agents should be guided by the matching of the side effects profile of the individual drugs to the patient-specific risk factor profile [47]. In patients with CKD, meta-analyses of adult trials demonstrated superior renoprotection by ACE inhibitors, even after adjustment for blood pressure and urine protein excretion [70]. There is no conclusive evidence as to whether the inhibition of the renin–angiotensin system is superior to other antihypertensives in non-proteinuric CKD patients [59,71,72]. Recommendations for adults prefer to err on the side of caution and suggest a low threshold for considering CKD patients proteinuric (200 mg protein/g creatinine in spot urine) [42]. Diuretics are less commonly used in children with CKD than in adults with CKD, due to the preponderance of hypodysplastic kidney disorders, which frequently present as salt-losing nephropathies. In patients with evidence of hypervolemia, thiazides and loop diuretics have proven most useful for controlling volume overload, and they have a very good side-effects profile. It should be remembered that, while thiazides are a popular first-line therapy in mild-to-moderate CKD, they are less effective when glomerular filtration rate (GFR) falls below 60 ml/min per 1.73 m2 body surface area, and they are ineffective below 30 ml/min per 1.73 m2. Therefore, furosemide should be preferred and used in adequate doses for CKD stages 4 and 5. Mineralocorticoid receptor antagonists (e.g. spironolactone) are theoretically attractive in CKD, due to their synergistic actions with RAS antagonists, and the new selective receptor blocker eplerenone is devoid of anti-androgenic side effects. However, monotherapy (and, even more, combination therapy with RAS antagonists) is limited by the potentiated risk of hyperkalemia. During diuretic therapy, patients should be monitored for volume depletion and electrolyte disturbances. Long-acting formulations help to increase patient compliance [42]. Calcium channel blockers are very potent anti-hypertensive drugs and, therefore, useful as add-on therapy in children with resistant hypertension. Dihydropyridine (DHP) drugs (e.g. nifedipine, amlodipine) act mainly as vasodilators and do not have cardiac side effects. Amlodipine has pediatric labeling and is available as a suspension, and doses need not to be adjusted to renal function. However, DHP-type calcium channel blockers (CCBs) increase intra-glomerular pressure and proteinuria, while non-DHP-type calcium channel blockers (e.g. verapamil and diltiazem) have an additional anti-proteinuric effect. In a long-term clinical trial of elderly patients with type II diabetic nephropathy, non-DHP calcium channel blockers showed as equally an effective slowing of CKD progression, reduction of proteinuria and antihypertensive efficacy as did the ACE inhibitor lisinopril (while both were superior to the beta blocker atenolol [73]). However, there are no published safety data on any of the non-DHPs in children with hypertension, so they should be used with caution due to their known prolongation of the PR interval in adults [74]. Intravenous administration of nicardipine is an option for controlling hypertensive crises, especially when the level of renal function is unclear or changing rapidly. It has been used safely, even in very small children with hypertension, despite reports of hypotension in normotensive newborns with asphyxia [74]. Beta blockers can be used as second-line therapy for renal hypertension in children. However, they are contraindicated in asthma and can cause fatigue. All beta blockers require dose reduction with progression of CKD. They should be used with caution in heart failure, and their adverse metabolic effects make them less suitable for diabetics. The largest clinical experience, especially for infants, is available for propranolol. A sustained-release form of this drug allows once daily administration in larger children. However, other agents, such as atenolol, which have the advantage of being both long acting and β1-selective, may be preferred in clinical practice. Other drugs are used less commonly, mainly due to their more severe side-effects profiles. Alpha blockers (such as prazosin) can be used in patients who also require them for control of bladder emptying or Raynaud’s phenomenon. Centrally acting alpha agonists (such as clonidine) act via reduction of sympathetic nervous outflow. Rebound hypertension after discontinuation is a major problem of this drug. The vasodilators hydralazine and minoxidil are less suitable in CKD as they are less effective and cause salt and water retention. The aim of antihypertensive management is the regression of end-organ damage (especially LVH) and the lowering of blood pressure below target values while minimizing drug side effects. The currently recommended treatment goal in hypertensive patients with CKD is a blood pressure of < 130/80 mmHg in adults [42, 47] and < 90th percentile in children [46]. However, a meta-analysis of 11 randomized trials in non-diabetic adults with CKD showed differing results for proteinuric and non-proteinuric patients: while there was no increase in adverse renal outcomes with higher blood pressure in non-proteinuric patients, ideal systolic blood pressure for proteinuric patients (> 1 g per day) was 110–129 mmHg. In proteinuric patients, renal survival decreased, with systolic blood pressures below 110 mmHg and above 130 mmHg. These data are hard to interpret conclusively, as persistent hypertension during treatment may also be a reflection of more severe underlying disease [70]. In children, the pending results of the ESCAPE trial may help us to elucidate optimal blood pressure targets. Conclusion Adequate management of hypertension in CKD requires an active screening approach in order to prevent the significant renal deterioration and cardiovascular morbidity and mortality associated with high blood pressure. Owing to the plethora of different pathophysiological mechanisms involved, a whole range of therapeutic options is available. Non-pharmacological options should not be disregarded for obese children or for children on hemodialysis. Inhibitors of the renin–angiotensin system should be preferred for proteinuric patients and, probably, also for non-proteinuric patients. Multiple drug therapy is often necessary to reach target blood pressure below the 90th percentile. For this, diuretics and calcium channel blockers are the most suitable options.

Immunol_Lett-2-1-2430030

Murine neutrophils present Class II restricted antigen

Neutrophils were originally described as short lived, terminally differentiated phagocytes that contribute only to the innate immune response. Recent evidence of neutrophil cytokine production and expression of numerous cell surface proteins has suggested that neutrophils are likely to influence adaptive responses and may satisfy the criteria of antigen presenting cells. Under certain inflammatory conditions human neutrophils express major histocompatibilty complex (MHC) Class II and the costimulatory molecules CD80 and CD86. We have employed a murine T cell hybridoma with a transgenic T cell receptor specific for ovalbumin peptide 323–339 (OVA323–339), and a green fluorescent reporter of T cell receptor ligation, to directly investigate neutrophil-T cell interactions. These cells provide an ideal model system, allowing precise identification of antigen specificity and a clear readout of T cell activation. Additionally, whilst murine neutrophils have previously been shown to stimulate MHC Class I-dependent CD8+ T cell activation, CD4+ T cells stimulation via MHC Class II-expressing neutrophils has not been investigated. We addressed this by isolating murine neutrophils, loading with OVA323–339 and co-culturing with T cells specific for the OVA323–339/MHC Class II complex, and this resulted in T cell activation, as determined by expression of the green-fluorescent protein reporter. Antigen-pulsed neutrophils were also able to prime naïve OVA-specific CD4+ T cells in a contact-dependent manner, as shown by proliferation and cytokine production. Activation of lymphocytes was not due to contaminating macrophages. These studies demonstrate that murine neutrophils present MHC Class II-restricted peptides and induce T cell proliferation, confirming findings in human neutrophils, and demonstrate a novel pro inflammatory effect of murine neutrophils. 1 Introduction The range of functions ascribed to neutrophils has expanded significantly, from terminally differentiated, short-lived phagocytes that are solely involved in innate immune responses to cells that are now believed to influence the development of adaptive immune responses. Emerging evidence indicates that neutrophils may communicate with T cells through direct cell contact via major histocompatibility complex (MHC) Class I, Class II [1,2], and co-stimulatory molecules [3,4]. Additionally, a neutrophil-derived cytokine milieu [5] may have significant impact on the adaptive response to infection, for example via neutrophil-derived IL-12 [6,7]. There is considerable evidence to suggest that neutrophils satisfy the criteria of antigen presenting cells, with some capacity to process antigen. Human neutrophils express MHC Class II, either following in vitro activation via CD11b [8,9], with IFNγ, and GM-CSF [10,11] or following in vivo activation in RA synovial fluid [4,12], in patients with active Wegener's Granulomatosis [13], in patients receiving GM-CSF [14], and in persistent localized Staphylococcus aureus infections [15]. In addition to MHC Class II expression, stimulation of normal human PMN with IFNγ, or via CD11b induces CD80 and CD86 expression [8,11]. Human neutrophil presentation of superantigens is documented [2]. However, there are conflicting reports of neutrophils’ capacity to process and present tetanus toxoid on MHC Class II and activate T cells [1,2]. A capacity for extracellular antigen processing had been suggested as neutrophil gelatinase may cleave collagen to a Class II restricted immunodominant epitope [16]. Harding et al. demonstrated murine neutrophil Class I restricted antigen presentation [17] and additionally show that neutrophils processed phagocytosed bacteria via an alternate MHC Class I antigen-processing pathway, and such neutrophils may ‘regurgitate’ processed peptide into the extracellular space and this peptide may then bind MHC Class I on neighbouring cells (e.g. macrophages) for presentation to CD8 cells [17]. Human studies of antigen-specific neutrophil-T cell interactions are confounded by low frequency of antigen-specific T cells in peripheral blood, variation of MHC allele expression, which therefore impedes ideal peptide antigen identification. To overcome these complications, we have employed a murine model system. Moreover, despite the important role of the neutrophil in the pathogenesis of numerous inflammatory diseases, many of which are modelled in rodents, there is a relative paucity of information on murine Class II restricted antigen presentation. We, therefore, sought to investigate whether murine neutrophils present antigens via MHC Class II. 2 Materials and methods 2.1 Murine peritoneal neutrophil and macrophage separation Peritoneal exudate cells (PEC) were obtained as previously described [18]. Briefly, 6–8-week-old BALB/c mice (Harlan Olac, Bicester, UK) received 1.5 ml i.p. Brewers thioglycollate medium (Difco, East Molesey, UK) and 4 h later animals were sacrificed and cells harvested by peritoneal lavage. PEC were washed once then resuspended in 8 ml PBS, and layered over 3 ml Histopaque® 1083 (Sigma, Poole, UK), then centrifuged at 700 × g for 30 min at 20 °C. The enriched population of neutrophils contained within the cell pellet was washed twice in RPMI (Invitrogen, Paisley, UK) then resuspended at 2 × 106 ml−1 in medium supplemented with 2 mM l-glutamine, 100 IU penicillin, 100 μg/ml streptomycin (all Invitrogen) and 10% foetal calf serum (Harlan). Peritoneal macrophages were obtained by adhering PEC to a polystyrene flask for 3–4 h after which remaining adherent cells were incubated with ice cold PBS (without calcium or magnesium) at 4 °C for 10 min. Cells were harvested with a cell scraper, then washed in RPMI and resuspended at 2 × 106 ml−1. 2.2 Murine lymph node preparation DO11.10 mice, with CD4+ T cells specific for the OVA323–339 peptide in the context of the MHC Class II molecule I-Ad [19] were obtained originally from N. Lycke, University of Göteborg, Sweden. This T cell receptor is recognized by the KJ1.26 clonotypic antibody. Single cell suspensions of lymph nodes from DO11.10 mice were obtained by passing tissues through 100 μm Nitex membrane (Cadish, London, UK) using a 5 ml syringe plunger. To enrich for lymphocytes in lymph node cell suspensions, cells were resuspended at 4 × 106 ml−1 in complete RPMI, and cultured in 150 cm3 polystyrene flasks (Iwaki, Japan), for 2 h after which non-adherent cells were removed. Cells were washed three times in RPMI and resuspended at the desired cell concentration in complete RPMI with 25 mM HEPES. 2.3 Assessment of cell proliferation and cytokine production Cells were cultured under conditions indicated in figure legends. Where indicated, transwells (0.2 μm membrane, Fisher Scientific, UK) separated cell populations. Cell proliferation was assessed by incorporation of [3H]thymidine (0.5 μCi/well) for the last 18 h of culture. Cells were harvested using a Betaplate 96-well harvester (Wallac Oy, Turku, Finland) and [3H]thymidine incorporation measured on a Betaplate liquid scintillation counter (Wallac). Concentration of cytokines in supernatants were estimated by ELISA. Paired antibodies (Pharmingen, Oxford, UK) were used in accordance with the manufacturer's recommendations. All assays were optimised and validated prior to use. Maxisorp 96-well plates (Nunc) were coated with capture antibody in 0.1 M NaHCO3 overnight at 4 °C. Plates were blocked for 2 h; samples and standards added for 2 h, followed by biotinylated secondary antibody for 1 h, then ExtrAvidin peroxidase (Sigma) conjugate for 45 min. Plates were developed with TMB (3,3′,5,5′-tetra-methylbenzidine, KPL, Gaithersburg, USA), and then read at 630 nm with a Dynex MRX II plate reader and analysed using Dynex Revelation 3.2 software (Dynex, Middlesex, UK). The levels of cytokine in supernatants were calculated by comparison with recombinant cytokine standards (R&D Systems, Abingdon, UK). 2.4 Hybridoma cells The DO11.10-GFP hybridoma (kind gift of David Underhill, Department of Immunology, University of Washington, Seattle, Washington [20]) cells were grown in medium containing a 0.5 mg/ml geneticin. Co-cultures were established in which neutrophils or macrophages were incubated with 10 μg/ml OVA323–339 peptide and DO11.10-GFP hybridoma cells, in the ratio indicated in the figure legends. Cells were cultured for 14 h, washed, and then analysed by flow cytometry. 2.5 Flow cytometry Aliquots of 1 × 106 cells in 12 mm × 75 mm polystyrene tubes (Falcon BD, Oxford, UK) were resuspended in 100 μl FACS buffer (PBS, 2% FCS and 0.05% NaN3) containing Fc Block (Becton Dickenson, Oxford, UK) as well as appropriate antibody or isotype control. Neutrophils were identified with anti GR-1 (RB6-8C5, Caltag), macrophages with anti F4/80 (C1:A3-1, Caltag), and MHC Class II assessed (2G9, BD Biosciences), and T cells identified using anti CD4+ (RM4-5, BD) and a clonotypic antibody specific for the OVA323–339 TCR (KJ1.26). After washing, samples were analyzed using a FACSCalibur flow cytometer equipped with a 488 nm argon laser and a 635 nm red diode laser and analyzed using CellQuest software (both BD). 2.6 Statistical analysis Statistical analysis was performed on data as indicated in the figure legends, using Minitab software. 3 Results 3.1 Purification of murine neutrophils Neutrophils were purified from thioglycollate-induced peritoneal exudate cells (PEC), as there is a relative paucity of neutrophils in murine peripheral blood (data not shown). PEC were harvested 4 h post-thioglycollate administration, previously shown to be the peak of neutrophil recruitment, and after which an increasing proportion of macrophages are recruited [18]. Approximately 65% of thioglycollate-induced PEC were identified as neutrophils on the basis of GR1 staining (data not shown). PEC were further enhanced for neutrophils with a Ficoll gradient purification as previously described [18]. Following purification, samples were routinely 85% GR1 positive (Fig. 1a). Importantly, samples contained few macrophages (<4%) as determined by F4/80 staining (Fig. 1b). Cytospins of neutrophil preparations were prepared and demonstrated characteristic multilobed nuclei in >85% or cells (data not shown). Macrophage samples were 50–60% F4/80 positive cells (data not shown). Murine neutrophils, expressing GR-1, were found to express MHC Class II (Fig. 1c), albeit at low levels relative to F4/80 positive macrophages (Fig. 1d). 3.2 Murine neutrophils are able to present peptide antigen The murine T cell hybridoma, DO11.10-GFP, is activated by OVA323–339 peptide in the context of I-Ad, without any requirement for co-stimulation. The hybridoma does not respond to peptide alone (Fig. 2, ‘background’). Neutrophils or macrophages alone had no effect on GFP expression by DO11.10-GFP. However, neutrophils or macrophages loaded with OVA323–339 stimulated significant GFP expression by hybridoma cells when cultured at a 1:1 ratio. A dilution of macrophages was performed to estimate the contribution of contaminating macrophages to the results observed with neutrophil populations, demonstrating that at a ratio of 1 macrophage to 10 DO11.10-GFP cells, there was significantly less expression of GFP protein compared to that induced by neutrophils at a 1:1 ratio (Fig. 2). At a ratio of 1:100, there was no consistent GFP expression detected (data not shown). As there were <5% F4/80 positive macrophages contaminating the neutrophil samples, these titration experiments indicated that only a small proportion of the T cell hybridoma activation observed in the ‘neutrophil presentation’ assays could be attributable to macrophage contamination. 3.3 Neutrophils can drive naïve T cell proliferation and cytokine production To investigate neutrophils antigen presentation in a co-stimulation dependent system, and the ability of neutrophils not only to present but also to prime T cells, neutrophils were co-cultured with primary T cells. Lymph nodes were obtained from DO11.10 transgenic BALB/c mice. Single cell suspensions of lymph nodes were allowed to adhere to plastic for 3 h to remove the majority of adherent antigen presenting cells, after which the population was routinely 70% CD3 positive (data not shown). The presence of remaining endogenous APCs in the lymph node facilitated a strong response to OVA323–339 peptide alone (Fig. 3a, c and e). Therefore, neutrophils and macrophages were pulsed with OVA323–339 for 2 h then washed three times to remove unbound peptide prior to the addition of transgenic T cells. Cells were co-cultured for 72 h at the ratios indicated in the figure legends. Neutrophils pulsed with OVA323–339-induced proliferation (Fig. 3b), IL-2 (Fig. 3d) and IFNγ (Fig. 3f) production. Low levels of IL-5 were detected in some experiments (data not shown). The neutrophil-induced responses were less than those evoked by macrophages at the same cell ratio. A titration of macrophages was performed to give an estimate of the contribution of contaminating cells, demonstrating that only a small proportion of neutrophil-induced T cell responses could be attributable to contaminating macrophages. 3.4 Cell contact dependence of neutrophil T cell interactions To determine the relative contribution of cell contact dependent mechanisms and soluble factors in neutrophil-T cell interactions, cells were separated by 0.2 μm membranes placed between the cell populations, allowing diffusion of soluble proteins but preventing cell contact. The responding OVA323–339-specific T cells were not a 100% pure T cell population and contained some APCs, demonstrated by the lymph node response to OVA323–339 alone (Fig. 4a and b ‘OVA’). Thus, the transwells permitted investigation of whether lymph node-derived APCs were presenting soluble proteins derived from dead or degraded neutrophils. Bone marrow-derived dendritic cells (DC) were used as a positive control. Free OVA323–339 or ConA crossed the transwell membrane and stimulated cell proliferation and cytokine production (Fig. 4a). However, when antigen pulsed neutrophils were separated from the lymph node cells by the transwell, antigen-specific proliferation and cytokine production was reduced (Fig. 4a and b). Importantly, antigen pulsed cell cultures did not release sufficient free antigen to stimulate a response from the OVA323–339-specific T cells below (Fig. 4a). Thus, the antigen-specific T cell proliferation and cytokine production induced by neutrophil antigen presentation is cell contact dependent. 4 Discussion Data presented here demonstrate that murine neutrophils are indeed capable of directly presenting antigen on Class II MHC and activating T cells independently of other APCs. The DO11.10-GFP hybridoma cells respond exclusively in a Class II restricted manner, and our data demonstrate that neutrophils, rather than contaminating cells, were mediating this response. At comparable cell ratios, neutrophil-mediated T cell activation was lower than that mediated by macrophages, commensurate with reduced expression of MHC Class II by neutrophils. Although neutrophils have been reported to express co-stimulatory molecules, we detected only low levels of CD40 and were unable to consistently detect CD80 or CD86 in our neutrophil preparations (in contrast to the F4/80 positive macrophages which expressed CD40, CD80 and CD86; data not shown). Thus it is possible that: co-stimulatory molecule expression below the limits of detection of methods used here is sufficient for inducing a response, that T cells are responding in a co-stimulation independent manner, or that co-stimulation has occurred via alternative molecules. Neutrophil antigen processing capacity was investigated using whole ovalbumin protein compared to OVA323–339. These studies indicated that these neutrophils did not process the ovalbumin effectively, demonstrated by a lack of response by DO11.10-GFP hybridoma (data not shown). An MHC Class I processing pathway within neutrophils was elegantly defined with E. coli transfected to express a MHC Class I restricted epitope of OVA. This demonstrated that neutrophils are capable of non-classical antigen processing for presentation on MHC Class I [17]. Although neutrophil enzyme mediated degradation of collagen generates a Class II epitopes (14), there is no current evidence of any Class II processing pathway. It has been proposed that as macrophage phagosomes are adequate for Class II antigen processing [21], therefore a parallel system may exist in neutrophils. A definitive investigation of murine neutrophil MHC Class II antigen processing awaits transfection of bacteria with OVA323–339, thus elucidating the effects on antigen processing of concomitant neutrophil activation by bacteria. There is increasing evidence documenting the potential for neutrophils to acquire macrophage [22] or dendritic cell (3) characteristics, often in response to in vitro conditions. We document in vitro antigen presentation by in vivo activated neutrophils. It will be of interest to delineate whether such neutrophil behaviour can be observed strictly in vivo. Physiologically, the location of such neutrophil-T cell interactions is likely to be restricted to sites of chronic or established inflammation, for example persistent bacterial infection [15]. Neutrophils have been shown to degrade proteins to release antigens recognized by auto reactive T cells [16], and such antigens may then be transported to the draining LN by migrating DCs. Due to the neutrophils’ unique proteolytic activities, this may represent a mechanism of presentation of novel or cryptic antigens. Furthermore, there is evidence that neutrophils may passively acquire membrane proteins, including MHC, from other cells [23]. In our GFP-hybridoma studies (Fig. 2), this seems unlikely as the neutrophils were harvested from the peritoneum before there was significant in vivo macrophage migration, and immediately purified in vitro. However, this phenomenon may have contributed to the neutrophil mediated T cell activation in co-culture (Fig. 3). There are overall fewer studies of murine neutrophils compared to human neutrophils, which is likely to be a reflection of the ease of acquiring human neutrophils that constitute 70% of circulating leucocytes, compared to limited quantity of murine peripheral blood available and the relative paucity of neutrophils, contributing only 20% of circulating leukocytes in rodents. These studies were only possible through use of rodent cells, as the TCR transgenic T cells absolutely control antigen specificity of response, which was central to this investigation of neutrophil-T cell interactions. However, the significantly different proportion of neutrophils in murine and human peripheral blood raises questions as to whether murine and human neutrophils perform parallel functions as is assumed. Our data document a novel, Class II restricted antigen presenting function of murine neutrophils, confirming findings in other human systems, and expanding our understanding of the role of the neutrophil in the adaptive immune response.

Eur_J_Clin_Pharmacol-3-1-2039832

Evaluation of limited sampling strategies for tacrolimus

Objective In literature, a great diversity of limited sampling strategies (LSS) have been recommended for tacrolimus monitoring, however proper validation of these strategies to accurately predict the area under the time concentration curve (AUC0–12) is limited. The aim of this study was to determine whether these LSS might be useful for AUC prediction of other patient populations. Introduction The calcineurin-inhibitor tacrolimus, used widely after organ transplantation, has a narrow therapeutic index and highly variable pharmacokinetic characteristics. Close monitoring of the drug concentration is required to achieve an optimum efficiency by minimizing the risk of subtherapeutic and toxic blood concentrations. Efficacy and side effects of tacrolimus are highly correlated with the area under the time concentration curve (AUC0–12) [1]. Elevated tacrolimus concentrations may lead to severe side effects such as nephrotoxicity, neurotoxicity and hyperglycaemia [2–4], while subtherapeutic tacrolimus concentrations increase the risk of transplant rejection enormously [5–7]. The most exact way to monitor the total tacrolimus exposure is by creating 12-h pharmacokinetic profiles, which implies that the tacrolimus concentration should be measured at at least six different time points. The AUC0–12 can then be calculated according to the trapezoidal rule using the tacrolimus concentrations measured at different time points. Since recording a complete 12-h pharmacokinetic profile for every patient is not feasible in clinical practice, traditionally many transplant centres have used tacrolimus trough (C0) concentrations to estimate the tacrolimus exposure. Although tacrolimus C0 concentrations are generally considered to be a good indication of the total systemic drug exposure [1, 8], its usefulness in differentiating graft rejection episodes from nephrotoxicity has been questioned [6, 9–11]. Recently, the correlation between individual tacrolimus concentrations and AUC0–12 has been studied in kidney [12–18], liver [19], heart [20, 21] and lung [22] transplant recipients. In these studies, a poor association was found between the tacrolimus C0 concentrations and the AUC0–12, while tacrolimus concentrations measured at other time points showed much better correlations with the AUC0–12. Additionally, strategies have been developed that included a limited number of sampling time points within a short time post-dose, the so-called limited sampling strategies (LSS). Several two- and three-time-point sampling strategies showed a high correlation with the AUC0–12 in published studies and were able to predict the AUC0–12 more accurately than the C0 concentration alone [12, 15–18, 20, 22]. Based on the number of published studies regarding LSS for tacrolimus, there seems to be a growing interest for non-C0 concentration measurements as an indicator of between-patient variability and as a guide for dose adjustments. Most of these studies recommend different LSS, but these strategies have not been validated with an independent population. Ting et al. [23] recently reported that validation of the different LSS with an independent transplant population is an absolute prerequisite. The question was whether the LSS described in literature could be used in other centres with different populations. Predictive value of LSS from literature was evaluated using our own renal post-transplant group of 37 patients with known AUCs. Also the predictive value of trough levels (C0 and C12) determined in our own population was investigated. Materials and methods Patient population In total, 37 Caucasian renal transplant recipients for whom a complete 12-h time tacrolimus concentration curve had been determined in a past clinical trial were included in this study (Table 1). The transplant recipients underwent a renal transplantation at least 1 year ago. Patients taking medication known to interact with tacrolimus, who suffered from gastro-intestinal or liver disease, pre-transplantation diabetes mellitus or other disorders that could have altered the absorption of tacrolimus were excluded from this study as illustrated in Table 1. Table 1Demographic characteristics of renal-transplant recipientsDemographic characteristicsPatients (n = 37)Gender (male/female)24/13Age (years, mean ± SD)51.3 ± 10.9Length (cm, mean ± SD)174 ± 8.4Weight (kg, mean ± SD)77.4 ± 13.5Body mass index (kg/m2, mean ± SD)25.6 ± 3.42Primary kidney disease (n) Glomerulonephritis1 Chronic pyelonephritis2 IgA nephropathy4 Hypertensive nephropathy7 Diabetes mellitus nephropathy0 Polycystic kidney disease8 Unknown4 Other11Transplantation number (n) First30 Second6 Third or more1Tacrolimus mono therapy (n)29Tacrolimus dose (mg kg−1 day−1, mean ± SD)0.054 ± 0.029C0 (ng/mL, mean ± SD)6.59 ± 1.39AUC0–12 (ng×h/mL, mean ± SD)122.5 ± 31.1Cmax (ng/mL, mean ± SD)20.9 ± 6.5Tmax (h, mean ± SD)1.24 ± 0.43Use of azothioprine/MMF/rapamycine/steroids3/4/0/2Time since transplantation (days, mean and range)1,542 (453–4,128)Haemoglobin (mmol/L, ref. M: 8.2–11.0, F: 7.3–9.7)8.52 ± 0.83Haematocrit fraction (ref. M: 0.41–0.52, F: 0.36–0.48)0.41 ± 0.04ALAT (units/L, ref. M: <45, F: <35)24 ± 13ASAT (units/L, ref. M: <35, F: <30)17 ± 10Serum albumin (g/L, ref. 34–45)37.0 ± 3.84Serum creatinine (μmol/L, ref. M: 71–110, F: 53–97)128 ± 29Creatinine clearance (Cockcroft-Gault; mL/min, ref. 90–140)58.4 ± 26.6Ref. are the reference values applied in the Clinical Chemistry and Haematology Laboratory of the University Hospital in Maastricht. M Male, F female, MMF mycophenolate mofetil Prior to the blood sample collection, there was no tacrolimus dose change for at least 1 week. After overnight fasting, the blood samples were collected immediately before (C0) and 0.5 (C0.5), 1 (C1), 2 (C2), 3 (C3), 4 (C4), 5 (C5), 7.5 (C7.5) and 12 (C12) h after the morning tacrolimus administration. Patients were not allowed to take food until 1 h after ingesting the tacrolimus dose and were advised to avoid grapefruit intake to prevent alterations in the tacrolimus metabolism. Demographic as well as clinical data were determined at the time of recording the 12-h time tacrolimus concentration curve. The study was performed in accordance with the Declaration of Helsinki and its amendments. The protocol was approved by the local Medical Ethics Committee and written informed consent for participation in this study was obtained from all patients. Determination of tacrolimus concentrations The tacrolimus blood concentrations were determined in ethylene diamine tetra-acetic acid (EDTA) whole blood, using a method based on high-pressure liquid chromatography tandem mass spectrometry (HPLC-MS/MS). The assay is linear from 1 to 300 μg/l. Intra-assay precision and accuracy were 3.4, 2.2, and 3.0% and 102, 94 and 94% respectively at 3.04, 6.23 and 13.0 μg/l (n = 6). Inter-assay precision and accuracy were 8.2, 5.2, and 4.6% and 102, 94 and 93% (n = 9) respectively. Lower limit of quantification was 1.0 μg/l. The laboratory participates in the International Tacrolimus Proficiency Testing Scheme. Limited sampling strategies investigated In our opinion, a suitable limited sampling strategy for tacrolimus should consist of two or three time concentration points within a short time post-dose (≤4 h) including a trough level. We selected 24 LLS from the literature [12, 15, 16, 18, 20, 22]. Eighteen of these strategies were based on regression analysis [12, 16, 18, 20, 22], and six other strategies were based on Bayesian fitting [15]. Also strategies based on the tacrolimus C0 and C12 concentrations were developed for our own renal transplant patient population and compared with above-mentioned LSS. Pharmacokinetics and statistical analysis The area under the tacrolimus time concentration curve (AUC0–12) was calculated from the time versus tacrolimus concentration plot using the linear trapezoidal rule in MWPharm 3.50 (Mediware, Groningen, The Netherlands). The predicted AUC0–12 (AUCpred), calculated with the 24 different LSS, were validated by determining the predictive performance as described by Sheiner and Beal [24]. The percentage of the prediction error (PE) and the percentage of the absolute prediction error (APE) are parameters often used for validation in LSS [12, 14–16, 18, 20, 22]. Given the high pharmacokinetic variability, an APE of less than 15% was considered clinically acceptable [16, 25, 26]. Prediction bias was measured as a percentage of the prediction error [PE (%)] using the following formula: Prediction precision was measured as a percentage of the APE using the following formula: The variance in the strength of association between the AUCpred and the AUCactual was reflected by the linear regression coefficient of multiple determination (R2). All values are expressed as mean ± SD. All statistical analyses were performed with use of SPSS 12.0 software for windows (Chicago, IL, USA). Results Evaluation of predictive performances of the limited sampling strategies Table 2 shows an overview of the studies describing the LSS evaluated in the present study. The regression equations and the R2 found by the investigators for the evaluated LSS are summarised in Table 3. Table 4 describes the R2, which represents the association between AUCpred and AUCactual and the calculated PE and APE of the 24 evaluated LSS for our 37 pharmacokinetic profiles. Thirteen of the 18 LSS (72%) based on regression analysis had a predictivity of >90%. Additionally, all except three of the LSS examined gave a better prediction of the complete AUC0–12 in comparison with LSS based on a trough concentration C0 and C12 (mean 62%). Predictivity of all six LSS based on Bayesian fitting was <90% (mean 66.8%). Additionally, Fig. 1 illustrates an overview of the performances of the 26 LSS evaluated in our well-characterized population of renal transplant recipients. Table 2Overview of the characteristics of transplant patients included in the studies that described limited sampling strategiesStudyTransplanted organNumber of patientsaNumber of AUC0–12 curves for validation (Ib/NIc)Analytical methoddTime since transplantationeInclusion criteriafWong et al. [16]Kidney180/18Imx II2.5 years1,2Aumente Rubio et al. [20]Heart220/25Imx<1 year–Pisitkun et al. [18]Kidney150/15Imx II8.7 months1,2,3Armendariz et al. [12]Kidney2213/14ImxUnknown–Scholten et al. [15]Kidney4364/20ImxDiffersg2Ragette et al. [22]Lung150/31Imx7.3 months–aNumber of transplant patients used in the included study for both developing and validating the limited sampling strategies.bNumber of AUC0–12 used for developing the limited sampling strategies.cNumber of independent (I) and dependent (NI) AUC0–12 used in the study to validate the created limited sampling strategies.dThe analytical method used to determine the whole blood tacrolimus concentration.eThe mean time after transplantation.fThe inclusion criteria used for the transplant patients in the different studies. 1 Tacrolimus administrated when patients were in the fasting state, 2 patients selected for using no interfering medication with tacrolimus, 3 patients selected with a normal liver function test.gTwenty-two pharmacokinetic profiles were obtained within 2 weeks after transplantation, and 42 pharmacokinetic profiles were obtained between 6 and 52 weeks after transplantation.Table 3Overview of limited sampling strategies and their reported coefficients of correlation (R2) with the complete tacrolimus AUC0–12EquationTime pointsRegression equationsR2Ref1.C00.542.C120.793.aC0, C2, C40.93[16]4.aC2, C40.93[16]5.aC0, C2, C40.97[20]6.C0, C40.95[20]7.C0, C1, C20.93[18]8.aC0, C1, C40.97[18]9.aC0, C2, C40.97[18]10.aC1, C2, C40.96[18]11.C0, C10.91[18]12.aC0, C40.95[18]13.a C1, C40.95[18]14.aC2, C40.96[18]15.a,bC0, C1, C40.97[18]16.a,bC0, C2, C40.96[18]17.C0, C1, C40.97[12]18.C0, C1, C3Bayesian estimation of the actual AUC0–120.97[15]19.C0, C2, C3Bayesian estimation of the actual AUC0–120.96[15]20.C0, C2, C4Bayesian estimation of the actual AUC0–120.97[15]21.C0, C2Bayesian estimation of the actual AUC0–120.94[15]22.C0, C3Bayesian estimation of the actual AUC0–120.96[15]23.C0, C4Bayesian estimation of the actual AUC0–120.95[15]24.aC0, C2, C40.98[22]25.C0, C40.96[22]26.aC2, C40.94[22]Limited sample strategies derived from the linear trapezoidal rule and the complete 12-h AUC.aLimited sampling strategies that are able to predict 90% of complete AUC0–12 of the renal transplant recipients within the absolute prediction error (APE) of 15%.bLimited sample strategies derived from the linear trapezoidal rule and the actual AUC0–12.Table 4Evaluation of predictive performance of limited sampling strategies to estimate the complete AUC0–12 in the 37 renal transplant recipientsEquationTime pointsR2Mean PE (%)Mean APE (%)≤15%a23.bC0, C40.760−14.9 ± 13.8 (−46.0–33.2)17.9 ± 9.43 (1.12–46.0)13 (35%)22.bC0, C30.779−11.5 ± 14.0 (−41.9 to 33.1)15.7 ± 8.83 (2.0–41.9)21 (57%)1.C00.5362.11 ± 14.8 (−27.1 to 24.4)12.3 ± 8.22 (0.7–27.1)22 (59%)11.C0, C10.7036.58 ± 14.8 (−26.5 to 43.7)12.6 ± 10.1 (0.1–43.7)24 (65%)2.C120.809.56 ± 11.6 (−12.7 to 29.9)12.0 ± 8.97 (0.3–29.9)24 (65%)19.bC0, C2, C30.502−4.44 ± 17.4 (−45.3 to 50.6)13.7 ± 11.4 (0.4–50.6)25 (68%)20.bC0, C2, C40.537−5.11 ± 16.3 (−43.1 to 50.3)12.9 ± 10.4 (0.2–50.3)28 (76%)18.bC0, C1, C30.5259.95 ± 19.4 (−29.7 to 88.8)13.1 ± 17.4 (0.4–88.8)30 (81%)25.C0, C40.911−7.83 ± 6.36 (−21.3 to 2.4)8.08 ± 6.02 (0.1–21.3)30 (81%)7.C0, C1, C20.8692.35 ± 9.96 (−17.2 to 27.3)8.03 ± 6.22 (0.0–27.3)31 (84%)6.C0, C40.896−5.97 ± 6.71 (−20.1 to 4.7)6.63 ± 6.04 (0.6–20.1)31 (84%)21.bC0, C20.802−3.69 ± 10.2 (−19.6 to 18.6)9.10 ± 5.67 (0.4–19.6)31 (84%)17.C0, C1, C40.9435.91 ± 7.06 (−8.8 to 26.3)7.02 ± 5.93 (0.2–26.3)33 (89%)15.cC0, C1, C40.9345.00 ± 7.28 (−9.8 to 25.8)6.81 ± 5.57 (0.2–25.8)34 (92%)14.C2, C40.9642.28 ± 6.58 (−17.1 to 16.1)5.45 ± 4.24 (0.7–17.1)35 (95%)24.C0, C2, C40.941−4.81 ± 5.26 (−17.3 to 2.8)5.32 ± 4.73 (0.1–17.3)35 (95%)13.C1, C40.9736.30 ± 4.84 (−5.9 to 17.8)6.68 ± 4.28 (0.3–17.8)36 (97%)8.C0, C1, C40.9673.37 ± 5.21 (−5.2 to 17.7)4.87 ± 3.80 (0.2–17.7)36 (97%)9.C0, C2, C40.9620.10 ± 6.37 (−16.7 to 14.7)4.71 ± 4.22 (0.3–16.7)36 (97%)26.C2, C40.9593.38 ± 5.24 (−7.6 to 15.5)5.20 ± 3.37 (0.0–15.5)36 (97%)10.C1, C2, C40.9763.07 ± 5.40 (−14.9 to 13.2)4.99 ± 3.64 (0.1–14.9)37 (100%)16.cC0, C2, C40.953−1.58 ± 5.29 (−14.9 to 10.1)4.00 ± 3.75 (0.0–14.9)37 (100%)12.C0, C40.9303.55 ± 6.30 (−9.8 to 14.3)6.29 ± 3.46 (0.1–14.3)37 (100%)4.C2, C40.963−1.66 ± 4.99 (−12.0 to 14.3)4.13 ± 3.20 (0.2–14.3)37 (100%)5.C0, C2, C40.9591.33 ± 5.24 (−11.8 to 14.0)4.22 ± 3.32 (0.5–14.0)37 (100%)3.C0, C2, C40.965−0.20 ± 4.79 (−10.4 to 13.7)3.64 ± 3.06 (0.2–13.7)37 (100%)aNumber and percentage of calculated AUC0–12 with a prediction error within 15%.bBayesian estimation of the actual AUC0–12.cLimited sample strategies derived from the linear trapezoidal rule and the actual AUC0–12.Fig. 1An overview of the predictive performances of the limited sampling strategies published using 37 different pharmacokinetic profiles recorded in our renal transplant recipients. The number of limited sampling strategies categorized into the percentage of AUC0–12 within a prediction error of <15% is plotted on the x-axis, the number of limited sampling strategies is plotted on the y-axis Discussion Our results confirm the results of several other studies [12, 14–16, 18, 20, 22] that trough concentrations C0 and C12 have a lower predictive value for the complete 12-h AUC than almost all other studied LSS. The predictivity of LSS based on Bayesian estimation of the AUCactual was lower than the LSS based on regression analysis. Therefore a trough level and one or two time points in the early phase (≤4 h) post-dose seem not to be sufficient for a Bayesian estimation strategy to fit correctly most of the AUC0–12 and thus predict the complete AUC0–12 reliably. The differences in variability and shape between the curves of post-transplant recipients combined with just two or three sample points might have been caused by the large differences found between the AUCpred calculated according to the Bayesian estimation strategy and the complete AUCactual. In contrast to most studies that describe LSS for tacrolimus in literature, we used an HPLC-MS/MS assay to determine the tacrolimus concentration. Because there seems to be a fixed difference of about 15% between the immunoassay and the HPLC-MS/MS, the prediction will change proportionally, and the predictivity of the LSS will be the same. Also potential interfering drug-drug interactions will have an equal influence on the different tacrolimus concentrations, which consequently has no effect on the predictivity of the different LSS. Ting et al. [23] recently suggested that LSS should only be applied on transplant patient populations that are comparable with the transplant patient population that was used to develop the LSS. However, the renal transplant patient group examined in the present study was not exactly comparable with the transplant patient populations in which the equations for the LSS were developed. For example Aumente Rubio et al. [20] and Ragette et al. [22] used heart and lung transplant recipients respectively to develop and validate their LSS. Despite the fact that the LSS were developed with the pharmacokinetic profiles of patients who underwent a different kind of transplantation, Eqs. (5), (24) and (26) were able to predict at least 90% of the AUC0–12 within an APE of 15%, which suggests that these LSS are more robust than expected by Ting et al. [23]. Even though LSS gave a better reflection of the tacrolimus exposure, they are currently not often applied by clinical transplant practioners, possibly for logistical and financial reasons. In conclusion, after validating several LSS from the literature, the present study indicates that all but three LSS gave a better prediction of the complete AUC0–12 than the trough concentrations C0 or C12. Moreover, LSS could produce satisfactory predictions for AUC0–12 recorded in an independent renal transplant patient population, although further evaluation of their reliability is necessary.

Appl_Microbiol_Biotechnol-3-1-1914294

Functional analysis of genes involved in the biosynthesis of isoprene in Bacillus subtilis

In comparison to other bacteria Bacillus subtilis emits the volatile compound isoprene in high concentrations. Isoprene is the smallest representative of the natural product group of terpenoids. A search in the genome of B. subtilis resulted in a set of genes with yet unknown function, but putatively involved in the methylerythritol phosphate (MEP) pathway to isoprene. Further identification of these genes would give the possibility to engineer B. subtilis as a host cell for the production of terpenoids like the valuable plant-produced drugs artemisinin and paclitaxel. Conditional knock-out strains of putative genes were analyzed for the amount of isoprene emitted. Differences in isoprene emission were used to identify the function of the enzymes and of the corresponding selected genes in the MEP pathway. We give proof on a biochemical level that several of these selected genes from this species are involved in isoprene biosynthesis. This opens the possibilities to investigate the physiological function of isoprene emission and to increase the endogenous flux to the terpenoid precursors, isopentenyl diphosphate and dimethylallyl diphosphate, for the heterologous production of more complex terpenoids in B. subtilis. Introduction Isoprene (2-methyl-1,3-butadiene; I) or actually isopentenyl diphosphate (IDP; II) is the general precursor of all terpenoids, that represent a very diverse class of natural products. Two evolutionary distinct routes occur in nature for the biosynthesis of IDP. The spread of the two pathways is well investigated for organisms with a sequenced genome (Lange et al. 2000; Boucher and Doolittle 2000; Rohdich et al. 2001). In eukaryotes and archaea IDP and its isomer dimethylallyl diphosphate (DMADP; III) are formed via the mevalonate pathway (reviewed by Kuzuyama 2002). This pathway is well studied and for many organisms the enzymes are characterized and the encoding genes identified. More recently, another pathway to IDP was discovered in some eubacteria and in plastids of higher plants, that proceeds via the intermediate methylerythritol phosphate (MEP) (Rohmer et al. 1993; Rohmer 1999). Most Gram-negative bacteria, including Escherichia coli, use the MEP pathway and for E. coli this is now well documented. For Gram-positive bacteria the situation is less clear. Gram-positive cocci have been reported to only possess genes for the mevalonate pathway (Wilding et al. 2000). In Streptomyces aeriouvifer both pathways are found (Seto et al. 1996). In Bacillus subtilis, being regarded as the prototype for Gram-positives, only homologues of the genes for the MEP pathway are present (Wagner et al. 2000). For E. coli the complete MEP pathway has been elucidated and the genes involved have been identified and the corresponding enzymes described (reviewed by Eisenreich et al. 2004). The pathway exists of seven subsequent enzymatic steps (Fig. 1). The first reaction of the MEP pathway is catalyzed by a transketolase (DXS) and concerns the condensation of pyruvate (IV) and d-glyceraldehyde-3-phosphate (V) to 1-deoxy-d-xylulose 5-phosphate (VI) (Sprenger et al. 1997; Lois et al. 1998). 1-deoxy-d-xylulose is a branching intermediate in the biosynthesis of isoprenoids, thiamine (vitamin B1) (White 1978; David et al. 1981, 1982), and pyridoxol (vitamin B6), respectively (Hill et al. 1989). The second step is catalyzed by the IspC protein, initiating a rearrangement in the carbon skeleton followed by a reduction in the keto function of (VI), and finally delivering 2C-methyl-d-erythritol 4-phosphate (Takahashi et al. 1998). Five other subsequently acting enzymes are needed for the synthesis of IDP, including a phosphocytidyl transferase (IspD), a kinase (IspE), a cyclophosphate synthase (IspF), and two reductases (IspG, IspH) (Fig. 1). Fig. 1Methylerythritol phosphate pathway of isoprenoids [Dxs (1), 1 deoxy-d-xylulose 5-phosphate synthase; IspC (2), 2C-methyl-d-erythritol 4-phosphate synthase; IspD (3), 4-diphosphocytidyl-2C-methyl-d-erythritol 4-phosphate synthase; IspE (4), 4-diphosphocytidyl-2C-methyl-d-erythritol kinase; IspF (5), 2C-methyl-d-erythritol 2,4-diphosphate synthase; IspG (6), 2C-methyl-d-erythritol 2,4-cyclodiphosphate; IspH (7), 1-hydroxy-2-methyl-butenyl 4-diphosphate reductase; Idi (8), isopentyl diphosphate isomerase; IspA (9), farnesyl diphosphate synthase]. The putative genes are mentioned in italics for every biosynthetic step, including the essential gene YpgA, which is indicated with an *. Numbering of compounds and biosynthetic steps refers to the text Bacteria use the precursors IDP and DMADP for the synthesis of several compounds including the side chains of ubiquinone or menaquinone. Next to that, several bacterial species use these precursors to synthesize isoprene, that is emitted to their environment (Kuzma et al. 1995). B. subtilis emits isoprene in high levels compared to other bacterial species as has been described for B. subtilis 6051, B. subtilis 23059, and B. subtilis 23856 (Kuzma et al. 1995). Using a bioreactor system, the emission of isoprene by B. subtilis 6051 was found to occur in three phases during the growth curve (Wagner et al. 1999). The phases correspond with respectively, glucose metabolism and secretion of acetoin, catabolism of acetoin, and the early stage of sporulation. The general laboratory strain B. subtilis 168 showed another pattern of isoprene emission, lacking phases 2 and 3 (Fall and Copley 2000). The function of the emitted isoprene has been postulated as being a signal molecule in the natural environment of the microorganism. Another possible explanation for the emitted isoprene is the efflux as an overflow metabolite in the bacterial pathway to isoprenoid structures (Fall and Copley 2000). The uptake of isoprene by microorganisms present in soil samples has been described as a sink of atmospheric isoprene (Cleveland and Yavitt 1998). However, there is no full evidence supporting one of the hypotheses mentioned. Isoprene originates from DMADP. The conversion of DMADP in isoprene is known to be an enzymatic process in the poplar tree from which a gene has been isolated and characterized (Miller et al. 2001; Schnitzler et al. 2005). Attempts to prove the enzymatic conversion to isoprene in B. subtilis suggested the involvement of an enzymatic step. Enzymatic activity has been partially purified, but the enzyme turned out to be very labile (Sivy et al. 2002). Until now the encoding gene could not be identified. A search for a homologue protein of the isoprene synthase from poplar tree in the genome of the B. subtilis did not yield a candidate. Based on homology with known genes, mostly from E. coli, candidate genes for the other enzymatic steps in the MEP pathway in B. subtilis have been suggested. Identification of the genes involved in this pathway should shed some light on the function of isoprene synthesis in Bacillus physiology. Due to the essentiality of the products downstream of the isoprenoid biosynthesis, like menaquinone, it is not remarkable that the candidate genes, except for the isomerase, have been shown to be essential genes for the survival of the bacterial cell (Kobayashi et al. 2003). In the construction of the BFAN collection these knock-out strains were not viable and therefore conditional mutants have been made using a pMUTIN3-vector. The expression in these strains can be regulated by the isopropyl-beta-d-thiogalactopyranoside (IPTG) inducible Pspac promoter (Yansura and Henner 1984; Vagner et al. 1998). Using this mutant system, expression levels can be regulated to a minimum for survival by decreasing the amounts of IPTG. In that way, the function of the genes can be investigated. Identification of the involvement of certain genes in the isoprenoid biosynthetic pathway, may create possibilities to redirect the metabolic flux toward IDP. The possibilities of pathway engineering have shown their strength already in other microorganisms, like E. coli and Saccharomyces cerevisiae. The metabolic flux was dependent mainly on the deoxyxylulose 5-phosphate synthase (DXS), the isopentenyl diphosphate isomerase (Idi), and the prenyltransferase delivering farnesyl diphosphate (IspA) (Harker and Bramley 1999; Kim and Keasling 2001; Kajiwara et al. 1997; Martin et al. 2003). By up-regulation of the MEP pathway, B. subtilis can be developed as an interesting host organism for the production of complex terpenoid compounds, such as the valuable drugs artemisinin and paclitaxel. In the present study we describe the functional analysis of genes in B. subtilis putatively encoding enzymes involved in the biosynthesis of the terpenoid precursor isopentenyl diphosphate. We used quantitative isoprene emission of different well controllable knock-out Bacillus strains in their environment to determine the function of the tested genes. Materials and methods Bacterial strains and media Bacterial strains used are listed in Table 1. B. subtilis strains 6051, 23059 and 23856 were obtained from the American Type Culture Collection (ATCC) (Rockville, USA). B. subtilis mutant strains were obtained through a chromosomal integration of pMUTIN3 derived plasmids (Vagner et al. 1998). Experiments were performed using Luria-Bertani (LB)-medium containing 1% bacto-tryptone, 0.5% yeast extract, and 0.5% NaCl (all purchased from Duchefa, Zwijndrecht, the Netherlands). If required, medium was supplemented with erythromycin (2 μg/ml; Duchefa, the Netherlands) or the specified concentration of isopropyl-β-d-thiogalactopyranoside (IPTG; Duchefa, the Netherlands). For growing on plates the medium was solidified with 1.5% agar. Table 1Putative genes for the MEP pathway in the genome of B. subtilis and the percentage of identities with known proteins from E. coli, the essentiality of each gene, the corresponding BFAN mutant strain (explained in detail in the text, with numbering of the respective enzymatic steps)EnzymeE. coliB. subtilisEssential geneMutant strainDxs (1)dxsdxs/yqiE (43%)+168IyqiEIspC/Dxr (2)dxryluB (43%)+168IyluBIspD (3)ygbPyacM (36%)+168IyacMIspE (4)ychByabH (27%)+168IyabHIspF (5)ygbByacN (56%)+168IyacNIspG (6)gcpEyqfY (46%)+168IyqfYIspH (7)lytByqfP (35%)+168IyqfPIdi Iidi––Idi II (8)–ypgA (39%)a–168ypgAIpsA/FPPS (9)ispAyqiD (43%)+168IyqiDaHomologue of Streptomyces protein (Kaneda et al. 2001) Detection of bacterial isoprene emission A single colony of the different B. subtilis strains was transferred from a plate to 10 ml Luria-Bertani (LB) medium (if required supplemented with 2 μg/ml erythromycin and 100 μM IPTG) and grown over night (37°C; 300 rpm). Before the inoculation (1:100) of fresh LB medium containing different specified concentrations of IPTG, cells were gently washed three times with fresh LB-medium without IPTG by resuspending and centrifugating. Isoprene accumulation was measured on-line by sampling every 15 min, during a period of nine hours, 15 ml of the air above 50 ml bacterial culture, growing (37°C; 300 rpm) in a 500 ml Erlenmeyer air tight flask (CBN, the Netherlands), and transferring into a gas-chromatography system suitable for the sensitive detection of isoprene (Syntech Spectras GC955 series 601, Synspec BV, the Netherlands) (Loreto and Delfine 2000). The air was pumped through a Tenax GA trap, desorbed at 180°C and transferred to an AT 5 column under a flow of 2.5 ml/min nitrogen (3.7 bars; quality 5). The temperature program used was 3 min at 50°C followed by an increase in temperature to 70°C at 5 min; kept at this temperature until 12 min and than lowered to 50°C again. The isoprene present was detected by photo ionization at 10.6 eV. The gas chromatograph was calibrated using the dynamic gas dilution principle with several concentrations of gaseous isoprene using liquid isoprene (Sigma, USA) diluted in methanol and evaporated with a gas dilutor (MK5, MCZ Umwelttechnik, Germany). During the isoprene detection the growth of the bacterial culture was determined by measuring the optical density at 600 nm (OD600 nm) every hour. Results Isoprene emission in wild-type Bacillus strains The emission of isoprene from B. subtilis wild type strains 168, 6051, 23059, and 23856 was investigated. For all four strains isoprene accumulated in the logarithmic phase of growth, leading to a high increase in isoprene in the flask. The absolute maximum of wild type strains varied between 400 and 700 μg/m3. Starting from the late logarithmic phase to the beginning of the stationary phase the amount of isoprene in the flask decreased slowly. Figure 2 shows the accumulation of isoprene during the growth of the wild type strains 168 and 6051. B. subtilis 6051 reached higher levels of isoprene, but this is probably not caused by a higher production rate rather than the amount of cells present, as reflected by the optical density of these strains. Corrected for the amount of cells, determined as the optical density at 600 nm, B. subtilis 168 accumulated at maximum 148 μg/ m3 / OD600 nm compared to 166 μg/ m3 / OD600 nm for B. subtilis 6051. The other two wild type strains tested, 23059 and 23856, showed a similar accumulation of isoprene as the strains described before. Corrected for the amount of cells, 23059 produced 192 μg/ m3 / OD600 nm and 23856 produced 112 μg/ m3 / OD600 nm isoprene at maximum accumulation (Table 2). Fig. 2Isoprene emitted (solid lines) by wild type strains B. subtilis 168 (•) and B. subtilis 6051 (○) during growth (dashed lines). Zero time (t = 0) indicates the transition point between the exponential and the post-exponential growth phasesTable 2Maximum concentration of isoprene accumulated after emission by the different wild type Bacillus subtilis strains corrected for the amount of cells (OD600 nm)Bacillus subtilis strainMaximal isoprene accumulation (μg/m3/OD600 nm)16814960511662305919223856111 Mutant strains To find the optimal conditions in which expression of the genes can be limited without killing the cells, each conditional mutant strain was subjected to IPTG depletion by growing the strains on agar plates containing different concentrations of IPTG varying from 0 to 1 mM. Decreasing the concentration of IPTG caused growth inhibition after overnight incubation resulting in smaller and finally no colonies (on plates without IPTG) at 37°C. Table 3 shows the minimal concentration of IPTG at which still some growth was observed for the different conditional mutant strains. This concentration of IPTG was used in our experiments and defined as the condition causing limited induction. Most mutant strains indeed showed growth inhibition at lower concentration of IPTG. The mutant strain168IyqfY, the mutant for the putative gene encoding a homologue of the IspG protein (step 6) did not show any dependency on IPTG on plates (varying IPTG concentrations from 0 to 1 mM). Plates without IPTG incubated overnight resulted in the growth of normal colonies and the liquid cultures obtained growth curves comparable with wild type B. subtilis 168, independent of the concentration of IPTG. For some of the other mutant strains there was some growth of the bacterial cultures without IPTG as well, although severely impaired. Table 3Maximum amount of isoprene accumulated after emission by the different mutant Bacillus strains, corrected for the amount of cells (OD600 nm)Bacillus strainLimited expression (μM IPTG)Maximal isoprene accumulation (μg/m3/OD600 nm)Relative decreaseLimited inductionFull inductionFull repression168IyqiE (1)1081192715.2168IyluB (2)131167135.4168IyacM (3)504113427.9168IyabH (4)252916075.6168IyacN (5)1014119208.8168IyqfY (6)–n.d.194147–168IyqfP (7)113130179.8168ypgA (8)–––103–168IyqiD (9)101221661391.4The relative decrease is calculated as the maximal amount of isoprene at full induction devided by the maximal amount of isoprene at limited induction (dashes indicate that measurement did not apply here; explained in detail in the text, with numbering of the respective enzymatic steps).n.d., not determined For all conditional knock out strains isoprene emission was measured at the concentration of IPTG with limited expression in comparison to 1 mM (full induction) and no IPTG (full repression). Figure 3 represents the data for strain 168IyqiE, the conditional mutant strain for the first step in the biosynthetic pathway. Full induction of the dxs gene resulted in a normal growth curve of the bacterial culture and in an isoprene accumulation profile comparable to the profile of the wild type stain B. subtilis 168. Depletion of IPTG inhibited cell growth. The maximal concentration of isoprene accumulated in the flask and corrected for the amount of cells was 15 fold lower for the culture supplemented with 10 μM IPTG (8 μg/ m3 / OD600 nm) than for the culture with 1 mM IPTG (119 μg/ m3 / OD600 nm). Fig. 3Isoprene emitted (solid lines) by the B. subtilis mutant 168IyqiE during growth (dashed lines) in the presence of 1 mM IPTG (▾), 10 μM IPTG (○), and no IPTG (•). Zero time (t = 0) indicates the transition point between the exponential and the post-exponential growth phases Table 3 shows the maximum concentration of isoprene accumulated in the flask calculated on the amount of cells present at that time point for all tested mutant strains. Mutant strains for putative genes yluB, yacM, yabH, yacN, and yqfP encoding enzymes for respectively, steps 2, 3, 4, 5, and 7 showed significant differences in isoprene emission, varying from a 5–28 fold decrease at limited expression levels in comparison to full induction. The knock-out strain of ypgA, putatively encoding an isopentenyl diphosphate isomerase (step 8), is the only knock-out of a nonessential gene. The mutant strain showed a normal growth curve and did not show a significant difference in the emission of isoprene in comparison to the wild type strain B. subtilis 168. We also tested the function of yqiD, encoding a homologue for farnesyl diphosphate synthase (FDP; 7), synthase IspA (step 9) of E. coli. Depletion of IPTG to 10 μM did not result in a significant change in the isoprene emission by the mutant strain, while it did show to be dependent on IPTG for growth. Discussion To identify genes involved in the biosynthesis of the highly volatile compound isoprene, an efficient online detection system for this terpenoid (precursor) in the air above bacterial cultures was developed. Since the B. subtilis strains 6051, 23059, and 23856 were known to emit isoprene (Kuzma et al. 1995), we used these strains to set up the detection system for B. subtilis 168 for which the genome has been sequenced (Kunst et al. 1997) and functional knock outs have been made. All four strains emitted isoprene in our experiment. The differences in the amounts of isoprene emitted corresponded with the levels as described before, where B. subtilis 6051 and B. subtilis 23059 emitted higher amounts than B. subtilis 23856 (Kuzma et al. 1995). Since B. subtilis 168 showed emission of isoprene, the use of the experimental set up to investigate the isoprene synthesis by the mutant strains was justified. From the profile of isoprene emission with a maximum at the transition of the growth curve from the logarithmic phase to the stationary phase, it was concluded that from this time point on maximum isoprene concentrations had been reached and that amounts sampled from the culture every 15 min exceeded the production of new isoprene. We could not detect isoprene production in three phases using standard incubation in a shaking flask as has been described for B. subtilis 6051 grown in a bioreactor. The conditional knock out strain of the yqiE encoding DXS, the transketolase responsible for the first step in the biosynthesis of isoprenoids, appeared to be highly dependent on IPTG. The dramatic decrease in isoprene production at low concentration of IPTG supported the role of this gene in the biosynthetic pathway of isoprene. yqiE has been identified as dxs encoding 1-d-deoxyxylulose-5-phosphate in B. subtilis before (Harker and Bramley 1999; Hecht et al. 2001). However, the confirming results for this mutant strain proved that the experimental set up of our work was valid to investigate the function of the other putative genes. Apparently, the isoprene emission is severely decreased when genes in the pathway are expressed at lower level than in the wild type strain. Growth at the end of the incubation after several hours, as shown for many of the liquid cultures under IPTG depletion or even without suppletion of IPTG, may be explained by the occurrence of reversions in the mutant strain (Zanen et al. 2006). Another explanation could be the presence of a not fully repressed promoter in the complementation strain. It has been reported that in the used vector pMUTIN3 the Pspac promoter can give some low expression even in the absence of IPTG (Vagner et al. 1998; Petit et al. 1998). For genes showing low levels of expression in the wild type strain it is easy to get a pseudo wild type level of expression in the mutant strain. This may be in particular the case for highly efficient enzymes in the biosynthetic pathway. For the putative genes yluB, yacM, yabH, yacN, and yqfP, encoding homologues of the enzymatic steps 2, 3, 4, 5 and 7, the isoprene accumulation of the conditional knock outs by depletion of IPTG clearly proved the involvement of these genes in the biosynthetic pathway isoprene. B. subtilis 168IyqfY (step 6) did not depend on IPTG and therefore nothing can be concluded from the data obtained with this mutant strain other than that this enzyme may be highly efficient leading to complementation already obtained at the low expression levels caused by leaking activity of the Pspac promoter (Zanen et al. 2006). The nonessential gene ypgA encodes a homologue protein for an isopentenyl diphosphate isomerase (Idi) (step 8). The B. subtilis gene ypgA encodes for a so-called isopentenyl diphosphate isomerase type II protein (Takagi et al. 2004), while the genome of E. coli harbours a gene for a different Idi type I enzyme (Hahn et al. 1999). Interestingly, all archaea contain the Idi type II and all eukaryotes contain Idi type I. In genomes of eubacteria both are found, but there are also genomes without any of the described genes (Boucher and Doolittle 2000). In contrast to all other genes investigated in this study, the gene ypgA encoding for the IdiII was shown to be a nonessential gene in B. subtilis (Takagi et al. 2004). We also observed that the knock-out mutant of the bacterial strain is viable and produces isoprene. The idiI gene of E. coli has also been shown to be nonessential. Deletion mutants of this gene were viable on minimal medium (Hahn et al. 1999). Our findings, that the isoprene emission is not influenced by the knock out of the ypgA gene, can be explained by the enzymatic mechanism of the IspH protein, step 7, before the isomerase. IDP and DMADP can be synthesized independently by the catalytic action of IspH. A hypothetical mechanism for this reaction has been described (Eisenreich et al. 2004). This supports the nonessential character of the gene. The presence of isoprene proves that the isomer DMADP is still synthesized in the mutant and that the cells do not depend on the isomerase for the production of DMADP from IDP only. The isomerase functions in the balance of IDP and DMADP as a salvage protein (Eisenreich et al. 2004). The first enzymatic biosynthetic step downstream the formation of the precursors IDP and DMADP is catalyzed by IspA, farnesyl diphosphate synthase (step 9). Therefore the putative gene is not involved in the biosynthesis of isoprene itself. The observation that isoprene is still emitted while the expression of the gene was depleted is easily explained by its function in the pathway downstream of IDP biosynthesis. Although some accumulation of the precursors IDP and DMADP could be expected by blocking yqiD, no significant differences in the flux toward the isoprene emission were detected. Apparently the efflux of isoprene is not influenced by the changes in FDP synthesis downstream in the isoprenoid pathway. This can be regarded as a contradiction to the hypothesis that isoprene emission functions as an efflux of an overflow metabolite. In this study the involvement of several B. subtilis genes in the MEP pathway to isoprene was established by studying isoprene emission of mutants. Five genes, essential for viability, yluB, yacM, yabH, yacN, and yqfP, were shown to be essential for the isoprene production as well. Where these genes were at first instance depicted based on homology of the encoding proteins, the results of this study proved the functional involvement of the genes in the biosynthesis of isoprenoids on a biochemical level. A knockout of the sixth candidate gene, yqfY, did not yield a reduction in isoprene emission, nor a growth retardation, which might be explained by a not fully repressed Pspac promoter present in vector pMUTIN3. Knock-outs of the two other genes investigated in this study, showing no growth retardation as well as no reduction in isoprene emission, behaved in accordance with expectation. Although it remains unknown why isoprene is emitted by bacteria, the knowledge about the pathway to isoprene can be used for further investigations towards a better understanding of the metabolic flux to IDP and its physiological function. Next to that the IDP pool can be optimized by metabolic pathway engineering, creating a B. subtilis strain as an efficient Gram-positive host for the heterologous production of terpenoids, like the valuable plant-derived pharmaceuticals artemisinin and paclitaxel. The potential of using heterologous production organisms for the supply of terpenoids is already under investigation for E. coli and for several yeast strains (Martin et al. 2003; Lindahl et al. 2006; Ro et al. 2006; Dejong et al. 2006). For the purpose of metabolic pathway engineering it is interesting to search for a gene encoding isoprene synthase in B. subtilis as well. Evidence for an enzymatic bioconversion and the identification of this specific enzyme could give the possibility to block the efflux of isoprene thereby increasing the amount of isoprenoid available for the synthesis of terpenoids.

Eur_J_Nucl_Med_Mol_Imaging-3-1-2121116

Left ventricular dyssynchrony assessed by two three-dimensional imaging modalities: phase analysis of gated myocardial perfusion SPECT and tri-plane tissue Doppler imaging

Purpose To compare left ventricular (LV) dyssynchrony assessment by phase analysis from gated myocardial perfusion SPECT (GMPS) with LV dyssynchrony assessment by tri-plane tissue Doppler imaging (TDI). Baseline LV dyssynchrony assessed with standard deviation (SD) of time-to-peak systolic velocity of 12 LV segments (Ts-SD) with TDI has proven to be a powerful predictor of response to CRT. Information on LV dyssynchrony can also be provided by GMPS with phase analysis of regional LV maximal count changes throughout the cardiac cycle. Introduction Cardiac resynchronization therapy (CRT) has become an established therapy for patients with end-stage heart failure [1]. However, up to 30% of the patients do not respond to CRT when selection is based on the traditional clinical and electrocardiographic criteria [New York Heart Association (NYHA) functional class III or IV, left ventricular ejection fraction (LVEF) <35%, and QRS duration >120 ms] [2]. In the search for better selection criteria for CRT, it has been shown that patients with LV dyssynchrony have a higher likelihood of a positive response to CRT [3–8]. Over the past few years, the assessment of LV dyssynchrony has been studied extensively with echocardiography. The most frequently used technique is tissue Doppler imaging (TDI), which permits evaluation of timing of regional myocardial velocity and comparison of different regions yields information on LV dyssynchrony. Bax et al. [3] have used a four-segment model to assess LV dyssynchrony (four basal LV segments: anterior, lateral, inferior and septal) and reported that a delay ≥ 65 ms among the four segments was a strong predictor of response to CRT. Other authors extended the analysis by evaluating the activation pattern of the entire LV. This method, originally described by Yu and co-workers [6–8], uses the standard deviation (SD) of electromechanical activation times based on a 12-segment model of LV (Ts-SD). The authors investigated 56 patients with heart failure who underwent CRT implantation and demonstrated that a cut-off value of 34.4 ms for Ts-SD could predict reverse LV remodeling after CRT with a sensitivity of 87% and a specificity of 81% [7]. However, to calculate this parameter of LV dyssynchrony, three apical views need to be acquired separately. A 3-dimensional (3D) TDI imaging modality (tri-plane TDI) has become available to overcome this limitation. Tri-plane TDI allows simultaneous acquisition of all LV segments during the same heartbeat rendering the technique more precise than the 2-dimensional (2D) TDI equivalent. Another recent development is the use of gated myocardial perfusion single photon emission computed tomography (GMPS) for assessment of LV dyssynchrony [9, 10]. A count-based method has been developed to extract phase information from the regional LV count changes throughout the cardiac cycle, and this phase information is related to the time interval when a region, in the 3D LV wall starts to contract (phase analysis). This technique has recently been implemented in the Emory Cardiac Toolbox as a diagnostic tool for assessment of LV mechanical dyssynchrony [9]. To further validate the use of GMPS with phase analysis for the assessment of LV dyssynchrony, a direct comparison with tri-plane TDI was performed in a cohort of heart failure patients. Materials and methods Patients and study protocol Forty consecutive patients with heart failure, referred for evaluation of potential eligibility for CRT, were included in this study. Traditional selection criteria for CRT were applied, including moderate-to-severe drug refractory heart failure (NYHA class III or IV), depressed LVEF (<35%) and prolonged QRS duration (>120 ms). Patients with a recent myocardial infarction (<3 months) or decompensated heart failure were excluded. Evaluation of the clinical status included assessment of NYHA functional class, quality-of-life score (using the Minnesota quality-of-life questionnaire) and exercise capacity (using the 6-min walk test). Extensive echocardiographic analysis, using 3D transthoracic echocardiography, was performed to assess LV dyssynchrony. Resting GMPS with technetium-99m tetrofosmin was clinically performed to exclude extensive ischemia and/or viability [11, 12], in order to refer patients to revascularization if indicated. Thereafter, all scans were subjected to phase analysis to evaluate LV dyssynchrony. The results of the phase analysis with GMPS to assess LV dyssynchrony were subsequently compared with LV dyssynchrony derived from tri-plane TDI. Echocardiography 3D data set acquisition All patients underwent transthoracic echocardiography performed with a commercially available echocardiographic platform (VIVID 7, GE Vingmed Ultrasound, Horten, Norway) and equipped with a 3V-probe for 3D acquisition. Patients were scanned in left lateral decubitus position, from the apical window in tri-plane modus, acquiring simultaneously the apical four-, two- and three-chamber views. Care was taken to visualize the true LV apex. Color-coded TDI was applied to the tri-plane view to assess longitudinal myocardial regional function. Gain settings, filters and pulse repetition frequency were adjusted to optimize color saturation. Sector size and depth were optimized for the highest possible frame rate. At least two consecutive beats were recorded from each view, and the images were digitally stored for off-line analysis (EchoPac, GE Vingmed Ultrasound, Horten, Norway). The echocardiographic examination and the off-line analysis were performed by the same experienced echocardiographist, blinded to the GMPS and clinical data. 3D analysis of LV dyssynchrony During post-processing, the tri-plane TDI dataset was used to analyze myocardial velocity curves as previously described with 2D TDI echocardiography [6–8]. Sample volumes were placed in the basal and mid segments of the septal, lateral, inferior, anterior, posterior and anteroseptal LV walls to calculate the time from the beginning of the QRS complex to peak systolic velocity (Ts), using the same heart beat (Fig. 1). Subsequently, the standard deviation of Ts (Ts-SD) of the 12 myocardial segments was calculated and used as a parameter of LV dyssynchrony [6–8]. A cut-off value of Ts-SD ≥ 33 ms was used to define the presence of substantial LV dyssynchrony. Fig. 1Example of the myocardial velocity curves that can be derived by positioning the sample volume in any LV segment of the tri-plane dataset. This patient has substantial LV dyssynchrony: the postero-lateral and anterior walls (orange, light blue and red curves) are activated later than the septal and inferior walls (yellow and green curves); standard deviation of 12 LV segments' Ts is 57.5 ms Gated myocardial perfusion SPECT Resting GMPS with technetium-99m tetrofosmin (500 MBq, injected at rest) was performed using a triple head SPECT camera system (GCA 9300/HG, Toshiba Corp.) equipped with low energy high-resolution collimators. Around the 140-KeV energy peak of technetium-99m tetrofosmin, a 20% window was used. A total of 90 projections (step and shoot mode, 35 s per projection, imaging time 23 min) were obtained over a 360-degrees circular orbit. GMPS acquisition involved 16 frames per cardiac cycle. Data were reconstructed by filtered backprojection and then re-oriented to yield gated short-axis images. Data reconstruction was performed over 360 degrees. These images were then submitted to the Emory Cardiac Toolbox for phase analysis [9]. GMPS images were analyzed at the department of radiology of Emory University School of Medicine by J.C. and E.V.G., who were blinded to the echocardiographic and clinical data. A phase distribution was extracted from a GMPS study, representing the regional onset of mechanical contraction of the LV. It can be displayed in a polar map or in 3D and used to generate a phase histogram. An example of a patient without LV dyssynchrony is shown in Fig. 2a, whereas an example of a patient with substantial LV dyssynchrony as assessed with phase analysis by GMPS is shown in Fig. 2b. The following four quantitative indices were obtained from the phase analysis of all patients: (1) histogram bandwidth, which includes 95% of the elements of the phase distribution; (2) phase SD, which is the SD of the phase distribution; (3) phase histogram skewness, which indicates the symmetry of the histogram; (4) histogram kurtosis, which indicates the degree to which the histogram is peaked [9]. In a healthy individual the LV contracts in a coordinated manner, and the phase distribution is nearly uniform with a highly peaked distribution of the histograms. As the LV contraction becomes dyssynchronous, the phase SD and histogram bandwidth are expected to increase (Fig. 2b). Fig. 2a Example of a patient without LV dyssynchrony on GMPS. The non-normalized (upper panel) and normalized (lower panel) phase distributions are nearly uniform and the corresponding phase histograms are highly-peaked, narrow distributions. b Example of a patient (same as in Fig. 1) with LV dyssynchrony on GMPS. The non-normalized (upper panel) and normalized (lower panel) phase distributions show significant non-uniformity and the corresponding phase histograms have widespread distributions Statistical analysis Continuous data are presented as mean ± SD; dichotomous data are presented as numbers and percentages. Comparison of data was performed using the unpaired Student's t test or χ2 test when appropriate. Pearson’s correlation analysis was performed to evaluate the relation between histogram bandwidth, phase SD, histogram skewness and histogram kurtosis by the phase analysis of GMPS and LV dyssynchrony (Ts-SD) by tri-plane TDI. A p-value <0.05 was considered to be statistically significant. A statistical software program SPSS 12.0 (SPSS Inc, Chicago, Il) was used for statistical analysis. Results Study population A total of 40 consecutive patients with heart failure were included in this study [29 (72%) men; mean age 66±10 years]. Baseline characteristics are summarized in Table 1. Patients had severe LV dysfunction (mean LVEF 26 ±7%), with extensive LV dilatation (mean LV end-diastolic volume 210±48 ml). LV dyssynchrony assessment with tri-plane TDI demonstrated a mean value of Ts-SD of 34±15 ms. Table 1Baseline characteristics of the study population (n=40)Age (years)66±10Gender (M/F) 29/11NYHA class 3.0±0.46-MWT (m)282±126QoL score38±17QRS duration (ms)147±33Etiology, n (%) Ischemic25 (62) Idiopathic 15 (38)LVEF (%)26±7LVEDV (ml)210±48LVESV (ml)157±43Medication, n (%) ACE Inhibitors34 (85) β-blockers28 (70) Diuretics and/or spironolactone36 (90)LVEDV = left ventricular end-diastolic volume; LVEF = left ventricular ejection fraction; LVESV = left ventricular end-systolic volume; 6-MWT = 6-min walk test; QoL = quality of life score; Ts-SD = standard deviation of time to peak systolic velocity of 12 LV segments Histogram bandwidth The mean histogram bandwidth was 130°±69° (range 41° to 260°); the normal value is 38.7°±11.8° for men and 30.6°±9.6° for women [9]. A good correlation between histogram bandwidth and LV dyssynchrony, assessed with tri-plane TDI, was observed (r=0.77, p<0.0001, Fig. 3). Fig. 3Correlation between histogram bandwidth assessed with GMPS and LV dyssynchrony assessed with tri-plane TDI (Ts-SD) Phase SD The mean phase SD was 40.2°±18.8° (range 14.5° to 77.9°), whereas the normal value is 14.2°±5.1° for men and 11.8°±5.2° for women [9]. Pearson’s correlation showed a good correlation between phase SD and Ts-SD (r=0.74, p<0.0001, Fig. 4). Fig. 4Phase SD assessed with GMPS versus LV dyssynchrony assessed with tri-plane TDI (Ts-SD) Histogram skewness The mean histogram skewness was 2.50±0.81 (range 1.16 to 5.49), whereas the normal value is 4.19±0.68 for men and 4.60±0.72 for women [9]. A poor correlation with LV dyssynchrony assessed with tri-plane TDI was demonstrated by Pearson’s correlation (r=-0.30, p=0.06). Histogram kurtosis The mean histogram kurtosis was 7.38±6.23 (range 1.08 to 37.42), whereas the normal value is 19.72±7.68 for men and 23.21±8.16 for women [9]. No correlation between histogram kurtosis and Ts-SD was shown by Pearson’s correlation (r=−0.14, p=0.38). Substantial versus no substantial LV dyssynchrony Of the 40 patients included, 20 (50%) had a substantial LV dyssynchrony (defined as Ts-SD  ≥ 33 ms) [7, 8]. No significant differences in baseline characteristics were noted between the two patient groups (Table 2). Histogram bandwidth and phase SD were significantly higher in patients with Ts-SD  ≥ 33 ms compared to patients with Ts-SD <33 ms: 186°±52° versus 74°±24° (p<0.0001) and 55.3°±13.6° versus 25.1°±7.6° (p<0.0001), respectively (Fig. 5a and b). In addition, histogram skewness was significantly lower in patients with Ts-SD  ≥ 33 ms than in patients with Ts-SD <33 ms (2.23±0.93 versus 2.78±0.55, p=0.03). No significant difference between the two groups was observed for histogram kurtosis (6.36±7.94 versus 8.4±3.79, p=NS) (Fig. 5c and d). An example of a patient with substantial LV dyssynchrony on both tri-plane TDI as well as on phase analysis with GMPS is provided in Figs. 1 and 2b. Table 2Baseline characteristics of patients with substantial (≥33 ms) and no substantial LV dyssynchrony (<33 ms) assessed by SD of time-to-peak systolic velocity of 12 LV segments (Ts-SD) Ts-SD ≥33 ms (n=20)Ts-SD <33 ms (n=20)p-valueAge (years)67±1066±9NSGender (M/F)14/615/5NSQRS duration (ms)145±31149±36NSNYHA class 3.0±0.43.0±0.4NSEtiology, n (%) Ischemic 14 (70)11 (55)NS Idiopathic6 (30)9 (45)NSLVEF (%)25±726±7NSLVEDV (ml)209±47207±50NSLVESV (ml)159±46153±42NSHistogram bandwidth (°)186±5274±24<0.0001Phase SD (°)55.3±13.625.1±7.6<0.0001Histogram skewness2.23±0.932.78±0.550.03Histogram kurtosis6.36±7.948.40±3.79NSFor abbreviations see Table 1.Fig. 5a Patients with substantial versus no substantial LV dyssynchrony (Ts-SD), using a cut-off value ≥33 ms. Histogram bandwidth is significantly higher in patients with substantial LV dyssynchrony. b Patients with substantial versus no substantial LV dyssynchrony (Ts-SD), using a cut-off value ≥33 ms. Phase SD is significantly higher in patients with substantial LV dyssynchrony. c Patients with substantial versus no substantial LV dyssynchrony (Ts-SD), using a cut-off value ≥33 ms. Histogram skewness is significantly lower in patients with substantial LV dyssynchrony. d Patients with substantial versus no substantial LV dyssynchrony (Ts-SD), using a cut-off value ≥33 ms. No significant difference in histogram kurtosis was demonstrated between patients without and patients with substantial LV dyssynchrony as assessed with tri-plane echocardiography Ischemic versus idiopathic dilated cardiomyopathy In total, 25 patients (62%) had an ischemic cardiomyopathy and 15 patients (38%) had an idiopathic dilated cardiomyopathy. No differences in echocardiographic variables, including LV dyssynchrony assessed by tri-plane TDI, or GMPS phase analysis indices were found between ischemic and non-ischemic patients. No significant differences in correlation coefficients were observed between patients with ischemic cardiomyopathy and patients with idiopathic dilated cardiomyopathy for the four different GMPS variables (Table 3). Table 3Correlation coefficients of LV dyssynchrony, assessed with tri-plane TDI, versus the different LV dyssynchrony parameters, derived from phase analysis of GMPS (idiopathic dilated cardiomyopathy versus ischemic cardiomyopathy) Ischemic cardiomyopathy (n=25)Idiopathic dilated cardiomyopathy (n=15)p-valueHistogram bandwidthr=0.79r=0.78NSPhase SDr=0.77r=0.75NSHistogram skewnessr=−0.22r=−0.48NSHistogram kurtosisr=−0.07r=−0.35NS Discussion The main findings of the present study can be summarized as follows: (1) LV dyssynchrony assessment with phase analysis on GMPS is feasible in heart failure patients screened for eligibility for CRT implantation. The phase analysis variables histogram bandwidth and phase SD show good correlation with LV dyssynchrony as measured with tri-plane TDI (Ts-SD), suggesting that GMPS phase analysis can be used to quantify LV dyssynchrony. (2) Patients with substantial LV dyssynchrony on tri-plane TDI have significantly higher values of histogram bandwidth and phase SD derived from phase analysis with GMPS (histogram bandwidth: 186°±52° versus 74°±24°, p<0.0001) and phase SD: 55.3°±13.6° versus 25.1°±7.6° (p<0.0001). CRT is now considered an important therapeutic option in the treatment of patients with end-stage heart failure [1]. Although encouraging beneficial effects have been reported, including improvement in LVEF and reduction in heart failure symptoms, it has also been demonstrated that up to 30% of patients do not have a favorable response to CRT. The presence of LV dyssynchrony is considered a key issue in the identification of potential responders to CRT [3–6]. Echocardiography and in particular TDI have been extensively used to assess LV dyssynchrony. Several parameters derived from myocardial velocity curves of different LV segments have been proposed to quantify LV dyssynchrony and predict response to CRT [3–8]. Yu and coworkers described a model (derived from 2D TDI data) to quantify global LV dyssynchrony using Ts-SD of 12 LV segments [6–8]. The authors reported a cut-off value >32.6 ms to distinguish between responders and non-responders [8] and a cut-off value >34.4 ms as the most powerful predictor of LV reverse remodeling [7]. In this study LV dyssynchrony was evaluated by Ts-SD from tri-plane TDI and compared with LV dyssynchrony assessment with phase analysis by GMPS. The tri-plane TDI technique avoids the problem of heart rate variability since simultaneous acquisition of the three apical views is possible; in the present study a cut-off value  ≥ 33 ms was used to define substantial LV dyssynchrony on tri-plane TDI. Information on cardiac dyssynchrony can also be derived from nuclear imaging. Kerwin et al. [13] evaluated interventricular dyssynchrony in 13 heart failure patients using multi-gated equilibrium blood pool scintigraphy and phase analysis. Patients were imaged during biventricular pacing and in sinus rhythm. During biventricular pacing, a significant improvement in interventricular synchrony was observed. Moreover, an improvement in LVEF was reported for all 13 patients during biventricular pacing (17.2±7.9% to 22.5±8.3%, p<0.0001), and the increase in LVEF correlated significantly with the improvement in interventricular synchrony (r=0.86; p<0.001). Fauchier et al. [14] used equilibrium radionuclide angiography and phase analysis to evaluate the prognostic value of interventricular and intraventricular dyssynchrony in 103 patients with idiopathic dilated cardiomyopathy. Phase images were generated from the scintigraphic data, and mean phase angles and SDs were calculated for the right and left ventricle. During a follow-up period of 27±23 months, 18 major cardiac events occurred. Multivariate analysis revealed that LV intraventricular dyssynchrony and elevated pulmonary capillary wedge pressure were the only independent predictors of cardiac events. GMPS is currently used in patients with heart failure to exclude ischemia and/or viability before referring for CRT implantation and to assess total scar burden in patients with ischemic cardiomyopathy [11, 12]. Indeed, various studies have shown that the presence of severe resting perfusion defects, as assessed with GMPS, limits the response to CRT [15–17]. Recently, Chen et al. [9] developed with GMPS a count-based method to extract phase information from the regional LV count changes throughout the cardiac cycle. The phase information is related to the regional onset of mechanical contraction of the LV (phase analysis) and provides information on the synchronicity of the LV contraction. The authors assessed in 90 normal individuals the normal range for four quantitative indices that can be used as markers of LV dyssynchrony (histogram bandwidth, phase SD, histogram skewness and histogram kurtosis) [9]. Recently, these four GMPS indices have been compared with LV dyssynchrony assessment by TDI in 75 patients with severe heart failure. It was shown that among the four quantitative indices of phase analysis, the variables histogram bandwidth and phase SD correlated best with LV dyssynchrony as assessed by TDI [10]. The current study involved a different patient subset and provides further support that phase analysis with GMPS can be useful in the evaluation of LV dyssynchrony. The value of GMPS to assess LV dyssynchrony has been analyzed, and the indices derived from phase analysis were compared with Ts-SD measured by tri-plane TDI. It was shown that histogram bandwidth and phase SD correlated well with Ts-SD as evaluated by tri-plane TDI (histogram bandwidth: r=0.77 and phase SD: r=0.74). In addition, significantly higher values for histogram bandwidth and phase SD were observed in patients with substantial LV dyssynchrony on tri-plane TDI compared with patients without substantial LV dyssynchrony on tri-plane TDI. Limitations In the present study, follow-up data after CRT implantation were not available. Therefore, the value of phase analysis to predict response to CRT could not be determined. Furthermore, GMPS may be less suitable for repeated analysis during the follow-up after CRT implantation, due to the radiation burden of the technique. Conclusions The results of this study confirm the feasibility to evaluate LV dyssynchrony with phase analysis by GMPS and its applicability in the clinical setting. In particular, histogram bandwidth and phase SD showed a good correlation with LV dyssynchrony measured with Ts-SD as assessed with tri-plane TDI. Future prospective studies in larger patient populations with follow-up after CRT implantation are needed to elucidate the potential role of GMPS with phase analysis for prediction of response to CRT.

Neuropsychol_Rev-3-1-2039835

Motor-Skill Learning in Alzheimer’s Disease: A Review with an Eye to the Clinical Practice

Since elderly people suffering from dementia want to go on living independently for as long as possible, they need to be able to maintain familiar and learn new practical skills. Although explicit or declarative learning methods are mostly used to train new skills, it is hypothesized that implicit or procedural techniques may be more effective in this population. The present review discusses 23 experimental studies on implicit motor-skill learning in patients with Alzheimer’s disease (AD). All studies found intact implicit motor-learning capacities. Subsequently, it is elaborated how these intact learning abilities can be exploited in the patients’ rehabilitation with respect to the variables ‘practice’ and ‘feedback.’ Recommendations for future research are provided, and it is concluded that if training programs are adjusted to specific needs and abilities, older people with AD are well able to (re)learn practical motor skills, which may enhance their autonomy. Introduction The aging population is growing rapidly and by 2050 the number of elderly people aged 85 years old or older in Europe and North America is estimated to be approximately 19 million (Román 2002). Since age is a high risk factor for dementia (Smith and Rush 2006), the expansion of the aging population and thus the number of people suffering from dementia has momentous consequences for national care systems as well as a large economic impact (Schölzer-Dorenbosch 2005). A way to contain these costly effects is by helping older people to stay independent for as long as possible, implying that the elderly must not stop learning. They apply old skills differently or acquire new skills, like learning how to use a walking aid, which gives rise to the following questions: Are demented elderly people able to learn such new skills? And are we, health professionals, able to train them? As yet, psychopharmacological interventions, such as the use of cholinesterase inhibitors, may have some benefit in maintaining autonomy of elderly patients suffering from Alzheimer’s disease (AD), as demonstrated by delayed nursing home placement (Becker et al. 2006). Recent pharmacological research also shows promising results for cognition: the treatments evaluated produced a moderate positive effect on global cognitive functioning (Grimley Evans et al. 2004; Schölzer-Dorenbosch 2005; Takeda et al. 2006). In their 2002 review on the placebo-controlled effects of rivastigmine on the cognition of AD patients, Birks et al. (2002) reported statistically-significant increases of 0.8 points on the Mini Mental State (MMS) Examination and 2.1 points on the Alzheimer’s Disease Assessment Scale (ADAS-Cog). They also found benefits in the patients’ activities of daily living, although the difference with placebo was not significant. Takeda et al. (2006) also reported reductions in the cognitive impairment of AD patients for donepezil and galantamine, but again, although showing potential, the improvements did not suggest a major difference in the daily lives of the patients. These findings do not negate the importance of non-pharmacological approaches and it is possible that the interactions between medication and non-pharmacological approaches may be the most beneficial in maintaining patient’s autonomy. In their review, Luijpen et al. (2003) conclude that the effect of non-pharmacological interventions with respect to cognition and affective behavior in dementia is similar to the effect of pharmacological regimens. Non-pharmacological treatments to improve autonomy in this patient population should hence be considered an additional option, especially since rehabilitation is increasingly being advocated as a means to optimize patients’ overall functioning (De Vreese et al. 2001). Clare (2003) also concludes that neuropsychological rehabilitation applied in the context of progressive disorders like dementia do yield beneficial results. In the present review we will focus on the ability of Alzheimer’s patients to (re)learn practical motor skills. Currently, explicit or declarative learning methods are the starting points in most rehabilitation programs aimed at motor-skill learning in the cognitively unimpaired population (Van Cranenburg 2004). However, Zanetti et al. (2001) and Rösler et al. (2002) claimed that patients with dementia will profit more from implicit or procedural learning methods by showing that their AD cohorts were able to learn to waltz or to use a telephone when an implicit rather than an explicit learning approach was used. In implicit learning, skills are mastered without awareness, often simply by repeated exposure, and can be unconsciously revived from implicit memory (Buchner and Wippich 1998). The abovementioned studies were all focused on finding the best way to help older people with dementia learn or relearn practical (motor) skills, and although the results are encouraging, the patient samples were always small and it remains unclear how much was learned due to a lack of well-defined performance measures. In the first part of our review we looked for corroborating evidence in experimental research for intact implicit motor-learning capacity in cohorts of elderly patients diagnosed with AD. In the second part we will elaborate on how these intact learning abilities can be utilized for their rehabilitation while taking the principles from theories of motor-learning into account. Two of the theories’ core variables, i.e., practice and feedback (Schmidt and Wrisberg 2000) will be discussed more extensively, also in the light of research exploring these variables in AD. The results will be translated into practical instructions for more targeted rehabilitation training programs for this patient group. Materials and Methods Computerized searches of the literature using the databases of PubMed and PsycLIT were conducted spanning a 20-year period, from 1985 up to and including 2005. The search terms (any field) used were procedural learning, sequence learning, motor-skill learning, or motor learning in combination with Alzheimer’s disease. Only reports published in English were considered. For inclusion in this review the studies had to meet the following criteria: (a) a clinical diagnosis of Alzheimer’s disease based on specified and generally accepted criteria; (b) a procedural task with motor responses; and (c) task performance expressed in time or error measures, and not only in fMRI or other imaging data. Ultimately, 23 studies were included in this review. Three studies will only be discussed in the second part of the review since they explicitly examined the role of feedback and type of practice. Experimental Research of Implicit Motor-Skill Learning in Alzheimer’s Disease Implicit Learning Ability The main results of the studies generated by our search of the literature are shown in Table 1. Four studies using a Maze test in which blindfolded participants had to trace a complex pathway found that the AD patients were able to learn new motor-skills implicitly (Kuzis et al. 1999; Sabe et al. 1995; Starkstein et al. 1997; Taylor 1998). The nine studies that applied a Rotor-Pursuit task, in which participants had to maintain contact between a hand-held stylus and a rotating spot, also reported preserved learning abilities in their AD samples (Beatty et al. 1995; Deweer et al. 1994; Dick et al. 1995, 2001; Heindel et al. 1988; Heindel et al. 1989; Jacobs et al. 1999; Libon et al. 1998; Willingham et al. 1997). This is in agreement with the findings of Poe and Seifert (1997) based on a Puzzle-Assembly task and the results of Rouleau et al. (2002) involving a Mirror-Tracing task. Also, a Serial Reaction-Time Task (SRTT) was used in which participants needed to respond as fast as possible when a stimulus appeared in one of four places by pressing a corresponding response key (Grafman et al. 1990; Knopman and Nissen 1987; Knopman 1991; Willingham et al. 1997). Again, the AD patients showed implicit learning as reflected by the difference in reaction times (RTs) between blocks with a fixed sequence of stimuli presentation (decreasing RTs) and a random block (prolonged RTs). However, there are indications that the implicit learning ability in AD patients is affected because they generated inferior outcomes when accuracy was taken into account (Willingham et al. 1997) or when the data were log-transformed because of the unequal variance in RT (Knopman 1991). Ferraro et al. (1993) found preserved implicit SRTT learning only in the “very mildly demented” group, and less in the “mildly demented” group although it is relevant to mention that none of the other studies used such a subtle severity classification. Table 1Summary of results of experimental studies on motor-skill learning in Alzheimer’s diseaseAuthorYearSample size and typesTask(s)Amount of learningaResults on learning capacitySabe, L., et al.199520 AD with co-morbid depression, 35 AD without co-morbid depression, 14 depressive, non-demented patients, 16 healthy controlsMaze testAD: 19%Co: 22%The AD patients showed significant deficits in declarative learning but only a minor (although statistically significantly) drop in procedural learning. The AD group with comorbid depression showed a similar learning pattern as the non-depressed AD group.GroupxTrial:p < 0.05Starkstein, S.E., et al.199755 AD (13 with mild, 12 with severe and 30 without anosognosia)Maze testAD no: 48%There was no group difference in declarative learning. As to procedural learning, the patients with severe anosognosia showed a significantly poorer performance whereas the patients with mild or no anosognosia showed no deficits.AD mild: 39%AD severe: −16%–Taylor, R.199858 AD, 58 multi-infarct dementiaMaze test–When age and overall neuropsychological functioning were taken into account, Maze performance was better in the AD patients than in the patients with multi-infarct dementia–Kuzis, G., et al.199915 AD, 15 PD, 10 PD and dementia, 24 healthy controlsMaze testAD: 10%The AD group showed deficits on all measures of explicit memory. There were no significant between-group differences in the measures of implicit memory between the AD, control, and PD groups.Co: 39%–Heidel, W.C., et al.198810 AD, 10 HD, 4 amnestic 20 healthy controlsRotor PursuitAD: 147%The AD patients showed preserved motor-skill learning while the patients with HD showed no motor learning.Co: 115%GroupxTrial: n.s.Heindel, W.C. et al.198916 AD, 13 HD, 17 PD, 22 healthy controlsRotor PursuitAD: 101%The AD patients showed preserved motor-skill learning while the patients with HD showed impaired motor learning.Co: 118%GroupxTrial: n.s.Beatty, W.W., et al.19954 AD, 1 corticbasal degenerationRotor Pursuit–The AD patients showed preserved motor skill learning–Deweer, B., et al.199413 AD institutionalized, 10 healthy controls, 17 AD out patients, 9 healthy controlsRotor PursuitAD in.:86%Explicit memory was severely impaired in the AD patients but they showed normal procedural learning.Co: 48%GroupxTrial: n.s.AD out: 161%Co: 139%GroupxTrial: n.s.Dick, M.B., et al.199512 AD, 12 healthy controlsRotor PursuitAD: 47%Performance significantly improved during the first 40 trials but additional practice provided no further beneficial effects. The AD patients showed minimal retention problems across four retention tests.Co: 81%GroupxTrial: n.s.Libon, D.J., et al.199816 AD, 14 vascular dementiaRotor PursuitAD: 60%The AD patients obtained a lower score on a verbal-learning task-recognition index and high scores on the Rotor Pursuit.–Jacobs, D.H., et al.199912 AD, 12 healthy controlsRotor PursuitAD: 124%The AD patients and the controls were able to learn the motor task.Co: 106%GroupxTrial: p = 0.473Dick, M.B., et al.200118 AD, 18 healthy controlsRotor PursuitAD: 27%In normal-vision trials no differences in learning between the AD patients and the controls were found.Co: 36%GroupxTrial: n.s.Dick, M.B., et al.200399 AD, 100 healthy controlsRotor PursuitThe AD patients and controls receiving constant practice outperformed those in the blocked and random conditions. The AD patients only benefited from constant practice.Poe, M.K. et al.19979 AD, 14 healthy controlsPuzzle Assembly–Even when the subjects had no explicit memory of practicing the task, they all demonstrated savings upon relearning. GroupxTrial: n.s.Rouleau, I., et al.200212 AD, 12 healthy controlsMirror TracingAD: 44%Those AD patients that were able to perform the basic mirror-tracing task did not differ from the controls in level of performance, learning over trails, retention over a delay interval and generalization to other tasksCo: 49%GroupxTrial: n.s.Knopman, D.S., et al.198735 AD, 13 healthy controlsSRTTAD: 22%The AD patients showed learning of the repeated sequence, although they responded more slowly.Co: 38%GroupxTrial: n.s.Graftman, J., et al.199042 AD, 7 PSP, 44 healthy controlsSRTTAD: 36%The AD patients and controls showed motor-skill learning while the PSP patients did not.–Knopman, D.199116 AD, 17 healthy controlsSRTTAD: 37%The AD patients showed learning of the sequence but they showed an inferior level of learning when the data were log-transformed.Co: 33%Ferraro, F.R., et al.199327 very mild AD, 15 mild AD, 17 PD, 26 healthy controlsSRTTAD mild: 11%The very mildly AD patients showed preserved learning comparable with the controls. The mildly AD patients and PD patients showed less implicit learning.AD very mild:22%Co: 20%Willingham, D.B., et al.199720 AD, 20 healthy controlsSRTT, Incompatible SRTT, Pursuit Tracking (randomized and repetitive pattern)SRTT:The dementia ratings predicted the ability to perform tasks but not the ability to learn them. AD patients can have a performance deficit but they have no general deficit in motor-skill learning.AD: 52%Co: 60%GroupXTrial: p > 0.2Pursuit:AD: 12%Co: 17%GroupxTrial: p > 0.2Hirono, N., et al.199736 AD, 19 healthy controlsBi-manual coordinated Tracing taskAD: 37%Skill learning in those AD patients that completed the tasks was as good as in the controls. Co: 39%GroupxTrial: p = 0.193Dick, M.B., et al.199623 AD, 22 healthy controlsTossingThe AD patients given constant practice were able to learn and retain the tossing task similarly well as the controls. The AD patients showed less improvement when practicing at various distances.Dick, M.B., et al.200658 AD, 58 healthy controlsTossingThe AD patients showed significant improvements under constant practice only. None of the practice conditions facilitated intermediate transfer in the AD patients whereas constant practice did benefit them on tests assessing near transfer.AD= Alzheimer’s disease; HD= Huntington’s disease; PD= Parkinson’s disease; PSP=  Progressive supranuclear palsy.aexpressed as a percentage of the difference score between the last and first trial with respect to the score on the first trial. The GroupxTrial interaction for the AD and Co group is also reported when available. Thus, irrespective of the task used, the studies assessing implicit motor-skill learning in AD we reviewed yielded positive outcomes. Indeed, in their 1997 study, Hirono and colleagues found that patients with mild AD were able to acquire motor and perceptual as well as cognitive skills in various procedural learning tasks. It should be noted that in all studies the patients that could not perform the task were eliminated from the analyses. Yet, a failure to perform the prescribed task need not necessarily be related to learning problems. Willingham et al. (1997) attributed the phenomenon to other causes like the complexity of the instructions given or the type of skill to be performed. These factors may differ across tasks, which might explain their finding that the ability to complete one task did not predict the rate of improvement in another task. They conclude that AD patients have a performance deficit and not a generalized deficit in motor learning. Performance and Amount of Learning From the above discussion of results we can presume that at least a subgroup of AD patients show preserved implicit learning abilities, but to what extent? Here, two aspects in motor learning should be differentiated, i.e., overall performance level and amount of learning (e.g., the increment in Total Time on Target in the Rotor-Pursuit task). All the studies found preserved motor-skill learning in AD patients although their overall performance levels in terms of reaction and movement time were always inferior to those of the controls. However, when we take the level of learning into account, the results are less consistent. Some of the results were not reported with enough detail to show unambiguously the amount of learning the AD patients showed compared to the controls (Poe and Seifert 1997). Some comparative studies did not include a healthy control group in addition to the patient groups (Beatty et al. 1995; Grafman et al. 1990; Libon et al. 1998; Starkstein et al. 1997; Taylor 1998), preventing patient-control comparisons from being made. The AD patients in the SRTT studies showed the same amount of learning (decrease in RT during the blocks with the fixed sequence) as the normal controls (Ferraro et al. 1993; Knopman and Nissen 1987; Knopman 1991; Willingham et al. 1997). In the nine studies that used a Rotor-Pursuit or Tracking task there were also no patient-control differences in amount of learning (Deweer et al. 1994; Dick et al. 1995, 2001; Heindel et al. 1988; Heindel et al. 1989; Hirono et al. 1997; Jacobs et al. 1999; Rouleau et al. 2002; Willingham et al. 1997). Two of the studies using a Maze test reported learning abilities in the AD group but less improvement across trials compared to the controls, (Kuzis et al. 1999; Sabe et al. 1995) findings which are perhaps explainable by the use of a task without visual feedback. Taken together, the reviewed studies all showed preserved implicit motor-skill learning in AD patients regardless of the task used. Their performance levels, however, never reached the levels of the healthy controls, demonstrated by their prolonged reaction and movement times. The AD patients’ level of learning also differed depending on the task to be performed. Visual feedback appears to have a positive effect on their learning pace. They also seem to experience more problems with implicit learning when performing the SRTT, a task that involves two learning processes. The subjects have to master both spatial and motor regularities (Mayr 1996), and it is the learning of spatial regularities that may be compromised in AD patients, a process that is less implicated in the Rotor-Pursuit task in which normal implicit learning for the patients was found. Training Patients with Alzheimer’s Disease: Variables in Motor Learning The studies discussed provide evidence that AD patients can learn new motor skills in an implicit way. It is therefore worthwhile to establish what would be the best way to train them. In the next section we will give a brief account of the two variables practice and feedback that play a role in (re)training motor skills. We will subsequently discuss the variables in relation to the findings reported in the relevant AD studies and conclude by making recommendations of how to enhance the acquisition of new motor skills in this population. We will first, however, briefly address the existing views on the presence or absence of distinguishable learning stages in explicit and implicit learning. Generally, with explicit learning people tend to pass through three stages in the acquisition of motor skills (Fitts and Posner 1967). The first is the cognitive stage in which the focus is on understanding the task and developing strategies to perform it, requiring cognitive activity such as attention and executive functions. The second phase is the associative stage: the learner has selected the best strategy and now begins to refine the skill. Here, cognitive aspects are less important. And finally, there is the autonomous stage in which the skill becomes automatic, requiring a low degree of attention. Variables such as practice and feedback can be structured differently to enhance learning at each stage. Feedback in the cognitive stage, for example, may need to be more specific and applied more frequently to enhance learning, while feedback may be weaned toward the third stage of learning (Tse and Spaulding 1998). In implicit learning, on the other hand, there is no clear distinction between these three stages. It has been proposed that in implicit learning the three stages might overlap or be ordered differently. There is support for a parallel development of implicit and explicit knowledge in learning (Willingham and Goedert-Eschmann 1999). Practice: Theory and Outcome Studies with AD Patients The principle, “The more you practice, the more you learn,” implies that the amount of practice should be maximized in therapy. But does more practice indeed improve the performance in AD patients? Dick et al. (1995) found that on the Rotor Pursuit both the AD and control group had reached their optimal performance after 40 trials because subsequent practice failed to yield any additional augmenting effect. It would be interesting to determine whether this also holds for other tasks like the Maze test in which, relative to the controls, an inferior amount of learning was observed for AD patients (Kuzis et al. 1999; Sabe et al. 1995). Since fatigue also plays a role in learning, the next question is how to alternate practice with rest to maximize learning in patients. Schmidt and Wrisberg (2000) distinguish two types of practice. In ‘massed practice,’ the greater proportion of the sessions is dedicated to training, while in ‘distributed practice’ the duration of rest equals or is greater than that of practice. To date, the effects of alternating these two training methods in the generally older AD patient group still requires further investigation. Another factor that merits closer attention in the context of training programs for AD patients is whether the task should be learned as a whole or per constituent component. Training the components of a task separately before combining them into the whole pattern can be effective if the task itself can be naturally divided into components that reflect the inherent goal of the task (Schmidt 1988). For example, learning to drive a car can be easily divided into the components “learning to shift gear” and “learning to steer,” which can be trained individually. Learning to reach and grasp an item, on the other hand, does not lend itself well for phased training since reaching and grasping are integral components of a single, continuous movement. The amount of variation in the practice session(s) is also a topic for further study. Task variables like the beanbag’s weight and throwing distance in the Tossing task can be practiced in a random design so that the weight and distance can be varied systematically. Alternatively, they can be offered in a blocked design in which only one task variable per block is practiced repetitively. Another option is to use a constant design in which only one combination of task variables is trained. Note that over time, the connotation of the two terms has shifted: random and blocked practice now refer to the rehearsal of several distinct skills whereas varied and constant practice implies the rehearsal of different variations of the same skill (Schmidt and Wrisberg 2000). Nevertheless, in our report we will use the ‘old’ terms (random, blocked and constant) in their original meanings since these were terms and interpretations used in the reviewed literature. Early evidence suggests that random practice might be most effective for the acquisition and generalizability of a motor skill, whereas during the acquisition of a specific motor skill, performance benefits most from blocked practice (Schmidt 1988). All available studies reviewed on this matter (Dick et al. 1996, 2000, 2003) show that AD patients learn best under constant practice conditions. According to Dick and his 1996 team, humans use their episodic memory of the training trials to accurately perform a task while learning a skill. They suggest that because AD patients experience problems with episodic memory, constant practice is more effective because repeated running of the same motor program does not require an intact episodic memory. The second reason why random practice may be less effective is that other cognitive functions that play a role in random practice, like the ability to switch tasks and divide attention, are affected in AD patients. Dick et al. (1996, 2003) explained the AD patients’ superior learning performance under constant practice conditions in terms of the schema theory originally developed by Schmidt (1975), and likewise propose a more open-loop account of motor control. Schmidt assumes the existence of generalized motor programs (GMPs) that are acquired through practice and that define the “form” of the action. These GMPs can be altered to meet environmental demands by a closed-loop system using sensory feedback. Schemata, e.g., for varying weight and distances in tossing, are learned and allow the action to be scaled to the environment (Schmidt 2003). When they considered their results in terms of this theory, Dick et al. (1996, 2003) concluded that AD patients can develop and access a GMP in training situations that emphasize movement consistency. However, they do not form the motor schemas needed to successfully achieve a movement when the environmental demands change because they are unable to encode and to store the different types of information about a motor pattern. There are three other training approaches that can produce the desired learning effect: guidance, observation, and mental practice (Schmidt 1988). Guidance should only be used at the onset of training because experiments have shown that practice under unguided conditions seems to be more effective for retention and transfer (Shumway-Cook and Woollacott 1995). Observation conveys information about how a skill should be performed and seems to be especially beneficial for the acquisition of new movement patterns (Magill 1993). Our automated computer search and an extra search combining the three keywords with Alzheimer dementia both failed to generate any relevant studies that employed one of these training methods. The only study that provided some additional information on the topic is a report by Dick et al. (1988) which showed that AD patients could recall preselected (subject-defined) movements more accurately than constrained (experimenter-defined) movements on a linear positioning apparatus. This was explained by the patients’ ability to profit from mental preparation of the movement prior to its execution. Without further systematic investigation, however, it cannot be inferred that the ability to profit from mental preparation also means AD patients will profit from mental practice. More research into the effects of all three practice types in AD is needed. Feedback: Theory and Outcome Studies with AD Patients A second crucial variable that influences motor learning is type of feedback. Intrinsic feedback encompasses the sensory information generated by motion, and extrinsic feedback entails information from an external source like a therapist (Schmidt and Wrisberg 2000). There are various ways to provide extrinsic feedback. It can be delivered during or after the movement, immediately following movement completion or delayed, and in a verbal or a non-verbal fashion. It can contain information on average performance (summary feedback) or it may reflect each movement or performance (constant feedback; Schmidt 1988). It is generally believed that constant feedback enhances only motor performance, not the level of learning (Shumway-Cook and Woollacott 1995). With less frequent feedback, learners have to rely more on other cues, which entails more elaborate encoding (Schmidt 1988). Extrinsic feedback can moreover be divided into ‘knowledge of results,’ in which the movement outcome is given in terms of the goal, and ‘knowledge of performance,’ so that the feedback concerns the movement pattern itself (e.g., in a Tossing task: increase the swing of your arm). In almost all studies on motor-skill learning in AD, visual feedback was employed. Only the Maze tasks were administered under blindfolded conditions and the amount of learning in the AD patients proved inferior to the amount found for the controls (Sabe et al. 1995; Kuzis et al. 1999). In most Rotor-Pursuit tasks, the velocity of the target was individualized to equate initial performance. Controls generally tracked at a faster rate than the AD patients (Deweer et al. 1994; Dick et al. 1995; Jacobs et al. 1999; Libon et al. 1998). Possibly, AD patients can only perform this task at a slower rate because they rely more on visual feedback than controls. Only one study using a Rotor-Pursuit task explicitly examined the role of visual feedback on performance in AD patients, showing a drop in performance when the visibility of the moving target was reduced during the learning phase (Dick et al. 2001). In contrast to that of the normal controls, the patients’ performance did not improve across trials in the restricted-vision condition. In the full-vision condition the patients showed normal learning. It can be tentatively concluded that for AD patients, constant visual feedback is important in learning motor skills, but more research is needed to confirm this hypothesis. We did not find any studies that were concerned with the frequency of external feedback, and whether knowledge of results and knowledge of performance makes a difference in this patient group. Based on the results cited above, it may be hypothesized that both forms of feedback knowledge probably place too much weight on the cognitive abilities in AD patients and therefore contribute little to successful performance. Conclusions and Recommendations People with Alzheimer’s disease are able to implicitly (re)learn motor skills to a certain extent and under specific conditions. The experimental research to date shows preserved implicit motor learning irrespective of the task used. Patients are capable of acquiring motor skills without awareness simply by repeated exposure, although their performances will not reach normal levels. This is expressed in their protracted performance relative to that of unimpaired controls. Moreover, extent of learning will differ depending on the task to be mastered. The preserved implicit learning ability in AD can be of use for physical therapists working with this elderly patient group. Physical therapists can call upon neuropsychologists to provide information on their patients’ learning capacities since they have quantitative measures at their disposal to assess a patient’s level of functioning. However, the memory and learning tests currently available in the clinical practice evaluate explicit or declarative memory (Spaan et al. 2003). In order to get a satisfactory differential picture of the learning capacities in demented patients, implicit (motor) learning tasks need to be added to the neuropsychological assessment. The evidence of intact implicit learning in AD further prompted the question how these intact learning abilities can best be translated into rehabilitation programs targeting this patient group. Learning is central in rehabilitation and knowledge of the system under treatment, like the motor system, must be combined with knowledge of how learning principles must be applied to achieve a successful training program (Baddeley 1993). With respect to patients with dementia, apart from the subtype of dementia and its specific neuropsychological syndrome, the training programs should apply the principles that emerge from theories of learning. The studies we reviewed showed that in (re)learning motor skills, constant, or rather frequent and consistent practice is important in AD patients. This way of learning draws less on episodic memory and other cognitive functions compromised in AD patients. These data also suggest that practice under dual-task conditions should also be avoided. Because AD patients have difficulty in generalizing the motor skills learned during the sessions, training has to take place in an environment that closely resembles the one in which the skill is going to be used and presumably with tools used by the AD patient in his or her daily life. If, for instance, an AD patient is trained in the use of a microwave, the device used during the training should be the same as the one available in the patient’s household. The amount of training a patient needs will depend on the task being trained. The role of fatigue is also important in this respect. The effects of massed and distributed practice in this generally older patient group need to be addressed in future investigations. Patients with AD appear to remain dependent upon visual feedback throughout training and performance. Screening and subsequent correction of visual problems or the use of visual aids can be effective in the training process in this group for whom vision problems are very common (De Winter et al. 2004). The type and point in time when external feedback needs to be given and its effect on learning in AD also warrants attention in future research. In the introduction of our review we asked whether patients with Alzheimer’s disease might have intact motor-skill learning abilities. The answer is twofold. Clearly, AD patients show preserved implicit learning abilities that can be utilized in teaching (motor) skills, yet transfer to other skills is minimal. Accordingly, the professionals delivering the training programs should tailor the contents to the particular needs and abilities of this patient group or the individual patient. When the above guidelines are kept in mind and when our knowledge on this topic are widened, non-pharmacological interventions might contribute significantly in helping elderly people suffering from dementia to keep their autonomy. The extent to which pharmacological intervention may enhance these behavioral mechanisms and foster independent living in AD patients has yet to be determined.

Mol_Immunol-1-5-2080686

C1q deficiency promotes the production of transgenic-derived IgM and IgG3 autoantibodies in anti-DNA knock-in transgenic mice

C1q-deficient mice have been shown to develop a lupus-like disease and to display an impaired clearance of apoptotic cells that are enriched in lupus autoantigens. However, the role of C1q in the regulation of autoreactive B cells remains debatable. To explore this we crossed MRL/Mp C1q-deficient mice with knock-in transgenic (Tg) mice expressing an anti-ssDNA antibody (VH3H9R and VH3H9R/VLκ8R). Analysis of the VH3H9R mice showed that in the absence of C1q higher titres of Tg-derived IgM and IgG3 anti-ssDNA antibodies were detectable. In contrast, in the VH3H9R/VLκ8R C1q-deficient animals no increase in Tg antibody levels was observed. In both models the lack of C1q induced a marked reduction of marginal zone B cells and this was paralleled by a significant increase in the percentage of plasmocytes. Thus, one could postulate that in the absence of C1q the failure to clear efficiently dying cells provides an additional stimulus to the autoreactive Tg B cells resulting in their emigration from the marginal zone B cell compartment with subsequent increase in plasmocytes. However, the lack of C1q led to an increased production of Tg IgM and IgG3 antibodies only in VH3H9R mice indicating that additional genetic susceptibility factors are required to break self-tolerance. 1 Introduction Hereditary deficiencies of early components of the classical pathway of the complement system are known to predispose to systematic lupus erythematosus (SLE). Among these, C1q deficiency exhibits the strongest association with prevalence greater than 90% suggesting that a physiological activity of the early part of the classical pathway normally protects against the development of SLE (Pickering et al., 2000). Mice with targeted deletion of the C1q gene (C1qa−/−) developed a spontaneous lupus-like disease characterised by the development of anti-nuclear autoimmunity and glomerulonephritis associated with the presence of multiple apoptotic bodies (Botto et al., 1998). Introgression of C1q deficiency onto different genetic backgrounds revealed that in mice C1q operates as a disease modifier. C57BL/6.C1qa−/− mice displayed no increase of IgG autoantibodies (autoAbs) or glomerulonephritis (Mitchell et al., 2002), whilst C1q deficiency backcrossed onto the lupus-prone MRL/Mp background accelerated both the onset and the severity of the autoimmune disease (Mitchell et al., 2002). Consistent with these observations, C1q reconstitution by bone marrow transplant attenuated the autoimmune disease present in MRL/Mp.C1qa−/− mice (Cortes-Hernandez et al., 2004). Currently, there are two main hypotheses to explain the role of complement in the development of SLE, neither of which is mutually exclusive. The first model, defined as the tolerance hypothesis, proposes a role for complement in determining the activation thresholds of lymphocytes, whereby complement enhances presentation of autoantigens to self-reactive immature B cells resulting in their elimination (Prodeus et al., 1998). The second one, known as the ‘waste disposal’ hypothesis, suggests that in addition to its role in the clearance of immune complexes, complement is involved in the physiological disposal of apoptotic cells (Botto et al., 1998), that have been shown to express lupus autoantigens on their surface (Casciola-Rosen et al., 1994). C1q plays a significant role in the clearance pathway of cellular debris by binding directly or indirectly to apoptotic blebs where it activates complement and mediates phagocytosis by professional and non-professional phagocytes (Korb and Ahearn, 1997; Mevorach et al., 1998; Nauta et al., 2004; Quartier et al., 2005; Taylor et al., 2000). Therefore, improper removal of dying cells in the setting of C1q deficiency could result in the stimulation of autoreactive cells leading to autoimmunity. Immunoglobulin Tg models have been instrumental in understanding B cell regulation revealing several key mechanisms, including receptor editing, deletion, anergy and ignorance. The hen egg lysozyme (HEL)–anti-HEL (IgHEL) Tg model in particular has been widely used to demonstrate elimination of self-reactive clones by membrane-bound expressed antigen, anergy induction by soluble antigen (sHEL) or ignorance when the amount of antigen is so low that it does not reach the threshold to induce anergy. More recently, it has been shown that intracellular membrane-bound HEL failed to induce tolerance and was instead autoimmunogenic positively selecting IgHEL B1 cells and inducing large numbers of IgM autoantibody-secreting plasma cells (reviewed in Ferry et al., 2006). Taken together, these findings suggest that in this model the fate of self-reactive B cells is determined not only by the abundance, the avidity of the target self-antigen and the affinity of the B cell receptor but also by the location of the “auto”-antigen. The role of complement has been tested in the IgHEL Tg (MD4)–sHEL (ML5) model by crossing the double Tg mice with mice deficient in C1q, C4, C3 or CD21/CD35 (Cutler et al., 2001; Prodeus et al., 1998). IgG anti-HEL Abs remained undetectable in all complement deficient mice, but C4 and CD21/CD35 deficient B cells displayed reduced surface IgM modulation, indicating a lower degree of anergy induction in these mice (Prodeus et al., 1998). However, this was not observed in the C1q deficient mice (Cutler et al., 2001). This discrepancy could either indicate that C4 operates independently from C1q or reflect differences in the genetic background of the mice used (Cutler et al., 2001). Nevertheless sHEL is not the ideal model to study the maintenance of tolerance in SLE as it is neither a natural autoantigen, nor are soluble plasma proteins typically targeted by the SLE autoAbs. More recently, a model targeting an SLE antigen, DNA, was generated by Weigert and co-workers (Chen et al., 1995). In this model the rearranged variable heavy chain (VH) gene derived from a double stranded DNA-binding hybridoma developed in the autoimmune strain MRL/Mp.lpr/lpr, was inserted at the Igh locus and was referred to as VH3H9R. In contrast to conventional Tg mice, this knock-in model allows the Tg locus to undergo normal editing, isotype switching and somatic mutation. A variety of light chains can combine with the VH3H9 to yield anti-DNA Abs (Radic et al., 1991) but only few light chains are able to “silence” VH3H9R so that it no longer binds to DNA. By virtue of this characteristic the mice expressing only the VH3H9 chain (VH3H9R mice) can generate anti-DNA specificities. However, when the VH3H9R mice were crossed with a Tg knock-in light chain VLκ8 (Prak and Weigert, 1995) to generate monospecific Tg mice (VH3H9R/VLκ8R mice) (Chen et al., 1997a), this combination of heavy and light chain V regions (VH3H9/VLVκ8) bound only ssDNA and not dsDNA (Prak and Weigert, 1995). Previous studies with the VH3H9R mice have shown that the autoreactive Tg B cells accumulated in the splenic marginal zone and were regulated by anergy on non-autoimmune backgrounds such as BALB/c (Chen et al., 1995; Erikson et al., 1991) and C57BL/6 (Fukuyama et al., 2005; Sekiguchi et al., 2002). However, tolerance could be broken in this model if T cell help was provided in the form of a chronic graft versus host disease (Sekiguchi et al., 2002). Consistent with these observations, the double VH3H9R/VLκ8R knock-in Tg B cells were regulated by anergy in non-lupus prone mice (BALB/c), whilst in autoimmune prone MRL/Mp.lpr/lpr animals Tg B cells escaped tolerance induction and underwent class-switching and affinity maturation (Brard et al., 1999). These experiments suggested that the lpr mutation in the MRL background allowed the Tg autoreactive B cells to receive T cell help during a germinal center reaction. To determine whether C1q is involved in selection of self-reactive B cells, we bred the C1q-deficient mice with the VH3H9R and the VH3H9R/VLκ8R mice and monitored the regulation and activation of anti-DNA Tg B cells over a period of 10 months. The mice in this study were on the autoimmune prone background MRL/Mp expressing the CD95 (Fas) gene. The analysis of these mice revealed that the lack of C1q can influence the levels of IgM and IgG3 Tg-derived antibodies only in the VH3H9R model. 2 Materials and methods 2.1 Mice MRL/Mp mice were obtained from Harlan Olac, Bichester, UK. MRL/Mp.C1qa−/− deficient mice were generated as previously described (Mitchell et al., 2002). VH3H9R.MRL/Mp (Chen et al., 1995), VLκ8R.MRL/Mp (Prak and Weigert, 1995) and VH3H9R/VLκ8R.MRL/Mp mice were kindly provided by Prof. M. Weigert (Gwen Knapp Center for Lupus and Immunology Research, University of Chicago, Chicago, IL). MRL/Mp.C1qa−/− mice were crossed with the VH3H9R/Vκ8R.MRL/Mp mice and the resulting VH3H9R/VLκ8R.MRL/Mp.C1qa+/− were then crossed either with MRL/Mp or MRL/Mp.C1qa−/− in order to generate littermate animals. Mice were bled every 3 months starting from 2 months of age and at 10 months they were sacrificed. The mice were genotyped by PCR using specific primers. PCR primers were as follow: mC1qA/5′ (5′-GGGGCCTGTGATCCAGACAGG-3′), mC1qA/In2− (5′-TAACCATTGCCTCCAGGATGG-3′) and neo (5′-GGGGATCGGCAATAAAAAGAC-3′) for the C1q genotyping; MW114 (5′-CTGTCAGGAACTGCAGGTAAGG-3′) and MW162 (5′-CATAACATAGGAATATTTACTCCTCGC-3′) for the VH3H9R genotyping (Erikson et al., 1991); MW133 (5′-GGTACCTGTGGGGACATTGTG-3′) and MW157 (5′-AGCACCGAACGTGAGAGG-3′) for the VLκ8R genotyping (Carmack et al., 1991). Animals were kept under specific pathogen-free conditions. All animal care and procedures were conducted according to institutional guidelines and approved by the local ethical committee. 2.2 Flow cytometry Flow cytometry was performed using a four-color staining of cells and analyzed with a FACSCalibur™ (Becton Dickinson, Mountain View, CA). The following Abs were used: anti-B220 (RA3-6B2), anti-CD5 (53-7.3), anti-CD11b (M1-70), anti-CD19 (1D3), anti-CD23 (B3B4), anti-CD21/CD35 (7G6), anti-CD90.2 (53-2.1), anti-CD138 (281-2), anti-IgM (II/41). All Abs were purchased from BD Biosciences Pharmingen (San Diego, CA) with the exception of the anti-VH3H9 idiotype (1.209), a kind gift from Prof. M. Weigert (Gay et al., 1993). Biotinylated Abs were detected using an allophycocyanin-conjugated streptavidin Ab (BD Biosciences Pharmingen). Staining was performed in the presence of saturating concentration of 24G2 mAb (anti-FcRII/III). Data were analyzed using WinMDI software (Version 2.8; Scripps Institute). 2.3 Serological analyses Serum levels of IgM and IgG were quantified by ELISA, as described previously (Cortes-Hernandez et al., 2004). Anti-ssDNA Abs, anti-chromatin Abs, anti-histone Abs and anti-dsDNA Abs were measured by ELISA as described previously (Burlingame and Rubin, 1990; Emlen et al., 1990). Microtiter plates were coated with ssDNA prepared from calf thymus DNA (Sigma), chromatin (Lorne Laboratories Ltd., Reading, UK) or histone (Calbiochem, Merck Biosciences, Darmstadt, Germany). For detecting anti-dsDNA Abs, plates were coated with streptavidin (Sigma). ΦX174 double-stranded plasmid DNA (Promega, Southampton, UK) was biotinylated with Photoprobe biotin (Vector Laboratories, Peterborough, UK) and added to the streptavidin. Serum samples were diluted appropriately in PBS 2%BSA, 0.05%Tween-20, 0.02% NaN3. Bound Abs were detected with alkaline phosphatase conjugated goat anti-mouse IgG (γ-chain specific) (Sigma–Aldrich, Dorset, UK) and anti-mouse IgM (Southern Biotechnology Associates, Inc., Birmingham, AL). All autoAb results are expressed in arbitrary ELISA units (AEU) in reference to a standard curve derived from serum pools containing high titers of autoAbs. The idiotype+ Abs were captured with the anti-VH3H9 idiotype (1.209) (Gay et al., 1993) and detected with alkaline phosphatase conjugated goat anti-mouse IgM or IgG subclasses specific Abs (Southern Biotechnology Associates). The standard curve was derived from an IgM VH3H9R/Vκ8R mAb provided by Prof. Weigert or sera containing high IgG subclasses titers of the Tg. 2.4 Renal assessment Proteinuria was assessed using Haema-combistix (Bayer Diagnostics, Newbury, UK). Kidneys were fixed in Bouin's solution and paraffin embedded, and sections were stained with periodic acid–Schiff reagent. Glomerular histology was graded in a blind fashion scored on a scale of 0–4, as described before (Carlucci et al., 2007). 2.5 Statistics The data are presented as median, with range of values in brackets, unless otherwise stated. The non-parametric Mann–Whitney U-test was applied throughout with differences being considered significant for p-values < 0.05. Statistics were calculated using GraphPad Prism Version 3.0 (GraphPad Software, San Diego, CA). 3 Results 3.1 Expression of Tg autoantibodies C1q deficiency has been shown to accelerate the onset and progression of SLE in MRL/Mp mice (Mitchell et al., 2002), but the mechanisms underlying these effects are still uncertain. The VH3H9R/VLκ8R Tg alleles were transferred onto C1q-deficient mice by crossing VH3H9R/VLκ8R.MRL/Mp with MRL/Mp.C1qa−/− and experimental cohorts of littermate mice were generated. Whilst in previous studies the expression of the Tg alleles was monitored by measuring the Tg-specific immunoglobulin allotype (Steeves and Marion, 2004), this analysis could not be applied in these mice as the Tg alleles and the endogenous immunoglobulin allotype were indistinguishable. Nevertheless, we were able to detect the combination of the VH3H9 with the VLVκ8 using an idiotype specific monoclonal Ab named 1.209 (Gay et al., 1993). As the 1.209 mAb can also recognize the VH3H9 paired with other endogenous light chains such as VLκ4 (Gay et al., 1993), this anti-idiotype Ab allowed us not only to assess the VH3H9R/VLκ8R Tg B cells but also some of the VH3H9R Tg B cells. C1q deficiency did not affect the proportion of peripheral blood B cells expressing the Tg in neither of the two models (VH3H9R/VLκ8R.MRL/Mp.C1qa−/−: 89.1 ± 1.1% versus VH3H9R/VLκ8R.MRL/Mp: 90.7 ± 1.5%, p = 0.4118; VH3H9R.MRL/Mp.C1qa−/−: 25.3 ± 1.6% versus VH3H9R.MRL/Mp: 22.3 ± 2.3% p = 0.1669). The low proportion of idiotype+ B cells in the VH3H9R.MRL/Mp mice indicated that most of the B cells in these mice had either paired the VH3H9R Tg allele with a light chain that was not recognised by the anti-idiotype Ab, or edited the VH3H9R allele, or used the endogenous immunoglobulin allele. In the VH3H9R/VLκ8R.MRL/Mp.C1qa−/− mice the B cell receptor was down-modulated to the same extent as in the VH3H9R/VLκ8R.MRL/Mp mice (data not shown). Serological analyses were performed in order to establish whether the Tg autoAbs were expressed. The different cohorts of mice were bled at different time points, and the data presented here are those obtained at 10 month of age when the mice were sacrificed. Substantial levels of idiotype+ IgM, IgG3, IgG2a and IgG2b Abs were detected in the VH3H9R/VLκ8R.MRL/Mp and in VH3H9R.MRL/Mp mice indicating that the anti-DNA knock-in Tg alleles on this background were not regulated by anergy unlike when on the C57BL/6 background (Fig. 1). In the VH3H9R/VLκ8R.MRL/Mp.C1qa−/− mice the levels of idiotype+ Abs were similar to those detected in the strain-matched C1q-sufficient animals. In contrast, the VH3H9R.MRL/Mp.C1qa−/− mice had significantly increased levels of idiotype+ IgM and IgG3 Abs compared to the VH3H9R.MRL/Mp mice (Fig. 1A and B), but similar levels of idiotype+ IgG2a and IgG2b Abs (Fig. 1C and D). We then measured the levels of IgM and IgG autoAbs. A significant increase in the titre of IgM anti-ssDNA Abs was observed in the VH3H9R/VLκ8R.MRL/Mp.C1qa−/− mice compared to the VH3H9R/VLκ8R.MRL/Mp mice (Fig. 2A). Of note, the VH3H9R/VLκ8R.MRL/Mp.C1qa−/− mice displayed only markedly increased levels of IgM autoAbs directed against ssDNA (the specificity encoded by the Tg alleles) but not against other lupus autoantigens such as dsDNA, histone or chromatin (Fig. 2B–D). In the VH3H9R.MRL/Mp mice the pairing of the VH3H9 with endogenous L chains could yield to more specificities. Consistent with this, VH3H9R.MRL/Mp.C1q−/− mice had significantly increased levels of IgM anti-ssDNA, anti-dsDNA, anti-histone and anti-chromatin Abs (Fig. 2A–D). We then analysed the levels of IgG autoAbs and found that these were similar between the C1q-sufficient and -deficient cohorts (Supporting information Table S1). 3.2 Renal assessment At 10 month of age all the animals were sacrificed. VH3H9R/VLκ8R.MRL/Mp and VH3H9R.MRL/Mp animals showed histological evidence of a mild glomerulonephritis (median score 1.0, range: 0.0–3.0 and 1.5, range: 1.0–3.0, respectively) reminiscent of the disease observed in MRL/Mp non-Tg mice (Mitchell et al., 2002). C1q-deficiency in MRL/Mp mice has previously been shown to worsen the kidney pathology (Mitchell et al., 2002). However, in the Tg mice the absence of C1q did not exacerbate the renal disease compared to the strain-matched C1q-sufficient mice (VH3H9R/VLκ8R.MRL/Mp.C1qa−/− mice: median score 1.0, range: 0.0–3.0, p = 0.4287; VH3H9R.MRL/Mp.C1qa−/− mice: median range 2.0, range: 1.0–2.0, p = 0.4564). 3.3 Flow cytometric analysis of splenic and peritoneal B cells In order to determine if the serological data were accompanied by phenotypic changes in T and B lymphocytes, we performed a comprehensive analysis of the various splenic and peritoneal subpopulations. Cells of at least 7 mice from each cohort were analysed by FACS at the time of the sacrifice using the combinations of markers shown in Table 1. In agreement with the findings in other genetic backgrounds (Li et al., 2002), the VH3H9R.MRL/Mp mice had a larger marginal zone (MZ) B cell population compared to VH3H9R/VLκ8R.MRL/Mp (Fig. 3A) and an increased proportion of idiotype+ B cells in the MZ compartment compared to follicular (FO) compartment (5.9 ± 1.0 versus 4.2 ± 0.6, paired t-test p = 0.0157) (Table 1). More interestingly, in the VH3H9R.MRL/Mp.C1qa−/− mice, which produced higher levels of IgM autoAbs, the MZ B cell population was significantly decreased compared to the VH3H9R.MRL/Mp animals (Fig. 3A) and less idiotype+ B cells were found in the MZ compared to the FO compartment (3.4 ± 0.7 versus 4.8 ± 1.1, paired t-test p = 0.0042) (Table 1). In the VH3H9R/VLκ8R.MRL/Mp mice there was no preferential localization of idiotype+ B cells in the MZ zone (paired t-test p = 0.056) (Table 1). C1q deficiency was associated with a significant reduction in the percentage of MZ B cells (Fig. 3A) and a decrease of idiotype+ B cells in the MZ compared to the FO area (4.2 ± 0.9 versus 17.5 ± 3.2, paired t-test p = 0.0013). These findings indicated that MZ B cells might have been activated in the absence of C1q. Activated MZ B cells have been shown to migrate into the T cell zone and differentiate into plasma cells. Indeed, the percentage of plasma cells was increased in the two C1q-deficient cohorts compared to the respective MRL/Mp controls (Fig. 3B). The peritoneal cavity in mice has been shown to be a site where some self-reactive B cells can escape tolerance (Fagarasan et al., 2000) and B-1 cells, one of the major sources of circulating IgM, accumulate. More cells were recovered from the peritoneum of VH3H9R.MRL/Mp.C1qa-/-mice compared to VH3H9R.MRL.Mp mice (p = 0.0051) but not in the VH3H9R/VLκ8R.MRL/Mp.C1qa−/− versus VH3H9R/VLκ8R.MRL/Mp animals (p = 0.1923) (Table 1). When the peritoneal B cell subpopulations were analyzed in more detail, all subsets were increased in the VH3H9R.MRL/Mp.C1qa-/mice compared to the VH3H9R.MRL.Mp mice (Table 1). On the contrary, in the VH3H9R/VLκ8R C1q-deficient animals only B-1a cells were slightly increased compared to their wild type strain-matched controls (Table 1), but this did not reach statistical significance (p = 0.0518). We then analysed the VH3H9/VLκ8 idiotype expression on the different peritoneal B cells. Surprisingly in the peritoneum of the VH3H9R/VLκ8R.MRL.Mp and VH3H9R.MRL.Mp mice the majority of the B cells were not expressing the VH3H9/VLκ8 idiotype (Table 1). Of note the C1q-deficient mice had similar percentage of idiotype+ B cells in the different peritoneal B cell subpopulations compared to C1q-sufficient mice. The reduced proportion of idiotype+ B cells in the peritoneum compared to the peripheral blood and the spleen (data not shown) suggests that in this organ the B cells either preferentially used the endogenous alleles or had been activated and undergone editing. 4 Discussion C1q deficiency in humans and in mice has been associated with the development of a lupus-like illness. However, the role of C1q in the regulation of autoreactive B cells remains debatable. Here, we explored the regulation of autoreactive B cells by C1q using anti-DNA knock-in Tg models (VH3H9R/VLκ8R and VH3H9R) on the lupus prone MRL/Mp genetic background. The analysis of these mice revealed that the MRL/Mp background was in itself sufficient to allow the expression of anti-DNA Tg autoAbs and that the lack of C1q modified this effect only in the single Tg model (VH3H9R). However, in the absence of C1q the MZ B cell compartment was significantly reduced in both models and this was accompanied by an increase in plasmocytes. Recently the VH3H9R/VLκ8R anti-DNA knock-in Tg model has been widely used for investigating the mechanisms of regulation of autoAb production in murine models of SLE. There is an increasing evidence that the VH3H9R/VLκ8R and VH3H9R anti-DNA Tg B cells, which have a relatively weak affinity to ssDNA, are regulated by anergy on normal genetic backgrounds such as BALB/c and C57BL/6, but could be induced to lose tolerance when transferred onto lupus models such as (NZB × NZW)F1 or MRL/Mp.lpr/lpr (Brard et al., 1999; Chen et al., 1995; Erikson et al., 1991; Fukuyama et al., 2005; Sekiguchi et al., 2002). In light of these observations it was important for our study to establish whether the autoimmune prone MRL/Mp background was in itself capable of breaking tolerance. MRL/Mp mice are known to develop a mild autoimmune disease which can be accelerated with different disease-modifying genes such as Yaa (Merino et al., 1989), lpr and gld (Cohen and Eisenberg, 1991). The analysis of the idiotype+ (VH3H9R/VLκ8R) Abs revealed that the Tg MRL/Mp mice indeed had in circulation these idiotype+ Abs (IgM, IgG2a, IgG2b, IgG3) indicating that a break of tolerance had spontaneously occurred in these mice. One explanation for this is an intrinsic defect in MRL/Mp B cells and there is some evidence in support of this. MRL/Mp mice have been reported to exhibit a defect in maintaining the developmental arrest of VH3H9/VLλ anti-dsDNA conventional Tg B cells (Mandik-Nayak et al., 1999, 2000) and to have a spontaneous B cell hyperactivity in the absence of Ag in the IgHEL experimental model (Nijnik et al., 2006). However, the VH3H9/VLλ anti-dsDNA autoreactive B cells, despite not being any longer developmentally arrested as in a BALB/c mice, exhibited follicular exclusion and failed to differentiate into plasma cells (Mandik-Nayak et al., 1999). Furthermore Tg MRL/Mp mice expressing IgHEL have been shown to be able to down-regulate their B cell receptor and to be unable to secrete detectable levels of anti-HEL Abs in the presence of sufficient amount of sHEL (Nijnik et al., 2006). Similarly anti-laminin Tg MRL/Mp mice were found to be tolerant (Rudolph et al., 2002). Another potential explanation for the break of the B cell tolerance in the VH3H9R/VLκ8R anti-DNA Tg mice, is that in this model the MRL/Mp background was able to provide sufficient T cell help and the presence of idiotype+ IgG subclasses favour this hypothesis. We next examined whether C1q could modulate the phenotype of the anti-ssDNA Tg B cells. In the VH3H9R/VLκ8R.MRL/Mp mice the absence of C1q increased significantly the circulating levels of IgM against ssDNA but not against other autoantigens including dsDNA. As the pairing of VH3H9 with VLκ8 prevents the binding of VH3H9 to dsDNA, the elevated amount of IgM anti-ssDNA observed could have been the result of an increased Tg expression. However, the idiotype analysis failed to demonstrate a difference in the levels of IgM idiotype+ Abs between VH3H9R/VLκ8R.MRL/Mp.C1qa−/− and VH3H9R/VLκ8R.MRL/Mp mice, questioning whether the source of the increased levels of IgM anti-ssDNA was indeed the Tg B cells. On the other hand, the VH3H9R.MRL/Mp.C1qa−/− mice displayed significantly higher levels of IgM and IgG3 idiotype+ Abs. As the VH3H9R heavy chain can pair with different endogenous light chains generating a wider range of autoAbs, other specificities were tested. Indeed the absence of C1q increased significantly IgM levels against all the lupus autoantigens analysed. To gain insight into the mechanisms regulating the autoAb production in the C1q-deficient mice we then carried a detailed analysis of the different B cell populations. Several studies have proposed a possible role for MZ B cells in the development of lupus in mouse models. Although studies in immunoglobulin Tg mice have shown that autoreactive B cells can accumulate in the marginal zone under various experimental situations (Li et al., 2002; Mandik-Nayak et al., 1997, 1999, 2006; Qian et al., 2004; Wen et al., 2005), it remains to be established whether MZ B cells secrete pathogenic autoAbs in any model of lupus. B cells producing potentially pathogenic autoAbs are thought to home to the MZ (Chen et al., 1997b; Li et al., 2002) and sequestration to this site is believed to prevent them from entering into the germinal centres and developing the properties of pathogenic B cells. However, recent studies in lupus-prone mice have reported both enlargements (Grimaldi et al., 2001; Mackay et al., 1999; Wither et al., 2000) and impaired development of the MZ B cell compartment (Amano et al., 2003; Samardzic et al., 2002). Consistent with previous observations in the VH3H9R/VLκ8R model (Wen et al., 2005), the MZ was found to be enlarged in the VH3H9R.MRL/Mp mice but no accumulation of MZ B cells was observed in VH3H9R/VLκ8R.MRL/Mp mice. Importantly C1q deficiency decreased dramatically the proportion of MZ B cells in these mice and this was accompanied by an increase in the percentage of plasma cells. More interestingly, in the C1q-deficient Tg mice we observed a significant disappearance of idiotype+ B cells from the MZ suggesting that these cells had been activated. Marginal zone B cells bearing low affinity self-reactive BCR can react to repetitive Ag and produce natural autoAbs of the IgM isotype upon contact with blood-borne pathogens or self Ag (Oliver et al., 1999). Moreover, MZ B cells can undergo T cell independent switching to IgG, IgA and IgE in response to pathogen associated molecular patterns (PAMP). As C1q has been shown to be involved in the clearance of apoptotic cells, which are enriched in the typical lupus autoantigens, one could postulate that the increased IgM Tg secretion and the reduction of the MZ B cells in the VH3H9R.MRL/Mp.C1qa−/− mice might be related to a failure to clear antigens associated with dying cells. In the absence of C1q the ineffectively cleared autoAgs could stimulate the autoreactive MZ Tg B cells to differentiate into plasmocyte resulting into the decrease of MZ B cells and increase of circulating IgM autoAbs. Similarly, in the spleens of VH3H9R/VLκ8R.MRL/Mp.C1qa−/− mice significantly less idiotype+ B cells were present in the MZ compartment and more plasmocytes were found. However, these cellular changes were not paralleled by an increase of idiotype+ IgM Abs in circulation indicating in this model they were not sufficient to induce autoantibody production. Another potential source of IgM autoAbs are the B1 cells. B1 cells have long been associated with the secretion of natural Abs against self and foreign pathogens, which can occur without obvious inflammatory response. The importance of B1 cells in the pathogenesis of lupus continues to be debated. A major argument for a role of B1a cells in mouse lupus relates to the expansion of this subset in NZB and (NZB × NZW)F1 mice. Reduction of B1 cells for instance via intraperitoneal injection of H2O delayed disease onset and reduced disease severity in (NZB × NZW)F1 mice (Murakami et al., 1995). However, the expansion of B1a cells was shown not to be critical for the production of IgM or IgG autoAbs in this murine model of SLE (Atencio et al., 2004). The Abs secreted by B1 cells tend to be polyreactive with low-affinity cross-reactivity to a variety of self Ags and these characteristics are very different from the pathogenic IgG Abs produced by lupus mice. In our experimental model, we found an increase in the total number of peritoneal and B1 cells in the VH3H9R.MRL/Mp.C1qa−/− mice compared to their wild type counterparts (Table 1). A similar trend was observed for the B1a cells in the VH3H9R/VLκ8R.MRL/Mp.C1qa−/− mice but this did not reach statistical significance. However, in both models the percentages of idiotype+ B1 cells between C1q-deficient and -sufficient mice were similar. These findings would indicate the lack of C1q may favour the expansion of peritoneal B1 cells and that these cells might have contributed to the increase of serum IgM autoAb levels observed in the VH3H9R.MRL/Mp C1q-deficient mice. Supporting this hypothesis, a recent study showed that C1q deficiency increases the positive selection of B-1 cells and IgM autoAb production by a membrane-bound intracellular auto-Ag (Ferry et al., 2007). In conclusion, using mice expressing site-directed transgenes for anti-DNA autoAbs we have shown that: (i) the MRL/Mp background was in itself capable of inducing the expression of anti-DNA Tg autoAbs, and (ii) VH3H9R.MRL/Mp.C1qa−/− could influence the production of Tg-derived IgM and IgG3 autoAbs possibly as a result of an impaired disposal of cellular debris. However, we found no evidence of a direct role of C1q in the regulation of self-reactive conventional B cells. Further studies on non-lupus prone genetic backgrounds such as C57BL/6 will be necessary to determine if C1q deficiency can play a more substantial role in shaping the repertoire of the autoreactive B cells.

Ulster_Med_J-76-3-2075581

The use of mechanical bowel preparation in elective colorectal surgery

Background Mechanical bowel preparation (MBP) prior to elective colorectal surgery has been in use for many years. It is considered important in preventing post-operative infectious complications after colorectal surgery. The evidence to support these claims is lacking within the medical literature and yet this still remains standard practice in many hospitals. A literature search was undertaken to ascertain the evidence available regarding the use of MBP in elective colorectal surgery. BACKGROUND Mechanical bowel preparation (MBP) prior to elective colorectal surgery has been in use for many years. Early observational studies and long-standing clinical experience have shown that removal of faecal matter from the bowel lumen prior to surgery has been associated with decreased patient morbidity and mortality1. It is still commonly used in routine practice today2. In fact, in a recent survey of members of the American Society of Colon & Rectum Surgeons, 99% of respondents routinely use MBP although 10% question its use3. This is in keeping with common belief that clinical practice often is not evidence based but is based on tradition, previous teaching and anecdote. MBP is considered important in preventing post-operative infectious complications after colorectal surgery2,4–10. Important infectious complications include wound infection, intra-abdominal abscess formation and anastomotic leakage. There are a number of ways in which MBP is thought to act. It may decrease intraoperative contamination with faecal material thereby reducing the incidence of post-operative wound infection and residual intra-abdominal infection6,7,9,10. It may prevent mechanical disruption of the anastomosis by the passage of hard faeces9 and improves the handling of the bowel intra-operatively2,7. It may reduce the bacterial count within the colon7,10. Conversely, it may also be associated with bacterial translocation through the bowel wall hence possibly contributing to post-operative infectious complications5,11. The evidence to support these claims is lacking within the medical literature and yet this still remains standard practice in many hospitals4,12. Primary colonic anastomosis is considered unsafe in unprepared bowel but there is little data to suggest that infectious complications are decreased by MBP10. Bowel preparation is unpleasant for patients and can be associated with complications such as dehydration, nausea, vomiting, mucosal lesions, hypokalaemia and other electrolyte disturbances1,9,11. The omission of this practice from pre-operative preparation would be welcomed by nursing staff and patients alike.? METHODS A literature search was undertaken to ascertain the evidence available regarding the use of MBP in elective colorectal surgery. This included a search of PubMed, Medline and Embase using the keywords “mechanical bowel preparation”, “bowel cleansing” and “elective colorectal surgery”, a search of recent relevant journals including Diseases of the Colon and Rectum and British Journal of Surgery and backward chaining from articles obtained. The search was restricted to English language articles and a timescale of 10 years was chosen to give a balanced view of this topic. In this review, mechanical bowel preparation will be defined as an oral preparation given prior to surgery to clear faecal material from the bowel lumen. There are a number of different preparations available including polyethylene glycol, mannitol and sodium picosulphate. Rectal enemas may also be administered before low anterior resections to ensure that the rectum is empty. Elective colorectal surgery is defined as any surgery undertaken on a planned basis for any condition of the colon or rectum requiring bowel resection and primary anastomosis. This will include colorectal carcinoma and inflammatory bowel disease. There are a number of recent randomised controlled trials (RCTs) to evaluate the use of MBP prior to elective colorectal surgery,10,11,13,14 and specifically for left-sided resections5. Many of these studies are underpowered therefore introducing the possibility of a Type II error and limiting the use of these results in clinical decision-making (Table I). Table I RCTs examining MBP Zmora 2003 Fa-Si-Oen 2005 Ram 2005 Bucher 2005 Miettinen 2000 No. of patients included 415 250 329 153 267 No. of patients excluded 35 0 Not given 0 12 No. of pt (MBP/no MBP) 187/193 125/125 164/165 78/75 138/129 Mean age (MBP/no MBP) 68/68 68/70 (median) 68/68 63/63 61/64 Cancer % (MBP/no MBP) 78/78 90/92 75/88 32/28 46/55 L colon surgery % (MBP/no MBP) 68/72 48/58 89/85 100/100 45/47 Type of prep Polyethylene glycol Polyethylene glycol Sodium phosphate Polyethylene glycol Polyethylene glycol Antibiotic Yes Yes Yes Yes Yes Same length of prophylaxis No Yes Yes No Yes Rectal enema Yes No No Yes No Anastomosis % (stapled / handsewn) Not given 7/93 (MBP) 8/92 (no MBP) 94/6 (MBP) 98/2 (no MBP) Not given 60/30 (MBP) 62/28 (no MBP) Surgeon/trainee % Not given 42/59 (MBP) 50/50 (no MBP) 37/63 (MBP) 32/68 (no MBP) Not given Not given Anastomotic leak % (MBP / no MBP) 3.7/2.1 (NS) 5.6/4.8 (NS) 0.6/1.2 (NS) 6/1 (NS) 4/2 (NS) Wound infection % (MBP / no MBP) 6.4/5.7 (NS) 7.2/5.6 (NS) 9.8/6.1 (NS) 13/4 (NS) 4/2 (NS) Intra-abdominal abscess % (MBP / no MBP) 1.1/1 (NS) Not given 0.6/0.6 (NS) 1/3 (NS) 2/3 (NS) NS = not significant This lack of power in studies is somewhat overcome by the use of meta-analyses and systematic reviews of the literature but the reader must be aware that these methods also have their limitations. A number of meta-analyses and systematic reviews were used in this review1,2,7–9,15. DISCUSSION Six systematic reviews were identified in the literature assessing the role of MBP in preventing infectious complications following colorectal surgery1,2,7–9,15. The meta-analysis carried out by Platell & Hall7 found a statistically significant increase in the incidence of wound infection in those patients receiving MBP when considering the three included RCTs. This may have been influenced by the rate of wound infection seen in one trial that used a five day regime of MBP. The anastomotic leakage rate was also higher in the MBP group but not significantly so. Each RCT used a different type of MBP and this lack of standardisation affects the validity of the results. The included studies were also underpowered thereby introducing a high possibility that they failed to detect a significant difference in the results (type II error). Evidence from the prospective and retrospective studies was in favour of no MBP in pre-operative period. A subsequent review of the literature by Zmora et al9 appraised four RCTs. One of the studies found an increased risk of anastomotic leakage and intra-abdominal infection but no increased risk of wound infection in the group of patients receiving MBP. The remaining RCTs found no significant difference in intra-abdominal infection rate but a slight increase in wound infection rate in the MBP group. There were conflicting results in the non-randomised studies with some showing an increased rate of infection and others reporting no difference in infection rates between the groups. The remaining four meta-analyses only included RCTs1,2,8,15. All authors agreed that MBP was of no benefit in preparation for colorectal surgery and it may be detrimental to the patients' outcome1,2,8. Wille –Jorgensen et al15found that although initial analysis showed a significantly higher rate of anastomotic leakage in the MBP group, this significance disappears when sensitivity analyses are applied thereby weakening the conclusion that MBP leads to an increased rate of anastomotic leakage. Slim et al8 found there was significantly more anastomotic leakage in the group of patients receiving MBP and a tendency to a higher rate of wound infection but this was not statistically significant. This group repeated the analysis excluding the poor quality trials and the results still favoured a no MBP regime although this was not statistically significant. Only two meta-analyses looked at MBP in rectal surgery specifically1,15. Willie-Jorgensen et al15 found that when results were stratified for colonic and rectal surgery there was no trend in either direction. Guenaga et al1 found that the results of stratification favoured no MBP but this was not statistically significant. This is of more clinical importance as it may be difficult to perform a low anterior resection and anastomosis with a loaded rectum15. Both authors suggest that further trials evaluating the use of rectal preparation with enemas may be useful. Guenaga et al1also mention that the use of pre-operative radiotherapy would be an important consideration in assessment of bowel preparation for rectal surgery as many patients with rectal cancer undergo pre-operative radiotherapy. All but one of the five RCTs examining MBP5,10,11,13,14 found no significant difference in the rate of anastomotic leakage and wound infection between patients receiving MBP or not10,11,13,14. The largest trial was undertaken by Zmora et al10 with 415 patients recruited. There are several flaws in the methodology of this trial introducing bias and compromising the validity of the results. There was no difference found in the rate of post-operative infectious complications between the two groups. The rate of diarrhoea post-operatively was significantly more common in the group receiving MBP but this is of little clinical significance, as many patients will experience an increased stool frequency once the bowels become active. The authors acknowledge that separating the role of MBP in post-operative infection rate is difficult and ideally all other measures should be constant. They also note that the study is underpowered to detect a 5% difference in infection rate. The RCT conducted by Bucher et al5 comparing MBP with no MBP in patients undergoing elective left-sided colorectal surgery found an increase in the total incidence of infectious abdominal complications in the group receiving MBP (22% v 8%; p=0.028). This led the authors to conclude that there was good evidence to suggest that the practice of MBP should be re-evaluated. They gave an enema pre-operatively to all patients undergoing an anterior resection regardless of whether they had been randomised to MBP or not, decreasing the internal validity of the results. If anastomotic leak rate (a more clinically important outcome than wound infection) were to be used as the primary end-point then the study would need 514 patients in each group. The trial conducted by Ram et al14 was not properly randomised, introducing methodological bias and limiting the value of the results of this study. There was no definition of sample size and patients with low rectal anastomosis were excluded. Again, the assessor of outcome was not blinded to the intervention, introducing another source of bias. No statistically significant difference in the frequency of infectious complications was observed between the groups yet the authors concluded that “mechanical bowel preparation is unnecessary for safe elective colonic and colorectal surgery". But they recommend MBP in selected cases including the resection of small tumours when palpation of the colon may be necessary or when intra-operative colonoscopy may be performed. Fa-Si-Oen et al13 conducted a well-designed multi-centre RCT, reported in 2005. Approximately half the resections carried out in this study were left-sided. This is important as it is now generally accepted that right-sided anastomosis is safe without MBP. This study excluded patients undergoing rectal surgery. There was no significant difference in wound infection or anastomotic leak rate but the bacterial swab results used to define wound infection in this study were only correctly obtained in 185 out of 250 patients therefore this may not be an accurate reflection of the true rate of wound infections. This study could not demonstrate an additional protective effect for MBP but it was an interim analysis and was underpowered. As a result, conclusions for clinical practice cannot be drawn from these results. Miettinen et al11 reported the results of a prospective, randomised study including patients undergoing rectal surgery. There was no significant difference in infectious complications found between the two groups but it is difficult to conclude on the influence on anastomotic leakage from these results as the study included patients who did not undergo an anastomosis. A number of these reported trials are underpowered thereby limiting their ability to detect a clinically significant difference in outcome between the two study groups10,11,13. One way of overcoming the problem of small sample sizes is to carry out a multi-centre trial where a larger number of patients are easier to recruit. A limitation is that they introduce heterogeneity in operative and peri-operative techniques. This is important, as surgical technique may be the single most important factor in influencing the surgical outcome9. All these studies agree that elective colorectal surgery may be safely performed without MBP and that there is no evidence to continue this invasive practice with potentially negative side effects. Memon et al16 carried out a retrospective non-randomised trial based on operating surgeon preferences using a questionnaire. The validity of this questionnaire is unclear as no pilot study was carried out prior to the collection of definitive data. Follow-up of the patients was obtained using the hospital records therefore relying on accurate clinical notes, which are not always available. One hundred and thirty six patients who underwent elective left-sided colorectal procedures for non-obstructive large bowel pathologies were identified using the hospital computer system. Coding errors may mean that some eligible patients were excluded from the analysis. This, along with the lack of randomisation, would introduce significant bias. No statistical difference was found between the two groups for all infectious complications and mortality. The authors recognise the limitations of their results and do not recommend any changes in practice but do suggest that a prospective randomised trial should be performed to demonstrate the impact of MBP on morbidity and mortality in patients undergoing elective colorectal surgery. A prospective, observational trial performed by van Geldere et al17 assessed the outcome of 250 consecutive patients who underwent resection and primary anastomosis of the colon and upper rectum under the care of a single surgeon. None of these patients received MBP pre-operatively. Both emergency and elective procedures were included in analysis. Results were favourable with an overall wound infection rate of 3.3% and an anastomotic failure rate for left-sided resections of 1.2%. The authors recommend that more powerful randomised trials are needed but in the hands of a single surgeon, primary anastomosis of unprepared bowel is safe with relatively few complications. A small observational study conducted by Ahmad et al4 found an anastomotic leak rate of 4.2% and a wound infection rate of 8.5%. The average age of the sample was lower than that of the typical population undergoing colorectal surgery. This fact, plus the small sample size, compromises the external validity or the extent to which the results can be generalised to other samples or situations. CONCLUSION There are a number of meta-analyses, systematic reviews and RCTs looking at the efficacy of MBP in preventing post-operative infectious complications following elective colorectal surgery. Unfortunately many of these trials are underpowered and have a high chance of a type II error10,11,13. Most authors recommend that colorectal surgery is safe without pre-operative MBP but that there may some situations in which it may be beneficial (e.g. if there is a small tumour or the possible need for intra-operative colonoscopy)14. The implication for clinical practice in this situation is that there is not enough strength of evidence at present to recommend a change in practice. There is a need for further higher powered trials to try to answer this question definitively. The only way that this may be achieved is by multi-centre trials where it is easier to recruit a large number of patients but it must be taken into consideration that this will introduce heterogeneity in the operative and peri-operative techniques which may have an influence on overall outcome9. There is a need for larger clinical trials in this area to address whether MBP, with its potential side effects, is truly necessary prior to elective colorectal surgery. Further studies are required to assess the use of rectal preparation alone prior to rectal surgery1,15 and also to include patients who have undergone pre-operative radiotherapy1 as this is a common occurrence in patients who have rectal carcinoma and these patients may subsequently undergo resection with primary anastomosis. It is clear that further research is needed to clarify the role of MBP in elective colorectal surgery to ensure that the patients are receiving the most appropriate treatment with the least adverse effects.

Qual_Life_Res-3-1-2039860

Demographic characteristics and quality of life of patients with unexplained complaints: a descriptive study in general practice

Objective About 13% of GPs’ consultations involve unexplained complaints (UCs). These complaints can progress to chronic conditions like medically unexplained symptoms, chronic functional symptoms or somatoform disorders. Little is known about the demographic characteristics and quality of life of patients with early stage UCs. Our study objective was to describe these characteristics. Additionally we compared them with other patient groups to serve as a frame of reference. Introduction A patient’s complaints are said to be unexplained if the general practitioner (GP) cannot decide on a specific diagnosis after adequate history taking, physical examination and careful consideration of the patient’s psychosocial context [1]. On average, 13% of consultations involve complaints considered unexplained by Dutch GPs [2]. These unexplained complaints often concern fatigue, abdominal and musculoskeletal complaints. It is often assumed that unexplained complaints are mild and self-limiting, because patients do often not revisit their GP for them after a first consultation [3, 4]. However, an unexplained complaint can be a first sign of somatic or psychosocial pathology or a precursor of more chronic unexplained complaints such as chronic medically unexplained symptoms (MUS), chronic functional symptoms or somatoform disorders [5]. Patients with chronic unexplained conditions often have high levels of medication and other healthcare use and frequently show significant psychological distress [6]. Patients with multiple medically unexplained symptoms (MMUS) have impaired quality of life [7]. Furthermore, patients often do not feel taken seriously, do not feel helped and feel treated as malingerers. They, therefore, tend to lose trust in their GPs [8]. In brief, the long-term consequences of many unexplained complaints are unfavourable. GPs too consider unexplained complaints problematic. One study found that in consultations with such patients GPs feel frustrated and helpless though devoted to help [9]. Research on unexplained complaints is increasing. So far, most research has focused on patients with chronic consequences of unexplained complaints. This research varies from discussions on nomenclature and taxonomy [10–12], to research on the characterization of these patient groups [13, 14] and possible treatment options [15–17]. Little is known about patients with unexplained complaints in their early stages. We performed this study to describe the demographic characteristics and quality of life of patients with early stage unexplained complaints. To provide some frame of reference, we compared these descriptive measures for the unexplained complaints to those of an unselected group of patients visiting their GP, and a group of depressed patients. For practitioners, better delineation, in terms of quality of life, of a group of patients that frequently seeks their help may sensitize them to consider more structured monitoring in order to achieve a firmer diagnosis be it somatic, psychological, or psychosocial. Methods GPs and patients GPs were recruited from the southern and the western part of the Netherlands. GPs were given the Dutch College of General Practitioners’ definition of unexplained complaints: those complaints that remain of unclear origin for the GP after adequate history taking, physical examination and careful consideration of the patient’s psychosocial context [1]. This definition supposes that complaints are labelled “unexplained” early on in the clinical episode before elaborate investigations were performed. Between February 2002 and December 2003, participating GPs included adult patients, presenting with a complaint that the GP designated as “unexplained” at the end of the first consultation. Patients who had presented with the same complaint in the previous 6 months were not included. Patients were eligible if they presented with fatigue, abdominal complaints, musculoskeletal complaints, weight changes or itch as their main complaint. Individual GPs decided whether a complaint was “unexplained”. There was no standardization of this process other than that implied by the Dutch College of General Practitioners’ definition. GPs invited patients to participate in the study at the end of the consultation. Informed consent procedure and medical ethics committees All participating patients gave written informed consent after having read information provided by the GP directly after the consultation. The medical ethics committees of the University of Amsterdam and Maastricht University approved the study protocol. Data collection: study population The data for this study were collected as part of baseline measurements for a randomized diagnostic trial on unexplained complaints in general practice [18]. After the entry consultation, patients filled out a questionnaire on demographic characteristics. The RAND 36-item Health Survey (RAND-36) was used to measure generic health-related quality of life. The RAND-36 [19] is a Dutch version of the Medical Outcome Study 36-item Short Form Health Survey (SF-36) [20] and is composed of 36 questions and standardized response choices, organized into eight multi-item scales (domains). These eight domains are: physical functioning (PF), role limitations due to physical health problems (RP), social functioning (SF), general mental health (MH), role limitations due to emotional problems (RE), vitality (VT), bodily pain (BP) and general health perception (GH). Raw RAND-36 scores on the eight domains are linearly converted to 0–100 scales with higher scores indicating better quality of life. Data collection: reference populations Reference population for comparison of demographic characteristics The demographic characteristics of a Dutch general practice population were taken from the second Dutch National Survey of General Practice, which is a large representative population-based survey that takes place every 10 years. Methods of this study have been published elsewhere [21]. Briefly, this national survey contains data about health and healthcare-related behaviour of 375,899 persons, registered in 104 practices with 195 GPs in 2001. Reference populations for comparison of quality of life The results section of this manuscript shows that the quality of life of patients with early stage unexplained complaints is poor. Our initial objective was purely descriptive. However, when such low quality of life figures were found, we felt the need to compare these to other patient groups. First, we wondered whether maybe all patients consulting their GP would have such low quality of life figures and therefore compared the unexplained complaints group with a Dutch general practice population. We also thought that maybe our patient population included a lot of depressed patients which could explain the poor quality of life and therefore also compared our quality of life figures with those of a population of depressed patients. Figures of the Dutch general practice population came from a study on functional status, health problems, age and comorbidity in primary care patients [22]. In this study 60 GPs from 43 general practices handed out a written questionnaire to 100 patients of 18 years and older, consecutively visiting their practice. In total 4,024 patients responded. Figures of the depressed patients came from the Netherlands Mental Health Survey and Incidence Study (NEMESIS). This NEMESIS survey, was based on a random sample drawn from the Dutch general adult population aged 18–64 (n = 7,076) of whom 204 had been diagnosed with a major depression within the past month [23]. Statistical analysis First, we compared the demographic characteristics of the five unexplained complaint categories to each other. Those of the total UC study population were compared to those of a Dutch general practice population. Differences were tested using the Fisher’s exact test. Second, mean domain scores on the RAND-36 were compared for each of the five unexplained complaint groups using linear regression with the “fatigue” complaint group as the reference category. We calculated robust variance estimates (Huber-White sandwich estimator) [24] to allow for potential dependence of quality of life scores within a single GP. Lastly, we compared mean domain scores of the total UC study population to those of a Dutch general practice population sample and to a random sample of depressed patients. Differences were statistically tested with t-tests in which equal variances were not assumed. Analyses were performed using STATA, version 9.2. Results General characteristics of the study group Of the 91 GPs who intended to participate, 18 dropped out before including any patients. They predominantly reported lack of time as the reason for drop out. Ten GPs did not include any patients during the inclusion period, although they did not formally drop out. Thus, 63 GPs (69%) included 513 patients with unexplained complaints (range: 1–36 patients per GP). Questionnaires from 466 (91%) patients were available for analysis. Forty-seven patients stopped their participation to the study or did not fill out this particular questionnaire. Demographic characteristics of the study group The demographic characteristics are summarized in Table 1, for the total UC study population, per complaint group and for the Dutch general practice population. The mean age of the total UC study population was 44 years (interquartile range 31–55) and 74% were women. Most patients had completed secondary or higher level education (91%) and were married or living together with a partner (71.5%). Table 1Demographic characteristics of the total UC study population, per complaint group and of a Dutch general practice populationDemographic characteristicCategoriesTotal UC study population % (n = 466)Fatigue % (n = 295)Abdominal complaints % (n = 60)Musculoskeletal complaints % (n = 69)Other complaint groups % (n = 42)Dutch GP population % (n = 3,85,461)SexMale26.426.12524.633.349.5Female73.6a73.97575.466.750.5AgeMean (years)44.241.842.352.2 b50.3Unknown0–194.96.16.71.1023.420–3937.142451330.930.440–6445.9a41.74066.7503365+1210.28.318.819.113.2Educational levelNone1004.3 b2.416Elementary97.53.317.4 b14.318.9Secondary68.9a68.18059.5b64.348.1Higher22.1a24.416.718.81917Marital stateSingle28.529.82030.428.6UnknownMarried/cohabiting71.570.28069.571.4UnknownInsurance TypePublic65.265.868.369.65067.4Private34.834.231.730.45032.6NationalityDutch95.396.396.791.39386.2Not Dutch4.73.73.38.7713.8Region of the NetherlandsWest56.457.646.762.352.4UnknownSouth43.642.453.337.747.6Unknowna P < 0.05 (Fisher’s exact test, reference group is GP population)b P < 0.05 (Fisher’s exact test, reference groups are fatigue and abdominal complaints) The main reason for encounter was fatigue (63.3%) and only few patients with weight changes or itch were included. In all further analyses, these latter two groups were combined into an “other complaints” group. Patients with musculoskeletal complaints were significantly older and less educated than patients with fatigue or abdominal complaints (P < 0.05). Compared to patients from the Dutch general practice population (second Dutch National Survey of General Practice), unexplained complaint patients were more likely to be women (74% vs. 51%), older (46% vs. 33% in the age group 40–64) and more highly educated (91% vs. 65% secondary or higher level education). These differences were statistically significant (P < 0.05). Quality of life of the total UC study population and per unexplained complaint group The overall quality of life for the total UC study population was poor (Table 2). All domains showed a mean score of less than 70 (on a scale from 0 to 100), except Physical Functioning (mean 73.4). The lowest score was in the Role Functioning Physical domain (mean 37.2). Also the Vitality domain scored low (mean 40.5) both in the total UC study population and in the separate complaint groups. Table 2Quality of life of the total UC study population and per complaint groupAllFatigueAbdominalMusculoskeletalOtherN466295606942Domains RAND-36MeanMeanMeanΔ (95% CI)MeanΔ (95% CI)MeanΔ (95% CI)Physical functioning73.473.381.38.1 (2.2 to 14.0)a62.2−8.0 (−14.8 to −1.3)a76.33.0 (−4.3 to 10.4)Social functioning66.562.570.88.3 (−0.2 to 16.9)76.413.9 (6.9 to 21.0)a729.5 (−0.7to 19.7)Role functioning physical37.231.556.324.7 (11.9 to 37.5)a397.5 (−3.5 to 18.5)4715.4 (−6 to 31.4)Role functioning emotional60.254.174.420.3 (9.0 to 31.6)a74.220.1 (6.6 to 33.6)a59.35.2 (−10.4 to 20.8)Mental health63.861.869.47.6 (2.3 to 12.9)a67.96.1 (1.1 to 11.1)a62.91.0 (−6.8 to 8.8)Vitality40.533.653.720.1 (15.1 to 25.1)a54.120.5 (13.8 to 27.2)a48.514.9 (8.2 to 21.5)aBodily pain67.772.761.7−10.0 (−15.9 to −6)a55.5−17.2 (−25.2 to −9.2)a61.3−11.4 (−21.5 to −1.2)aGeneral health55.755.261.15.9 (3.4 to 8.4)a53.8−1.4 (−3.6 to 0.8)54.3−0.9 (−5 to 3.3)Differences (Δ) indicate differences with the fatigue subgroup, where the mean of the fatigue subgroup was subtracted from the other subgroup’s meanConfidence intervals (95% CI) were calculated using the Huber-White sandwich variance estimator which accounts for within physician correlation (for details, see main text)aP < 0.05 (linear regression, reference group is fatigue) Except for the Physical Functioning and Bodily Pain scales, patients with unexplained fatigue significantly scored the worst (P < 0.05) in every domain, especially on the Role Functioning Physical domain (mean 31.5). Patients with musculoskeletal complaints showed the significantly lowest score on Physical Functioning (mean 62.2) and Bodily Pain (mean 55.5) (P < 0.05). Comparison of quality of life with other patient groups Table 3 shows that the total UC study population scored lower on all domains than the Dutch general practice population. These differences were statistically significant (P < 0.05) for all domains except Bodily Pain. Table 3Comparison of RAND-36 scores of different patient groupsDomains RAND-36Total UC study populationDutch GP populationDepressionAge18–8718–80+18–64N4664,024204MeanMeanΔ (95% CI)MeanΔ (95% CI)Physical functioning73.478.55.1 (2.9 to 7.3)a81.27.8 (3.9 to 11.7)aSocial functioning66.574.58 (5.4 to 10,6)a62.5−4 (−8.4 to 0.4)aRole functioning physical37.262.425.2 (21.5 to 28.9)a63.426.2 (19.3 to 33.1)aRole functioning emotional60.27514.8 (10.6 to 19.0)a42.6−17.6 (−24.6 to −10.6)aMental health63.869.65.8 (4.0 to 7.6)a46.6−17.2 (−20.4 to −14.0)aVitality40.558.518 (16.1 to 19.9)a40.3−0.2 (−3.6 to 3.2)Bodily pain67.768.40.7 (−1.7 to 3.1)680.3 (−4.3 to 4.9)General health55.765.710 (8.9 to 11.1)a55.70 (−3.2 to 3.2)Differences (Δ) indicate differences with the total UC study population, where the mean of the total UC study population was subtracted from the other patient group’s mean95% CI: 95% confidence intervalaP < 0.05 (t-test with equal variances not assumed, reference group is total UC study population) Patients from the depression group scored significantly lower on the Role Functioning Emotional and Mental Health domains. On all the other domains the total UC study population scored equally high or significantly lower (Physical Functioning and Role Functioning Physical, P < 0.05) than patients from the depression group. In other words, unexplained complaint patients scored lowest on predominantly physically oriented domains, whereas depressed patients scored lowest on predominantly mentally/emotionally oriented domains. Discussion Our findings indicate that patients with early stage unexplained complaints are mainly women in their forties, with secondary or higher education levels and with an overall remarkably poor quality of life. Their quality of life in all but one domain of the RAND-36 is significantly worse than that of patients from a general practice population, even taking into account that such a population also includes (around 13% of) unexplained complaint patients. Patients with unexplained complaints predominantly score badly on physically oriented domains, compared to depressed patients, who predominantly score badly on mental/emotional oriented domains. For the remaining domains they score on a comparably low level. Therefore, practitioners may consider to pay attention to quality of life aspects of patients with early stage unexplained complaints, even though this may not always lead to a firmer diagnosis or instant improvement in treatment options for most patients. A more intense monitoring of these patients could, however, be advised. The total UC study population scored very low on the Vitality domain of the RAND-36 (mean 40.5). This domain consists of questions regarding spirit and energy. The Vitality domain of the SF/RAND-36 is known to be correlated with subjective feelings of fatigue [25] and we therefore interpret this finding as a logical consequence of the high prevalence of patients with fatigue in the study population. Patients with unexplained complaints do not seem to have predominantly psychosocial problems. The domains Social Functioning and Mental Health were not the worst scoring domains in the study population (mean 66.5 and 63.8 respectively). The Role Functioning Physical domain scored much lower (mean 37.2), suggesting that patients with unexplained complaints feel that their physical complaints hinder them in their daily functioning. Of the five categories of unexplained complaints, patients with unexplained fatigue have the poorest quality of life. A first potential limitation of our study can be that different definitions and conceptualizations of unexplained complaints limit the generalizability of our findings. In our study, unexplained complaints were labelled “unexplained” by the GPs in the first consultation after onset, before more elaborate diagnostics. This is in contrast to medically unexplained symptoms (MUS), which refer to symptoms that have been ruled out to have an explainable cause (diagnosis by exclusion). Our concept of early stage unexplained complaints does not allow one to draw firm inferences as to whether the lower quality of life levels can be attributed to the unexplained complaints or should be attributed to underlying, though not yet diagnosed conditions, or concurrent comorbidity. Second, we cannot fully exclude the possibility that some degree of selection bias is present. Participating GPs may have selectively included older patients or those with poorer quality of life. However, in a non-inclusion study in the participating general practices, we searched the electronic medical files by means of text words for eligible but not included patients with unexplained complaints. This non-inclusion study did not show major sex and age differences between included and not-included patients. However, differences may exist on other characteristics. A third limitation of our study is that no specific depression or anxiety questionnaire was used. At the start of the study, such a poor quality of life was not anticipated and, therefore, only a more general questionnaire was considered sufficient. The RAND-36 mentally/emotionally oriented domain figures and the differences we found when comparing the quality of life profile of unexplained complaint patients with depressed patients however, are not pointing towards the presence of depression or anxiety. Furthermore, the GPs did not consider the included patients to be depressed or suffering from an anxiety disorder, otherwise they would not have labelled the patient as having unexplained complaints by definition. It is not impossible however, that depression or anxiety disorders might play a role in the poor quality of life of patients with unexplained complaints. Maybe these diagnoses are established only by GPs over time, and do not become clear in (one of) the first consultations. Finally, the patient groups used for comparison of quality of life are perhaps not totally comparable to our study population. For example, there are older patients included in our study population than in the depressed patient groups. Since age has its influence on quality of life this can have influenced the contrast. Also, a depression is a treatable condition, whereas early stage unexplained complaints are not treated yet. The better quality of life in this patient group on some domains can therefore be a treatment effect. We did, however, not intend to study a fully comparable contrast in this sense beforehand, it was a result driven comparison. Although much research has been performed in patients with more chronic consequences of unexplained complaints, to our knowledge, no other study on demographic characteristics and quality of life of patients with early stage unexplained complaints in general practice has been published. Patients with unexplained complaints appear to be mainly highly educated women in their forties. They report remarkably poor levels of quality of life. Future research should explore whether and how quality of life scores and other characteristics could help in early identification of patients at risk of developing chronicity. Until then practitioners should at least be aware that early stage unexplained complaints may not always be as mild as is frequently assumed. Early stage unexplained complaints may be associated with considerable suffering on a daily basis. Awareness of potential poor quality of life may help physicians realise that they are dealing with a patient, at least, in need of more intense monitoring but maybe also of more intense treatment approaches.

Sleep_Breath-4-1-2276561

Low intense physical exercise in normobaric hypoxia leads to more weight loss in obese people than low intense physical exercise in normobaric sham hypoxia

Training in mild to moderate hypoxia (14–17% O2 in breathing air) and extended resting in moderate hypoxia (9–13% O2) have been shown to have effects in animals and humans on lipid and glucose metabolism, appetite loss, and, in part, on body weight. The causality for these effects is not yet known in detail, and the available data in humans from high-altitude and low-pressure chamber studies are scarce. New technical developments by German companies in the production of artificial climates with normobaric hypoxic conditions in larger rooms at reasonable energy costs allow now to perform hypoxia weight loss studies in obese humans with stable experimental conditions and protocols with a sham hypoxia control. Thirty-two obese people were recruited for a mild intense training study in normobaric hypoxia (15 vol.% O2) and normoxia/sham hypoxia (20.1 vol.% O2). Twenty of these [mean age 47.6 years, mean body mass index (BMI) 33.1, 16 m, 4 f) were willing to follow up on an 8-week, three times per week, 90-min low intense physical exercise in their individual fat burning mode, which has been determined by an exercise testing with spiro-ergometry upfront. The subjects were evenly randomized into a hypoxia and sham hypoxia group. The difference of the two groups in weight loss and changes in HBa1C values were analyzed before and after the training period. No nutritional diet was applied. Subjects in the hypoxia group in mean lost significantly more weight than in the sham hypoxia group (Δ1.14 kg vs Δ0.03 kg; p = 0.026). This resulted in a tendency to reduce the BMI more in the hypoxia group (p = 0.326). In the mean, there was no HbA1C exceeding normal values (mean 5.67 and 5.47%), and the HbA1C stayed basically unchanged after the 8-week training. Mild physical exercise three times per week for 90 min in normobaric hypoxia for 8 weeks led to significantly greater weight loss in obese persons than the exercise in sham hypoxia in this, to our knowledge, first sham hypoxia controlled study. Introduction High altitude exposure which cause hypoxaemia has various serious effects on the human body. Those effects are studied by physiologists since the nineteenth century, and science has shed light on a lot of physiological and pathophysiological body reactions [1]. Weight loss in high altitude under hypoxic climate conditions is one of those effects in which the causes remain still unclear, and it is not known if it is dose/altitude-dependent and affects everybody or only certain individuals [3–5]. The detection of the hypoxia-induced factor-alpha 1 (HIF-α1) and the chain reaction on mediators that this factor starts have prompted some speculations and investigations on various nutritional signals. One of the nutritional signals that seem to be triggered by HIF is the peptide hormone leptin [6–9]. This could lead to the direction that the human lipid metabolism is affected by hypoxia and that hypoxia causes signals to reduce food intake and also changes in fat mobilization and fat uptake to the body [10]. Because the societies in first and second world countries struggle with an increased body weight, pathological fat levels, and the increasing number of individuals with metabolic syndrome in their populations, the urge for various inexpensive treatment forms including a lifestyle change has risen. The idea is coming up whether training or stay at moderate altitude or simulated altitude could enhance weight loss and improvement of fat and sugar metabolism because of the known effects of hypoxia on weight loss. Several open trial field studies with low numbers of individuals have been carried out so far, but they do not have controls [11–13]. In the AMAS 2000 study, patients with metabolic syndrome have trained and stayed for 3 weeks at 1,700 m altitude or within a control group at sea level. In this randomized controlled study, significant improvements in fat metabolism and regarding weight loss have been found for the patients at altitude compared to patients at sea level [14]. The construction design of our facility with normobaric hypoxic climate rooms allows pretending hypoxic conditions without actually running the rooms in hypoxic climate (sham hypoxia) and to perform a single blind randomized controlled weight loss study in simulated altitude. Our aim was to perform a physical exercise weight loss study in obese subjects under moderate hypoxia (15 vol.% O2) and to compare the effects on body weight and fat metabolism to a control group training on the same level in sham hypoxic conditions. Materials and methods Site Our facility for simulated altitude in the renovated Kurmittelhaus der Moderne in Bad Reichenhall in southeast Bavaria has three rooms (90, 37, and 25 m2) in which a hypoxic climate at normobaric conditions can be produced up to an simulated altitude of 6,000 m (approximately 7 vol.% O2; see Fig. 1). Fig. 1The 90 m2 hypoxia training room with exercise devicesThe rooms are two well-equipped training facilities and one sleeping and rest room with a full bath. To reach a hypoxic climate, the air in the room is flushed with nitrogen, and the volume percent of oxygen is reduced to the determined level. To control the climate, oxygen and carbon dioxide are constantly controlled. The data are transmitted to a computer, and the computer controls via specialized software the electronic climate unit and the inflow of nitrogen and fresh air through the air condition (software by Low Oxygen Systems, Berlin-Buch, Germany; hardware by Atlas Copco, Walter Bau and Siemens and Fujitsu-Siemens, all Germany). The nitrogen is produced via a filtration system that eliminates oxygen from pressurized air. Because of the constant air flow, there is a low noise level from the air condition in the room. The level of simulated altitude is usually displaced in the rooms. To keep subjects uninformed about the oxygen level in the room while the air condition was running at all times, this display was covered during the study times and made invisible to the subjects. Subjects Thirty-two healthy subjects (22 men, 10 women) with a minimum age of 16 years and a minimum body mass index (BMI) of 27 were asked to participate in an 8-week trial with three times 90-min moderate physical exercise. Estimation of sample size was calculated to reach a statistical significance for a change in BMI of one point. Acquisition of subjects was done by local radio announcement and personal contact. Subjects did not receive any kind of reimbursement for participation. All subjects were asked to continue with their normal diet/nutrition which they had before the study.Twenty subjects were finally willing to perform the whole 8-week training period and gave written consent. Subjects were randomly allocated to the hypoxia training and the sham hypoxia training group. Subjects were not told to which group they were allocated. Antropometric data of the subjects in the two groups are shown in Figs. 2 and 3. Both groups were comparable in terms of age and sex. Fig. 2Mean age (SD), weight, and body mass index (BMI) of the subjects who finished the trial in the hypoxia groupFig. 3Mean Age, weight, and BMI of the subjects who finished the trial in the sham hypoxia groupIn all 20 subjects, spiro-ergometry was performed to determine the max O2 consumption (ml kg−1 min−1) at exercise and the training level for the study set at the heart frequency value at 60% of the max O2 (ml kg−1 min−1). In all subjects, an echocardiography was performed for safety reasons to exclude subjects with heart insufficiency levels NYHA 2–4. Study parameters Primary study parameter was bodyweight (kg) before and after the trial. Secondary study parameters were: BMI and the laboratory parameters (serum and EDTA blood sample); HbAC1, cholesterol, high- and low-density lipoproteins and triglycerides. Other exercise and blood parameters were not part of this trial but are available on request for reviewers from the laboratory and exercise test data sheets.Blood samples were drawn at the beginning and end of the 8-week trial; bodyweight and height were measured via a regular medical weight scale. Training period Subjects performed exercise at a heart rate level which corresponded to 60% of the heart rate level at their individual max O2 (ml kg−1 min−1) on three different training (stepper, treadmill, and bicycle ergometer) devices for 90 min on 3 days per week, 8 weeks, either in normobaric hypoxia (15 vol.% O2, equivalent to 2,500-m altitude) or in normobaric normoxia (20 vol.% O2, equivalent to 450-m altitude). For safety reasons, SaO2 was controlled regularly during the training via pulseoximeter (Konica Minolta, Japan) in all subjects, and the subjects were asked for the main symptoms of acute altitude sickness (headache, nausea, and training untypical weakness in the legs). Pulse oximetry was blinded to subjects. Training was under constant control of medical personnel. Statistical analysis The differences between the two groups were analyzed via Student’s t test, statistical significance determined as p < 0.05 (statistical software used: Excel, Microsoft, Redwood, WA, USA). Data are presented as descriptive data, and values are given as mean values. Ethics and informed consent The study protocol was approved by the ethical committee of the Paris Lodron University Salzburg. All subjects gave written informed consent to participate in the study. Subjects under 18 were finally excluded; therefore, written consent by the parents was not necessary. Results All 20 subjects who were included in the training sessions finished the study without problems. No adverse side effects were reported by the subjects, including no signs of acute mountain sickness in those who exercised in 15.2 vol.% oxygen level. The average heart rate at 60% O2max for all subjects was 118/min. Subjects kept this heart frequency level stable during exercise (±3/min). This meant that because of the slightly increased heart frequency at rest in hypoxia, the hypoxia group exercised at a lower wattage. Because of the neutral smell of the air-conditioned air, subjects could not tell if they are in hypoxia or sham hypoxia. More than 50% of the subjects guessed wrongly as to which group they belong. Weight loss: The hypoxia group lost 1.14 kg in mean; the sham hypoxia group lost no bodyweight at all (p = 0.02, see Figs. 4 and 5). This led to a non-statistical significant tendency in the hypoxia group to reduce the BMI more than the sham hypoxia group (p = 0.33). The number of subjects had been too low to reach statistical significance. Fig. 4Mean (SD) weight loss after 7 weeks in the hypoxia group vs the sham hypoxia groupFig. 5Weight reduction or increase in the single subjects (matched pairs by age) of the hypoxia vs sham hypoxia group (y-axis = kg, x-axis = number of subject). Weight loss occurred in six subjects of the hypoxia group and in four of the sham hypoxia group, weight increase in two of the hypoxia group and four of the sham hypoxia group, weight stayed unchanged in two subjects of hypoxia (four and eight) and two of the sham hypoxia group (eight and nine) Lipid levels and HbA1C: None of the differences in lipid levels reached statistical significance. However, the hypoxia group showed little non-statistical significant tendencies to lose more cholesterol (−10.7 mg/ml vs 6.0 mg/ml; p = 0.68), triglycerides, and low-density lipoproteins than the sham hypoxia group, but less high-density lipoproteins (Figs. 6 and 7). Fig. 6Mean (SD) reduction and increase in triglycerides values of the hypoxia vs sham hypoxia groupFig. 7Mean (SD) reduction in high-density lipoproteins (HDL) in the hypoxia vs sham hypoxia group The healthy subjects of both groups were not diabetics. They had HbA1C levels on the edge but still in normal values (mean 5.67 and 5.47%), and those levels did not change after the training period of 8 weeks. Discussion The here presented study on fat metabolism and weight loss in mild to moderate hypoxia is, to our knowledge, the first one with a sham hypoxia control protocol. We could show that obese subjects did lose significantly more weight during an 8-week training period with three times per week 90-min low intense exercise when this exercise took place in 15 vol.% O2 hypoxia than in normal air at 450-m altitude (20 vol.% O2). We could also show that there have been non-statistical differences in fat metabolism or at least in serum levels of lipoproteins and triglycerides between the hypoxia training and the sham hypoxia training group. There are, due to the nature of the study and above that, several limitations, which have to be addressed. The number of subjects is not big enough to show a reduction in BMI, and power analysis shows that we would need about 32 subjects to have a significant BMI reduction. We wanted to get more subjects into the study, but the reduction of subjects to a limited number due to time restrictions and willingness is a common problem in weight loss protocols with obese people. The difference in weight loss in mean with 1.14 kg seems, for 8 weeks of training, not really a lot. However, several subjects did significantly lose more weight in the hypoxia group, and the mean bodyweight loss is not due to natural swings in weight change. This also happened although the hypoxia group trained at a slightly lower wattage due to a slightly increased heart frequency in hypoxia. We cannot fully exclude that the normoxia group decreased body fat and increased muscle mass due to the slightly higher level of muscle load, but it seems unlikely with this very low intense training. Because we did not put our subjects on a stringent diet, there might have been the chance that some individuals did eat more than normally and some less. Keeping the subjects on their routine food seemed to us the natural choice of being uninfluenced in the weight change by a special nutritional diet. The other point is that weight loss in hypoxia might be caused by reduced food intake, and that was a part of the study. To let the subjects reduce their food intake, you have to have them on a longer leash. This is especially true if leptin and hypothalamic signals come into play and might let subjects be less hungry, respectively, less hungry for fat. To achieve more accuracy, the only chance would be to exactly control the nutrition of all subjects already before the study and during the training period. To do this, subjects have to be under full supervision all the time. This is an almost impossible achievement in ambulatory care or with independent volunteers. Our findings are consistent with previous studies which have shown that low intense training in moderate hypoxia leads to weight loss [11–14]. The protocol with a sham hypoxia control group, which did not lose weight with the same amount of exercise, strengthens this hypothesis. A definite reason why this weight loss occurs is still not known. Several investigations could prove that the body reduces fat as a substrate for energy production due to the higher use of oxygen necessary in fat burning [2, 13], but this is still controversially discussed [14]. One of the reasons for a controversial discussion is that so far, nobody did observe subjects on a longer term [14]. Most studies have dealt with exercise in hypoxia for hours or a maximum of a few days, not for 2 months as we did. In our investigation, the subjects in hypoxia had reduced their triglycerides, their cholesterol, and their low-density lipoproteins and kept their high-density lipoproteins stable, whereas the subjects in normoxia did not do so. However, none of the effects reached statistical significance. Due to the actual literature, this can be explained by several possible causes: Either the fat mobilization is reduced due to hypoxia over time and so the serum levels of lipids are lower at the point of measurement or the increased catecholamine levels during hypoxia lead to reduced lipolysis and a depot of intracellular fat [16, 17]. Most likely, it seems that increased leptin levels lead to a reduced fat intake via food, and that is the reason why triglyceride levels decrease [6, 10]. The answer is still open, and more research is needed to clarify this. Other mediators besides catecholamines, like insulin like growth factor, might influence both fat and carbohydrate metabolism in hypoxia [15]. In conclusion, there is more and more evidence that even mild moderate hypoxia in comfortable simulated altitude, if combined with exercise, leads to weight loss. Weight loss regimes with exercise and diets or just exercise could be combined with normobaric hypoxia in the future. Future investigations should include whether the stay in moderate hypoxia alone without exercise increases the weight loss and if such a regimen would still be safe.

Biotechnol_Lett-3-1-2151841

Engineering eukaryotic protein factories

The biopharmaceuticals market is currently outperforming the pharmaceuticals market and is now valued at US$ 48 billion with an average annual growth of 19%. Behind this success is a 100-fold increase in productivities of eukaryotic expression systems. However, the productivity per cell has remained unchanged for more than 10 years. The engineering of the ER-resident protein folding machinery is discussed together with an overview of signal transduction pathways activated by heterologous protein overexpression to increase cell specific productivities. Introduction The demand for therapeutic proteins, i.e. monoclonal antibodies, has been forecast to exceed production capacities (Werner 2004). Moderate increases in product titers and yields are expected to result in significant savings in capital investment and consumer costs (Werner 2004). Over the last decade product titers have increased from ∼20 mg l−1 to 2 g l−1 (Werner 2004; Wurm 2004). These improvements are largely attributable to increases in viable cell densities. However, cell specific productivities remained nearly constant during this time (Wurm 2004). To improve cell specific productivities the bottleneck for heterologous protein secretion has to be identified and resolved. In expression systems exploiting multiple gene copies or strong promoters the amount of secreted heterologous protein does not increase proportionally with gene copy number (Schröder 2007), messenger ribonucleic acid (mRNA) (Schröder et al. 1999), or even the intracellular amount of the heterologous protein (Schröder and Friedl 1997). In these cells, the protein accumulates in intracellular aggregates (Schröder and Kaufman 2005), associates with the molecular chaperone heavy chain-binding protein (BiP)/glucose-regulated protein of 78 kDa (GRP78)/karyogamy 2 protein (Kar2p) (Schröder and Kaufman 2005), and induces dilation of the endoplasmic reticulum (ER) (Dorner et al. 1989; Gennaro et al. 1991). Therefore, exit of the correctly folded polypeptide chain from the ER is the rate-limiting step for heterologous protein secretion. I will discuss strategies to improve cell specific productivities by engineering protein folding in the ER and signal transduction pathways activated by overexpression of heterologous proteins. Protein folding in the ER The ER-resident protein folding machinery has four components: (1) Molecular chaperones assisting other proteins to fold, (2) enzymes such as protein disulfide isomerases (PDIs) and cis-trans peptidyl prolyl isomerases (PPIs), (3) a degradation machinery, and (4) signal transduction pathways that co-ordinate protein folding demand with capacity. At least three hierarchically organized chaperone systems operate in the ER, the heat shock protein of 70 kDa (HSP70) chaperones BiP and luminal HSP seventy 1 protein (Lhs1p)/chaperone in the ER 1 protein/Ssi1p/GRP170/oxygen-regulated protein of 150 kDa (ORP150), the HSP90 chaperone GRP94/adenotin/endoplasmin/tumor rejection antigen glycoprotein of 96 kDa (gp96)/ER protein of 99 kDa/HSP108/Ca2+-binding protein 4/protein kinase of 80 kDa, and the lectin chaperones calnexin/p88, calreticulin/Ca2+-binding protein of 63 kDa (CAB-63)/calregulin/Ca2+-regulated protein of 55 kDa (CRP55)/high affinity Ca2+-binding protein (HACBP), and calmegin/calnexin-t (Fig. 1). BiP works on completely unfolded polypeptide chains and translocates nascent polypeptide chains into the ER, whereas GRP94 and the lectin chaperones work on partially folded substrates. Fig. 1Hierarchy of chaperone systems in the ER. Abbreviations: Glc = d-glucose, Pi = inorganic phosphate () BiP consists of an N-terminal ATPase domain and a C-terminal substrate binding domain. BiP cycles through rounds of adenosine triphosphate (ATP) hydrolysis and adenosine diphosphate (ADP) ATP exchange (Fig. 1), which makes BiP a chaperone foldase (Winter and Jakob 2004). At least six DnaJ or HSP40 co-chaperones stimulate the ATPase activity of BiP and two growth after phage induction E (GrpE) co-chaperones the ADP ATP exchange reaction. In the ADP-bound form BiP has high affinity for unfolded substrates. Substrates bound to BiP are conformationally locked. ADP ATP exchange decreases the affinity of BiP for unfolded substrates. Substrate binding stimulates the ATPase activity of BiP. Thus, protein folding requires ATP (Dorner and Kaufman 1990). BiP and Lhs1p coordinate their activities. Lhs1p is a nucleotide exchange factor for BiP, and BiP stimulates the ATPase activity of Lhs1p (Steel et al. 2004). Cytosolic HSP90 chaperones are ATP-consuming chaperone foldases (Fig. 1) regulated by several co-chaperones that affect its ATPase activity, load substrates onto HSP90 chaperones, assist in protein folding, and target substrates to degradation pathways (Strudwick and Schröder 2007). GRP94 co-chaperones have not been identified, but other ER luminal chaperones found in complexes with GRP94 may be GRP94 co-chaperones. The KM value of GRP94 for ATP is ∼100 μM and its intrinsic ATPase activity is barely detectable. These data indicate that GRP94 may be a chaperone holdase, a chaperone that binds to unfolded proteins, but does not cycle through ATP hydrolysis cycles. A drop of adenine nucleoside concentrations in stress situations may activate GRP94 (Rosser et al. 2004). The lectin chaperones share an N-terminal globular oligosaccharide-binding domain and a C-terminal extended hairpin loop, the P domain, which provides the chaperone function (Strudwick and Schröder 2007). Calnexin and calmegin are transmembrane proteins. Calreticulin is a soluble protein. Most proteins entering the ER are glycosylated by addition of the oligosaccharide Glc3Man9GlcNAc2 (Glc = d-glucose, Man = d-mannose, GlcNAc = N-Acetyl-2-d-glucosamine). α-Glucosidases I and II rapidly remove the two terminal d-glucose residues. Calnexin and calreticulin bind to this monoglucosylated oligosaccharide. Removal of the third d-glucose moiety by α-glucosidase II releases the protein from calnexin/calreticulin. Reglucosylation of unfolded proteins by uridine diphosphate (UDP)-glucose:glycoprotein glucosyl transferase (UGGT) triggers repeated interactions with calnexin/calreticulin (Fig. 1). This reglucosylation reaction consumes UDP-glucose. Therefore, the calnexin chaperones are chaperone foldases. The Saccharomyces cerevisiae genome encodes a calnexin homolog, but no homolog for UGGT, suggesting that the lectin chaperones may also function as holdases. Demannosylation of N-linked oligosaccharides trigger export of proteins to the Golgi complex or their targeting for retrotranslocation into the cytosol and proteasomal degradation (Lederkremer and Glickman 2005; Ruddock and Molinari 2006). An expanding family of PDIs catalyzes the formation and isomerization of disulfide bonds. Disulfide bond formation requires the regeneration of oxidized PDI catalyzed by the flavin adenine dinucleotide (FAD)-dependent oxidases ER oxidation 1 protein (Ero1p)/ERO1-Lα, ERO1-Lβ and essential for respiration and viability 2 protein (Erv2p) (Fig. 2). The final electron acceptor is molecular oxygen. Oxidative protein folding accounts for ∼25% of all reactive oxygen species (ROS) formed in a cell (Tu and Weissman 2004) and is increased by ER stress (Haynes et al. 2004). Disulfide bond isomerization is independent of the Ero1p oxidases. Reduced glutathione provides reducing equivalents to remove incorrectly formed disulfide bonds. PDI has chaperone holdase activity that is independent of its protein disulfide isomerase activity (Wilkinson and Gilbert 2004) and redox-regulated foldase activity (Tsai et al. 2001). Fig. 2Disulfide bond formation and isomerization reactions catalyzed by PDI Slowly folding proteins or folded incompetent proteins have to be removed from chaperone cycles to prevent them from poisoning the protein folding machinery. This is achieved by targeting these proteins for retrotranslocation into the cytosol and degradation by the proteasome in a process called ER associated protein degradation (ERAD) (Yoshida 2007) (Fig. 3). Proteins destined for ERAD are recognized by BiP, PDI, and lectin sensors such as ER-degradation enhancing α-mannosidase-like proteins 1–3 (EDEM1–3)/homologous to mannosidase I 1 protein (Htm1p)/mannosidase-like 1 protein (Mnl1p) and osteosarcoma-9 (OS-9)/yeast OS-9 homolog protein (Yos9p) (Gauss et al. 2006a). The lectin sensor EDEM extracts unfolded proteins from the calnexin cycle. ERAD substrates are then retrotranslocated into the cytosol via the secretory 61 (SEC61) channel or a channel formed by derlin proteins, ubiquitinated by action of ubiquitin-conjugating enzymes (E2 enzymes) and ubiquitin ligases (E3 enzymes), deglycosylated, and degraded by the proteasome. Redundancy exists in the ubiquitination machinery involved in ERAD. In yeast and mammalian cells at least three distinct E2·E3 complexes are present in the ER membrane. The substrate specificities of these complexes, as well as their substrate selection rules are poorly understood. In yeast, the E3 ligase Hrd1p co-operates with Hrd3p and Yos9p to select its substrates (Gauss et al. 2006b). Another selection criterion is the time an unfolded protein spends in foldase cycles. In the calnexin cycle, demannosylation of N-linked oligosaccharides generates glycostructures recognized by lectins such as EDEM1–3. Association of cytosolic HSP70s with the nucleotide exchange factor B cell leukemia/lymphoma 2 (Bcl-2) associated athanogene protein 1 (BAG-1) (Alberti et al. 2003) and of cytosolic HSP90s with the ubiquitin ligase carboxyl terminus of heat shock cognate protein of 70 kDa (HSC70) interacting protein (CHIP) (Höhfeld et al. 2001) targets proteins for proteasomal degradation. Association of the DnaJ co-chaperone 58 kDa inhibitor of double-stranded RNA activated protein kinase (PKR, p58IPK) with HSP70 chaperones at the translocation pore targets proteins, whose translocation into the ER is stalled, for proteasomal degradation (Oyadomari et al. 2006). If similar targeting mechanisms for proteins folding slowly in the BiP foldase cycle exist, is currently unknown. Fig. 3ER associated protein degradation. The exact nature of the retrotranslocation channel is unknown. The list of E2 and E3 enzymes catalyzing ubiquitination of ERAD substrates is not exhaustive. Abbreviations: NEF = HSP70 co-chaperone, Ub = ubiquitin Engineering of chaperone machineries Engineering of chaperone holdases should be straightforward. Holdases hold onto an unfolded protein and act in an equimolar ratio to the unfolded substrate, but are unable to support folding reactions. Holdases may provide a buffer capacity to prevent aggregation of unfolded proteins. Co-overexpression of chaperone foldases may be beneficial to overexpression of holdases (Smith et al. 2004; Zhang et al. 2006b). Overexpression data for calnexin/calreticulin and PDI are most easily explained by assuming that this buffer capacity is increased. Calnexin or calreticulin overexpression generally improved heterologous protein secretion (Chung et al. 2004; Conesa et al. 2002; Kato et al. 2005). Overexpression of one cycle constituent should not increase cycle capacity, if this constituent is not limiting, and if constituent levels are not co-regulated. The hypothesis that calnexin or calreticulin are limiting for the calnexin cycle seems unlikely, because of the high abundance of these proteins. Alternatively, overexpression of calnexin or calreticulin may inactivate lectins such as EDEM (Molinari et al. 2003; Oda et al. 2003), which target unfolded proteins for ERAD. Further, elevated calnexin or calreticulin levels not supported by concomitant increases in activities of other calnexin cycle components, may convert these surplus amounts of calnexin and calnexin into holdases or inactive chaperones. PDI overexpression increased secretion of some, but not all, heterologous proteins (Butz et al. 2003; Damasceno et al. 2006; Inan et al. 2006; Kato et al. 2005; Mohan et al. 2007; Schröder 2007; Zhang et al. 2006b). PDI overexpression provides increased holdase activity in the ER. For example, overexpression of catalytically inactive PDI improved heterologous protein secretion (Hayano et al. 1995). PDI overexpression also improved secretion of heterologous proteins containing no disulfide bonds (Powers and Robinson 2007; Smith et al. 2004; Vad et al. 2005). Substrate selectivity of PDI may explain why PDI overexpression failed to increase heterologous protein secretion in some studies (Butz et al. 2003; Damasceno et al. 2006; Mohan et al. 2007; Schröder 2007). If this is the case, co-expression of an alternative holdase, i.e. GRP94, may yield more consistent results for different heterologous proteins. Alternatively, increased rates of disulfide bond formation or isomerization in PDI overexpressing cells may explain increased heterologous protein secretion. As an enzyme, PDI should function at substochiometric ratios to its substrates. In disulfide bond formation, regeneration of oxidized PDI may become rate-limiting in PDI overexpressing cells. PDI is an abundant ER protein, making it unlikely that its overexpression can significantly augment its catalytic activities. Engineering of foldases seems more difficult, because of the complex regulation of these chaperones (see above). BiP overexpression increased heterologous protein secretion for about half of the heterologous proteins studied (Chung et al. 2004; Damasceno et al. 2006; Dorner et al. 1992; Hsu and Betenbaugh 1997; Kim et al. 2003; Schröder 2007; Smith and Robinson 2002; Zhang et al. 2006b), but can also have negative effects (Dorner et al. 1992; Schröder 2007). In cells overexpressing BiP one of its co-chaperones, Lhs1p, or ER luminal ATP may become limiting, converting BiP into a holdase or inactive chaperone. This may target proteins towards ERAD, stall translocation, convert Lhs1p into a holdase or inhibit Lhs1p. Increased BiP activity may stall the GRP94 and/or calnexin–calreticulin chaperone machineries, because of the hierarchy of ER luminal chaperone systems. This again may target heterologous proteins for ERAD. BiP overexpression attenuates the unfolded protein response (UPR, see below). Thus, BiP levels in WT and BiP overexpressing cells expressing a heterologous protein may not be dramatically different. Further, BiP overexpressing cells may experience an imbalance in their chaperone machineries, because the UPR co-ordinates expression of several chaperones. Moreover, if overexpression of BiP shifts the folding bottleneck to the GRP94 and calnexin chaperone systems, upregulation of these chaperone systems by the UPR may be blunted, because of suppression of UPR activation by elevated BiP levels. Unfolded protein response (UPR) ER stress is the perturbation of the balance between the folding capacity and the folding demand imposed on the ER. To restore ER homeostasis, the UPR activates expression of chaperone, PDI, PPI, and ERAD genes, attenuates transcription of genes encoding secretory proteins and general translation, induces phospholipid synthesis, and induces an inflammatory response. Apoptosis is induced if these actions fail to restore ER homeostasis or in response to prolonged ER stress (Schröder and Kaufman 2005). The ER membrane of higher eukaryotes harbors at least three classes of transmembrane proteins activated by ER stress, basic leucine zipper (bZIP) transcription factors synthesized as type II transmembrane proteins, of which the best characterized are activating transcription factor 6α (ATF6α) and ATF6β/cyclic adenosine monophosphate response element binding protein (CREB)-related protein (CREB-RP)/G13, the protein kinase eukaryotic translation initiation factor 2α (eIF2α) kinase 3 (EIF2AK3)/pancreatic eIF2α kinase (PEK)/PKR-like ER kinase (PERK), and the protein kinase-endoribonucleases ER to nucleus signaling 1α (ERN1α)/inositol-requiring 1α (IRE1α) and IRE1β (Fig. 4). Several ATF6 paralogs, CREB3, CREB4, CREB-H, box B binding factor 2 (BBF2), and old astrocyte specifically induced substance (OASIS) have recently been described. In unstressed cells, these stress sensors are bound to BiP and kept in an inactive state. Upon ER stress, BiP is released coinciding with their activation. Whether BiP release is triggered by sequestration by unfolded proteins, conformational changes in the ER luminal domains, or reactivation of the BiP ATPase cycle is currently being debated (Kimata et al. 2004; Shen et al. 2005). ER stress also alters interactions of pro- and antiapoptotic Bcl-2 proteins at the ER membrane. These proteins regulate ER luminal Ca2+ homeostasis, ER morphology, ER stress signaling, and apoptosis induced in response to ER stress (Hetz 2007). Fig. 4Signal transduction pathway in the UPR. In the human UPR caspase 4 substitutes for caspase 12. In yeast and filamentous fungi, the IRE1-XBP-1 (Hac1p/HACA) pathway is the only known UPR signal transduction pathway. [Reprinted in modified form with permission from Bentham Science Publishers from Schröder M, Kaufman, RJ (2006) Divergent roles of IRE1α and PERK in the unfolded protein response. Curr Mol Med 6:5–36] BiP release from ATF6 induces translocation of ATF6 to the Golgi complex where its bZIP transcription factor domain is proteolytically released from the Golgi membrane by site 1 and 2 proteases (S1P and S2P) (Fig. 4). This cytosolic fragment translocates to the nucleus and induces expression of chaperone genes (Strudwick and Schröder 2007). The functions of the ATF6 paralogs in the ER stress response is less well understood. OASIS contributes to activation of the BiP gene (Kondo et al. 2005), and ATF6·CREB-H heterodimers activate inflammatory genes (Zhang et al. 2006a). BiP release from PERK triggers oligomerization and activation of the protein kinase domain of PERK. PERK phosphorylates the bZIP transcription factor nuclear factor erythroid 2 (NF-E2) related factor 2 (NRF2) and eIF2α (Strudwick and Schröder 2007) (Fig. 4). NRF2 phosphorylation induces its translocation to the nucleus, where a NRF2·ATF4 heterodimer activates transcription of antioxidant response genes (Harding et al. 2003) to counteract ROS formation by repeated folding attempts of nascent polypeptide chains in ER stressed cells. eIF2α phosphorylation attenuates general translation and decreases the influx of nascent unfolded polypeptide chains into the ER. It also clears short-lived proteins from the cell, i.e. D-type cyclins, resulting in cell cycle arrest in G1 phase, and of inhibitors of nuclear factor κB (NF-κB) (IκB), activating the pro-inflammatory transcription factor NF-κB. eIF2α phosphorylation also induces preferential translation of capped mRNAs containing several short upstream open reading frames (uORFs) and cap-independent translation of mRNA via internal ribosomal entry sites (IRES). Currently, the only known mRNA whose translation is stimulated by ER stress is the mRNA encoding the bZIP transcription factor ATF4. ATF4 induces expression of CCAAT enhancer binding protein (C/EBP) homologous protein (CHOP)/CHOP-10/growth arrest and DNA damage-inducible gene 153 (GADD153). CHOP represses transcription of anti-apoptotic BCL-2 (McCullough et al. 2001) and induces expression of pro-apoptotic tribbles-related protein 3 (TRB3) (Ohoka et al. 2005). Translational attenuation by PERK is transient and countered by several eIF2α phosphatases. The regulatory subunit of protein phosphatase 1 (PP1), GADD34, is induced late in ER stress by ATF4, and targets PP1 to eIF2α. Upon BiP release, IRE1 autophosphorylates in trans and activates its endoribonuclease domain (Strudwick and Schröder 2007). Activated IRE1 cleaves exon-intron junctions in the mRNAs encoding the bZIP transcription factors homologous to ATF/CREB1 protein (Hac1p) in yeast, HACA in filamentous fungi, and X box-binding protein 1 (XBP-1)/hepatocarcinogenesis-related transcription factor (HTF)/tax-responsive element binding protein 5 (TREB5) in metazoans (Fig. 4). In yeast, the HAC1 exons are ligated by transfer RNA ligase. Both transcription factors activate expression of genes encoding ER-resident molecular chaperones, protein foldases, and genes encoding proteins involved in ERAD. Hac1p and XBP-1 also stimulate phospholipid biosynthesis. Spliced Hac1p represses genes under control of the transcriptional regulator unscheduled meiosis 6 protein (Ume6p), including the early meiotic genes and a large array of genes involved in carbon and nitrogen metabolism. This function of spliced Hac1p may explain the slow growth phenotype of yeast cells overexpressing Ire1p or spliced Hac1p (Mori et al. 2000). Attenuation of metabolism by Hac1p in yeast may substitute for attenuation of general translation by PERK. XBP-1 activation is delayed compared to ATF6 and PERK activation (Yoshida et al. 2003), allowing for an early folding only phase in the UPR followed by a folding and degradation phase. In mammals, phosphorylated IRE1 sequesters tumor necrosis factor receptor associated factor 2 (TRAF2) from pro-caspase 12, resulting in clustering and activation of this caspase and initiation of apoptosis (Strudwick and Schröder 2007) (Fig. 4). TRAF2 recruits apoptosis signal-regulating kinase 1 (ASK1) and IκB kinase to IRE1. ASK1 activates the mitogen-activated protein (MAP) kinases p38 and ju-nana (jun) N-terminal kinases (JNKs). p38 and JNKs phosphorylate and potentiate pro-apoptotic transcription factors, i.e. CHOP and c-Jun, and phosphorylate and inhibit anti-apoptotic Bcl-2 and Bcl-xL to induce apoptosis (Szegezdi et al. 2006). Inhibition of JNK phosphatases by ROS formed in oxidative protein folding also contributes to JNK activation (Kamata et al. 2005). Association of the pro-apoptotic Bcl-2 family proteins Bax and Bak with the cytosolic portion of IRE1 is required for efferent IRE1 signaling, i.e. XBP-1 mRNA splicing and JNK1 activation (Hetz et al. 2006). Engineering of the UPR Engineering of the UPR holds promise to increase concentrations of several chaperones in a functionally meaningful ratio, thus preventing inadvertent conversion of chaperone foldases into holdases, or even inactivating chaperones because of lack of concomitant increases in co-chaperones and co-factors. The UPR is as much a survival response as it is a response to boost protein secretion. Clearly, some aspects of the physiological UPR are not desirable in an expression system, for example inhibition of general translation by PERK, stimulation of ERAD and apoptosis. Overexpression of spliced Hac1p or its filamentous fungi homolog HACA increased heterologous protein secretion in Saccharomyces cerevisiae and Aspergillus niger (Valkonen et al. 2003a, b). In Chinese hamster ovary (CHO)-K1 cells XBP-1 overexpression improved secretion of several heterologous proteins (Ku et al. 2007; Tigges and Fussenegger 2006). In yeast and filamentous fungi, secretion of several heterologous proteins was not improved or even decreased by overexpression of Hac1p (Gasser et al. 2006; Rakestraw and Wittrup 2005; Valkonen et al. 2003a). One drawback of some of these studies is that wild-type (WT) cells with a functional UPR are compared to WT cells overexpressing spliced Hac1p, which may explain why in some cases only marginal differences between the two cell types have been observed. The more consistent effect of spliced XBP-1 overexpression on protein secretion (Ku et al. 2007; Tigges and Fussenegger 2006) may be caused by a preferential activation of UPR signaling pathways that stimulate cellular functions beneficial for heterologous protein secretion, i.e. increased chaperone gene expression. This increased ER-resident chaperone machinery attenuates all UPR signaling, and most importantly, those UPR signaling pathways likely to decrease protein production, for example general translation inhibition by PERK, stimulation of ERAD and of apoptosis (Özcan et al. 2004). Conclusions Engineering of chaperone systems by overexpressing a single component of the ER-resident protein folding machinery has overall yielded mixed results. Our basic understanding of protein folding in the ER is still incomplete. Addressing these open questions should underpin engineering approaches to improve the performance of chaperone systems. A more detailed understanding of chaperone function and regulation should inform future work to improve chaperone systems. Co-expression of different holdases or targeting of heterologous or cytosolic holdases to the ER may yield more consistent improvements for different heterologous proteins. The function of overexpressed chaperones may not be the same as at their normal physiological concentrations, because of the lack of a corresponding increase in co-chaperones and co-factors. To improve foldase function, concomitant elevation of BiP, its co-chaperones, and ATP levels should be attempted. Engineering of the UPR suggests that mimicking an UPR by expression of its activated signaling molecules does not consistently improve productivities. Dissection of UPR signaling activities may be necessary to improve heterologous protein secretion. To place engineering of the UPR on an informed basis, we need to understand the UPR in more detail. It is still not clear what the most upstream events in activation of the UPR are, how UPR signaling integrates into cellular signaling, and how the UPR decides between a prosurvival and an apoptotic response to ER stress. Cell engineering also needs to address the potential problem that increased oxidative protein folding may be inherently toxic to cells, because of elevated cellular ROS levels.

Pharm_World_Sci-4-1-2253652

Minimising treatment-associated risks in systemic cancer therapy

Aim of the review To review the consequences of drug-related problems (DRP) in systemic cancer therapy and identify specific contributions of the pharmacist to minimise treatment-associated risks. Method Searches in PubMed, Embase and the Cochrane Library were conducted. Bibliographies of retrieved articles were examined for additional references. Only papers in English between 1980 and 2007 were included. Results In systemic cancer therapy there is an enormous potential for DRP due to the high toxicity and the complexity of most therapeutic regimens. The most frequently reported DRP can be classified into adverse effects, drug–drug interactions, medication errors, and non-adherence. Pharmacists have enhanced efforts to assure quality and safety in systemic cancer therapy together with other health care providers. In consequence, oncology pharmacy has evolved as a novel specialist discipline. The endeavour to merge and co-ordinate individual activities and services of the pharmacist has led to pharmaceutical care concepts which aim at offering novel solutions to the various DRP. Conclusion Pharmaceutical care for cancer patients should be developed within research projects and integrated into disease management programs in order to ensure broad implementation. Impact of findings on practice There is an enormous potential for drug-related problems in cancer therapy The most frequently reported drug-related problems in systemic cancer therapy are adverse effects, drug–drug interactions, medication errors, and non-adherence Pharmacists can contribute substantially to risk minimisation in systemic cancer therapy by adding specific drug-related knowledge to the treatment team and offering patient-related services Pharmaceutical care for cancer patients offers novel solutions to the various drug-related problems and should be further developed in research projects Introduction In systemic cancer therapy, drug regimens are administered following established protocols which have been carefully evaluated in clinical trials. The administration of supportive medication is not as standardised as with antineoplastic therapy. Major components of the supportive therapy are selected by the general practitioners or the patients themselves rather than by the oncologist. Furthermore, patients tend to see more than one physician involved in the cancer care process as well as alternative practitioners. Patients are also exposed to a tremendous variety of drugs which are available to the customer without prescription (over-the-counter, OTC). The more complex drug therapy is, the higher the risk of experiencing drug-related problems (DRP) such as adverse effects, interactions, medication errors, and non-adherence. Drug-related problems in cancer chemotherapy can have severe consequences originating from the high toxicity and narrow therapeutic range of anticancer drugs. The tragic case of Betsy Lehman who died from an overdose of cyclophosphamide at the Dana Faber Cancer Center in Boston demonstrated this risk painfully [1]. It must be the goal of all health care providers to minimise treatment-associated risks as much as possible. Over the last few decades the pharmacy profession has experienced a change from traditional drug-oriented toward patient-oriented services. In oncology, pharmacists have established central services for compounding cytotoxic drugs and started to offer industry-independent drug information for physicians and patients as well as therapeutic drug monitoring for critical dose drugs, e.g. aminoglycosides in neutropenic patients. As the specialist knowledge of pharmacists in this field has increased continuously “oncology pharmacy” has evolved into a new pharmaceutical discipline with its own curriculum. In 1995 the International Society for Oncology Pharmacy Practitioners (ISOPP) was founded. The aim of the society is “to determine the optimal medical treatment for cancer patients, thereby improving their quality of life”. Currently, the concept of pharmaceutical care is being adapted to the needs of the cancer patient as a further step to optimising individual drug therapy [2]. The intention of this review is to summarise possible solutions for DRP in oncology focussing on the specific tasks of the pharmacist. Aim of the review It was the aim of this article to review and analyse the most frequent DRP and their consequences in systemic cancer therapy. Moreover, specific contributions of the pharmacist to minimise treatment-associated risks should be identified. Method Searches in PubMed, Embase and the Cochrane Library were conducted. Database terms included DRP, adverse effects, drug–drug interactions, medication errors, adherence, compliance, pharmaceutical care in connection with oncology or cancer. Bibliographies of retrieved articles were examined for additional references. Only papers in English between 1980 and 2007 were included. Results Various DRP can occur in the treatment path of systemic cancer therapy (Table 1). In a recently published study conducted on an oncology ward of a Swedish hospital 114 DRP in 58 patients were identified indicating their high incidence in cancer patients [3]. In this section, major problems related to adverse effects, drug–drug interactions, medication errors, and non-adherence are reviewed separately. Table 1Drug-related problems can originate from several steps of the treatment path Adverse effects According to the World Health Organization (WHO) an adverse drug reaction (ADR) is defined as a response to a drug which is noxious and unintended, and which occurs at doses normally used or tested in man for prophylaxis, diagnosis or therapy of disease. In cancer chemotherapy such ADRs are strongly connected to the treatment itself. Because of the fact that most cytotoxic agents cannot distinguish between normal and neoplastic cells, most ADRs seem to be unavoidable. They are often accepted not only by patients but also by health care providers. In order to illustrate the most common ADRs in oncology patients in a cancer centre in Australia were observed and the incidence as well as the predictability, preventability, and severity of the occurred ADRs were assessed. Among the ten most common ADRs constipation ranked first but was connected directly to the use of opioids rather than to cytotoxic chemotherapy. Nausea and vomiting, fatigue, alopecia, drowsiness and myelosuppression ranked second to sixth. Of these ADRs 88% were predictable and about 50% even probably preventable, because of inadequate use of preventative measures [4]. As well as the incidence of ADRs in modern chemotherapy the patients’ perceptions of the impact of these ADRs on well being and quality of life are increasingly taken into account. Patient perceptions have changed markedly over the last two decades. Whereas in 1983 physical symptoms such as nausea, vomiting and hair loss were most troubling from the patients’ point of view, in 2002 psychosocial complaints ranked among the top ten symptoms, with the complaint “affects my family or partner” rated as the most severe ADR. Alopecia and fatigue ranked second and third place [5, 6]. As a consequence, the management of ADRs has become a focus of clinical research and new drugs such as aprepitant, a neurokinin-1-receptor antagonist for the prevention of nausea and vomiting, were developed [7]. Despite these new therapeutic options the minimisation of toxic effects of chemotherapy still seems to be a challenging task. The first and most important step to a better management of ADRs is the formulation and implementation of evidence-based clinical practice guidelines for the different symptoms in a multidisciplinary approach. Several studies have shown a positive effect of guidelines for antiemetic prophylaxis and therapy on both clinical and economic outcomes [8–11]. The most important influencing factors seem to be the appropriate use of 5-HT3 receptor antagonists and the application of oral drug formulations. Nevertheless, altering the usage patterns of physicians to comply with evidence-based guidelines seems to be difficult. Methods such as the use of computerised decision support systems and educational outreach mechanisms appear to be most effective particularly when used in combination [12]. The pharmacist can support and promote the adherence to guidelines [13]. A second step to a better management of ADRs is the implementation of standardised chemotherapy order forms that also contain supportive care medication. By using standardised order forms a more appropriate prescribing of antiemetics based on the level of emetogenicity of administered chemotherapy and thus a reduction in drug expenditure can be achieved [14]. Such order forms should be elaborated following a multidisciplinary approach including physicians, pharmacists and nurses. Drug–drug interactions The large number of prescribed drugs administered to cancer patients leads to a high potential for drug–drug interactions. As indicated previously many cancer patients also use over-the-counter medication as well as alternative and complementary treatment options. Frequently, cancer patients suffer from concomitant chronic diseases that require the intake of other drugs, further increasing the risk of interactions. This problem is often underestimated in oncology as highlighted in several review articles [15–18]. Moreover, drug–food, drug–disease and drug–diagnostics interactions have to be considered that are, however, beyond the scope of this article. Drug–drug interactions must be avoided as they can lead to overdosing or underdosing of anticancer drugs with the consequence of toxicity or a loss of effectiveness, respectively. However, not all published interactions are also clinically relevant. A rational and safe drug therapy includes a check for potential drug–drug interactions based on the individual medication and a decision whether and how identified drug–drug interactions have to be considered or not. Knowledge of the mechanisms of interactions is crucial to assess the clinical relevance of an interaction. Drug–drug interactions are usually classified as pharmaceutical, pharmacokinetic or pharmacodynamic interactions (Table 2). The risk of interactions can differ substantially between individual drugs, even within the same class of drugs. This can be considered when selecting a particular drug for an individual patient. Table 2Classification of drug–drug interactionsPharmaceutical interactions Mostly physical or chemical incompatibilities, e.g. chemical reactions or precipitations due to drug admixtures.Pharmacokinetic interactions Occur at the level of drug absorption, distribution, excretion, and metabolism. Frequently, the cytochrome P450 metabolising enzyme system and drug transporters such as P-glycoprotein are involved in such interactions.Pharmacodynamic interactions Directly related to the desired or undesired drug effects, e.g. the antitumoral effect or drug-associated toxicity. For some drugs with a high potential for interactions, therapeutic drug monitoring can be useful to detect and control interactions. The advantage of this approach is that the dosage of the drug of risk can be adapted to the measured individual plasma concentrations. Thus the treatment of both interacting drugs may be continued. Examples are anticonvulsant, antidepressant and antifungal drugs which are often administered to cancer patients as supportive medication [19]. In general, it is almost impossible for the physician to keep in mind all of the interaction mechanisms and potential consequences for the individual patient. Regular interaction checks by the pharmacist may solve this problem as he is the only health care professional who may have an overview of the drugs prescribed by various physicians and the medication taken by the patient on his own initiative. The avoidance of drug–drug interactions should be regarded as a multidisciplinary task [20]. The basis for each interaction check is to take the medication history of each patient and to update it regularly. Nowadays software tools are available facilitating rapid interaction checks and providing information on the mechanism and clinical relevance of an interaction. Each hospital and community pharmacy should have access to at least one of these tools. Medication errors In addition to the DRP described above, there are numerous risks for the occurrence of medication errors along the therapeutic path in oncology. A medication error is defined as any preventable event that may cause or lead to inappropriate medication use or patient harm, while the medication is in the control of the health care professional, patient, or consumer [21]. In 2000 the American Institute of Medicine published their report “To err is human: Building a Safer Health System” which analysed flaws in the health system and offered suggestions for its improvement [22]. Based on the results of a study conducted in Utah and Colorado in 1992 an incidence of about 3% adverse events in hospitalised patients was found, of which the majority of the non-operative events were adverse drug events (ADE) [23]. The authors concluded that iatrogenic injury continues to be a significant public health problem. Due to the narrow therapeutic range of many anticancer drugs, the impact of an ADE is more serious (and at worst lethal) compared to other systemically administered drugs. Antineoplastic agents were found to be the second most common group of drugs which caused lethal medication errors [24]. As well as quality and safety issues, the reduction of medication errors has a financial effect. The first source of medication errors is the prescription and ordering process: the wrong protocol may be chosen, the cumulative dose may be mixed up with the single dose, the route of administration may not be clearly indicated or misinterpretation of the physicians’ handwriting may lead to errors. A study carried out in a cytotoxic preparation unit in a French hospital revealed in only a 6 month period more than 300 medication errors out of 1,262 prescriptions for 285 patients. Most errors (>70%) were found to be simple physicochemical incompatibilities, e.g. incompatibility of the drug with the used matrix. However, also over- and underdosage as well as wrong medications were identified [25]. The authors conclude that most of these errors could possibly be avoided by a computerised prescription network. Computer-assisted solutions enable physicians to order the medication electronically which is nowadays one way to reduce the incidence of those errors. Mekhjian et al. evaluated the benefits of physician order entry systems and found that transcription errors could be reduced and speed in the ordering process could be improved. Even the duration of the stay in hospital could be reduced by the system [26]. Particularly in oncology, the computerised physician order entry (CPOE) has obvious advantages. A recent study has shown that CPOE in electronic medical records improves completeness of the medical record compared to paper charts independent of regimen complexity [27]. Prepared protocols facilitate correct ordering and double checking by the programme itself and by the pharmacist responsible. In addition, supportive medication, such as hydration or the antiemetic or antiallergic medication can be automatically proposed depending on the selected regimen. Moreover, the risk of misinterpretation of handwriting is eliminated. Recently, a CPOE system with integrated clinical decision support has been implemented in Ontario (Canada). This system, for example, alerts the clinician when maximum cumulative doses are reached, calculates body surface area, and estimates the creatinine clearance of the patient [28]. With all the advantages of these systems it has to be taken into account that new risks may arise from those new processes such as information errors due to the lack of clearness on the monitor [29]. The next step after ordering, the compounding of cytotoxic drugs also holds various risks, especially when performed on the ward. The implementation of central cytotoxic preparation units in pharmacy departments in the late 1980s has been one of the first measures to standardise the process in order to increase safety. Nevertheless, of 30,819 preparations surveyed in a study on incidence and risk factors of preparation errors 140 were found to be defective (0.45%) [30]. Less than half of those cases were classified as major errors including wrong dosage, wrong labelling or the use of incompatible diluents. Although the incidence is fairly low it must still be the aim to reduce this further by analysing the risk factors such as workload. Centralisation itself offers many more options for safety improvement. Standardised pharmaceutical validation leads to an increased interception of medication errors in antineoplastic treatment. This can be explained by the improved knowledge of the pharmacists involved in the validation process. In consequence of a standardised pharmaceutical validation process in the hematology–oncology department of a 550-bed university hospital the number of detected medication errors that did not reach the patients could be increased by 41% (from 14.08 medication errors/1,000 patient days to 19.83 medication errors/1,000 patient days) in the second year after the introduction of pharmacists in the multidisciplinary oncology team [31]. Using their expertise, pharmacists can facilitate protocol development and adherence, dose verifications and education of other health care practitioners [32]. Finally, storage and administration errors of cytotoxic drugs form a group of major medication errors with serious consequences for the patient. Therefore, it is mandatory that the assigned staff are well trained in all aspects of chemotherapy storage and administration. For many drugs, special storage conditions, e.g. refrigeration or protection from light, have to be considered. Detailed knowledge of the administered drugs is also very important in order to educate the patient and to react appropriately when incidents such as extravasations occur. In case of extravasation the necessary measures have to be initiated. Different drugs require different measures which the oncologically trained staff needs to know. An extravasation kit including essential items for first aid must be made available on site [33]. Incorrect or delayed treatment can cause serious damage to the patient, resulting in even the loss of a limb. Various surveys have been undertaken in order to find solutions for the prevention of medication errors. The non-punitive reporting of errors is a necessary requirement to be able to detect, analyse and consequently minimise the incidence and severity of medication errors in oncology. Safety measures with error reporting and analysis have been developed and implemented in order to improve the entire system [34]. The most effective improvements of treatment performance have been achieved by multidisciplinary system approaches integrating physicians, pharmacists and nurses [2, 32, 34–36]. Non-adherence Adherence (or compliance) is generally defined as the extent to which a person’s behaviour, taking medication, following a diet, and/or executing lifestyle changes, corresponds with agreed recommendations from a health care provider [37]. Non-adherence strongly compromises the success of a patient’s therapy, results in additional, potentially unnecessary, diagnostic and therapeutic procedures and thus generates further costs and possibly health problems as a consequence of the treatments themselves. It is important to note that the traditional concept that patients are solely responsible for taking their medication (and therefore non-adherence is a patient-driven problem) is misleading. The WHO understands adherence as a phenomenon consisting of five dimensions (Table 3). The magnitude of this problem is further emphasised by the fact that adherence rates for many long-term drug therapies range from only 40 to 78% [38, 39]. Table 3The five dimensions of adherence and examples of associated factors (adapted from [34]) Social/economic factorsHealth care team/system-related factorsCondition-related factorsTherapy-related factorsPatient-related factorsEconomic statusPatient-provider relationshipSeverity of symptomsComplexity of regimenAnxiety about side effectsCultural beliefsEducation of providersLevel of disabilityTreatment durationPatients’ motivationIlliteracyCapacity of systemRate of progressionChanges in treatmentPatients’ expectationsAgeDuration of consultationsCo-morbiditiesSide effectsForgetfulnessDistance from treatment centerMedication distribution systemAvailability of effective treatmentsPrevious treatment failuresPatients’ knowledge about illness It is a difficult task to assess patient adherence reliably. Patient self-reports, rates of prescription refills, patient diaries or pill counts as the sole basis for the measurement of adherence have been shown to be inadequate [38–40]. It has been demonstrated that these methods overestimate the degree to which patients adhere to their tamoxifen regimen [40]. The use of microelectronic adherence monitoring (Medication Event Monitoring System, MEMSTM, Fig. 1) provides a valuable estimate of the timing of events and insights into patients’ behaviour in taking medication. However, to prove actual intake of the medication, additional plasma concentration monitoring would be necessary. To obtain an optimal measurement of patient adherence, a combination of several methods is suggested [38]. Fig. 1Medication event monitoring system, MEMSTM: The pill bottle cap contains a micro-electronic circuit that registers when the bottle is opened. This data may then be transferred to a computer via a reading device Studies conducted in various disease areas have shown that a wide variety of methods to improve adherence is available. A recently published review classifies these methods into four general categories: patient education, improved dosage, increased hours when clinics are open, and improved communication between health care providers and patients [38]. All four categories require a collaborative approach including all members of the multidisciplinary health care team. Due to their position in the chain of health care providers, this requires the contribution of pharmacists, not only in the hospital setting but also in community pharmacies [41]. It has been shown that no single intervention strategy appears consistently more effective than another. Successful methods are complex and comprehensive as well as labour-intensive [38, 42]. The Cochrane Review on Interventions to Enhance Medication Adherence concludes that there is no evidence that low adherence can be “cured” and hence efforts to improve adherence must be maintained for as long as the treatment is needed [43]. Cancer patients seem to benefit especially from interventions towards an optimised adherence, resulting in improved outcomes [42]. Research investigating adherence in oncology settings has been mostly focussed on palliative care and supportive medication because chemotherapy has mainly been administered intravenously in hospitals or clinics. However, with the increasing availability and importance of oral anticancer drugs such as capecitabine, etoposide, vinorelbine, erlotinib and sorafenib, patients increasingly take over the responsibility for the correct administration of their prescribed therapy. Although cancer patients might be more adherent than other patient groups due to a high level of motivation, the slightest non-adherence can endanger the therapeutic goals [39]. Only few studies have been published investigating the level of adherence of cancer patients. In a cohort of outpatients receiving chemotherapy for haematologic malignancies adherence to the allopurinol and prednisolone prescription was measured based on plasma concentrations. It was shown that control patients without intervention were adherent only 17% of the time with allopurinol and 27% of the time with prednisone. Three specially developed intervention packages including education and home visits resulted in increased adherence rates of 44–48% and 33–38% of the time for allopurinol and prednisone, respectively [44]. Another study assessed adherence in patients with breast cancer receiving oral cyclophosphamide and found that 43% of the patient population met criteria for non-adherence according to both behavioural and dosage definitions [45]. In a recently published study in a cohort of 2816 women adherence to tamoxifen treatment was investigated using prescription refill data: 22.1% of the patients exhibited a discontinuation of treatment (non-persistence) within 1 year [46]. An open controlled trial is currently undertaken at the University of Bonn investigating the impact of intensified pharmaceutical care provision on the adherence of patients receiving oral capecitabine using the MEMS™-technology. Pharmacists’ interventions include the extensive provision of drug information to the patients and health care professionals, routine checks for drug interactions, preparation of written drug intake plans for patients, regular pharmacist consultations and structured documentation of the pharmaceutical care process [47]. Discussion The recognition of the described risks of the individual patient associated with complex drug therapies has led to the development of a conceptual framework for an advanced pharmacy practice philosophy. The concept of pharmaceutical care was introduced as a further development of the pharmaceutical profession gaining acceptance in Europe and worldwide [48, 49]. The American Society of Health System Pharmacists set up guidelines for standardised pharmaceutical care to ensure that pharmacists practicing pharmaceutical care work to the same standard [50]. Patients with complex drug regimens and/or chronic diseases and those who frequently need to be hospitalised might benefit from pharmaceutical care in particular. These criteria apply to many cancer patients. Within the pharmaceutical care process the application of agreed therapeutic algorithms can be assured on the individual basis. The adherence of the patient can be improved by patient education before and during the treatment cycles combined with patient counselling regarding drug therapy, adverse effects and complementary treatment options. The London oncology pharmacy group introduced guidelines for pharmaceutical care of cancer patients which not only include the actual ‘pharmaceutical care’ as such, but standardise the clinical pharmacy activities, dispensing, updating therapeutic policies, cytotoxic reconstitution, drug information, clinical trials, and the oncology training of pharmacists [51]. A group of British experts has drawn up a policy framework for commissioning cancer services. They suggest the establishment of structures which support the seamless care of cancer patients in the community setting in a network of all parties in order to make use of the respective specialist knowledge [52]. Accordingly, information flow at discharge from hospital to the community should be optimised utilising pharmaceutical care plans to ensure the efficient distribution of medication to the patient is not interrupted. Although there are some reports on the implementation of pharmaceutical care for cancer patients there is still little scientific evidence on the feasibility of pharmaceutical care and its actual benefit to the patient. In Canada, projects have been carried out which suggest to implement suitable outcome parameters to evaluate the impact of pharmaceutical services in oncology [53]. These have stimulated a wide discussion in Canadian health care politics regarding the necessity of the services offered. In Scotland, research is concentrating on the documentation and thereby standardisation of the provision of pharmaceutical care to cancer patients [54]. In Germany, various research projects are being carried out which investigate the feasibility and benefit of pharmaceutical care for patients with different indications. Patients with breast and colorectal cancer as well as patients receiving oral chemotherapy are the focus of projects at the University of Bonn. Another project is being carried out in Hamburg developing the provision of pharmaceutical care to lung cancer patients [2]. In terms of cancer many disciplines contribute to the care process. Thus, cross-profession and cross-sector cooperation is crucial in order to improve information flow. Over the last few years disease management programs (DMPs) have increasingly been introduced to cover the whole care process which begins with the early diagnosis of the disease. They aim at providing the optimal medical care by the implementation of evidence-based guidelines. Pharmaceutical care concepts seem to have the potential to support the goals of DMPs and could be easily integrated in programmes for cancer patients. However, it is necessary to document the impact of pharmaceutical care on patient outcomes in order to comply with the demand for transparency. Conclusions Systemic cancer therapy is particularly complex and hence associated with multiple risks for the patient as described above. Many of these risks are preventable by specific measures which can be taken by a particular health care provider or the patient. The pharmacist with his central position relating to drug dispensing and utilisation can contribute substantially by adding specific drug-related knowledge and offering patient-related services. Pharmaceutical care is designed as a framework that integrates individual contributions of the pharmacist into the entire therapeutic path. Model projects are urgently needed to assess the clinical, humanistic and economic outcome of patient-oriented pharmaceutical services. The integration of pharmaceutical care into disease management programmes might facilitate efficient collaboration between all health care professionals and hence improve effectiveness and safety of systemic cancer therapy.

Arch_Dermatol_Res-3-1-1910890

German evidence-based guidelines for the treatment of Psoriasis vulgaris (short version)

Psoriasis vulgaris is a common and chronic inflammatory skin disease which has the potential to significantly reduce the quality of life in severely affected patients. The incidence of psoriasis in Western industrialized countries ranges from 1.5 to 2%. Despite the large variety of treatment options available, patient surveys have revealed insufficient satisfaction with the efficacy of available treatments and a high rate of medication non-compliance. To optimize the treatment of psoriasis in Germany, the Deutsche Dermatologische Gesellschaft and the Berufsverband Deutscher Dermatologen (BVDD) have initiated a project to develop evidence-based guidelines for the management of psoriasis. The guidelines focus on induction therapy in cases of mild, moderate, and severe plaque-type psoriasis in adults. The short version of the guidelines reported here consist of a series of therapeutic recommendations that are based on a systematic literature search and subsequent discussion with experts in the field; they have been approved by a team of dermatology experts. In addition to the therapeutic recommendations provided in this short version, the full version of the guidelines includes information on contraindications, adverse events, drug interactions, practicality, and costs as well as detailed information on how best to apply the treatments described (for full version, please see Nast et al., JDDG, Suppl 2:S1–S126, 2006; or http://www.psoriasis-leitlinie.de). Introduction The Deutsche Dermatologische Gesellschaft (DDG) and the Berufsverband Deutscher Dermatologen (BVDD) have initiated a project to develop evidence-based guidelines for the treatment of plaque psoriasis. The full version of the Guidelines has been published in the Journal der Deutschen Dermatologischen Gesellschaft (JDDG 2006 Suppl 2) and is available at http://www.psoriasis-leitlinie.de [149]. This article summarizes the key points of the Guidelines. Background Psoriasis vulgaris is a common and chronic inflammatory skin disease which has the potential to significantly reduce the quality of life in severely affected patients. The incidence of psoriasis in Western industrialized countries is 1.5–2% [2]. Studies on the impairment of life quality have shown that, depending on the severity of the disease, a significant burden may exist in the form of a disability or psychosocial stigmatization [3]. Patient surveys have shown that the mental and physical impairment associated with psoriasis is comparable to that of other significant chronic conditions, such as type 2 diabetes or chronic respiratory diseases [4]. Patient surveys have shown that only 25% of psoriasis patients are highly satisfied with the outcome of their treatment, another 50% indicate moderate satisfaction, and approximately 20% report low treatment satisfaction [5]. In addition, there is a high non-compliance rate in the intake of medication of up to 40% [1]. Experience has shown that treatment selection for patients with psoriasis vulgaris is more commonly based on traditional concepts than on evidence-based data on the efficacy of various therapeutic options. In addition, it appears that systemic therapies are occasionally not applied in situations where they are needed due to the increased effort involved in monitoring the patients for unwanted side effects and possible interactions with other drugs. Goal of the Guidelines The overall objective of the Guidelines is to provide dermatologists in clinics and private practice with an accepted and evidence-based decision-making tool for the selection and implementation of a suitable and efficacious treatment for patients with psoriasis vulgaris. The focus of the Guidelines is on the induction therapy of mild, moderate, and severe plaque-type psoriasis in adults.Physicians’ personal experiences and traditional therapeutic concepts should be supplemented and, if necessary, replaced by an evidence-based assessment of the efficacy of individual therapies in psoriasis vulgaris.The guidelines provide in-depth explanations of the available systemic and topical treatments for psoriasis, including the different photo- and photochemical therapies, and provide detailed descriptions of the administration and safety aspects.By providing background information on the profile of the available drugs, including efficacy, safety, and aspects of practicality and costs, the Guidelines should facilitate the process of selecting an appropriate treatment for each individual patient. This should help increase compliance and optimize the benefit/risk ratio of psoriasis therapies. Methods A detailed description of the methodology employed in developing the Guidelines can be found in the Method Report on the Guidelines (http://www.psoriasis-leitlinie.de). Basis of data A systematic search of the literature was carried out in May 2005 with the objective of assessing the effectiveness of individual therapeutic options. This search yielded 6,224 publications of which 142 studies fulfilled the inclusion criteria for the Guidelines (see Box 1) and were therefore included in the assessment of the effectiveness of the relevant treatment. Various other aspects were evaluated on the basis of information presented in available literature, without a systematic analysis, and the years of personal experience of the experts. Evidence assessment The efficacy and effectiveness of each intervention was evaluated using evidence-based criteria. The methodological quality of each individual study was assessed using the following grades of evidence (GE):A1Meta-analysis that includes at least one randomized study with grade A2 evidence. In addition, the results of the different studies included in the meta-analysis must be consistent with one another.A2A high-quality (e.g. sample-size calculation, flow chart, intention-to-treat (ITT) analysis, sufficient size) randomized, double-blind comparative clinical study.BRandomized clinical study of lesser quality or other comparative study (non-randomized, cohort, or case-control study).CNon-comparative study.DExpert opinion.Individual interventions (i.e., as monotherapy) were rated according to the following levels of evidence (LE):Intervention is supported by studies with grade A1 evidence or studies with grade A2 evidence whose results are predominantly consistent with one another.Intervention is supported by studies with grade A2 evidence or studies with grade B evidence whose results are predominantly consistent with one another.Intervention is supported by studies with grade B evidence or studies with grade C evidence whose results are predominantly consistent with one another.Little or no systematic empirical evidence Therapeutic recommendation A distinct rating of the therapeutic options or a strict clinical algorithm cannot be defined for the treatment of psoriasis vulgaris. The criteria for selecting a particular therapy are complex. The decision for a specific treatment should be based on the profile of the available drugs and the characteristics of a given patient. The decision for or against a therapy remains a case-by-case decision. These Guidelines provide a reasonable form of assistance in deciding on a suitable therapy and are an instrument for optimizing the required therapeutic process. The recommendations formulated in the text are supported graphically in selected key recommendations by the following indications of the strength of the therapeutic recommendation:↑↑Measure is highly recommended↑Measure is recommended→Neutral↓Measure is not recommended↓↓Measure is highly inadvisableThe strength of the recommendation reflects both a treatment’s efficacy and the level of evidence supporting it as well as aspects of safety, practicality, and the cost/benefit ratio. A consensus on the strength of the recommendation was reached during the Consensus Conference. Results Therapeutic options are named and discussed in alphabetical order. Therapeutic strategies Fig. 1 Overview of the therapeutic options evaluated for chronic plaque psoriasis (the therapeutic options are listed alphabetically and do not represent a ranking) Evaluation of topical and systemic therapies in tabular form These tables are intended to provide a rough orientation for evaluating the therapeutic options. Cumulative calculations of the individual aspects for the overall evaluation of the therapeutic options are not possible and cannot be drawn upon for the final analysis of a therapeutic option. The product assessment for each individual patient may deviate greatly. A direct comparison between systemic and topical therapies is not possible because of the different severity of the psoriasis lesions of patients treated with topical or systemic treatments. The evaluations reported here were made on the basis of data extracted from the literature and expert opinions. For further details refer to the Methods Report at www.psoriasis-leitlinie.de Topical monotherapy Phototherapy and systemic monotherapy (a) Efficacy The evaluation of the efficacy column reflects the percentage of patients who achieved a reduction in the baseline Psoriasis Area and Severity Index (PASI) of ≥75%. ScaleSystemic therapyTopical therapy++++Approx. 90%Approx. 60%+++Approx. 70%Approx. 45%++Approx. 50%Approx. 30%+Approx. 30%Approx. 15%+/−Approx. 10%Approx. 5%–Not definedNot defined The evidence level applies only to the estimate of efficacy. (b) Safety/tolerance in induction therapy or maintenance therapy This refers to the risk of occurrence of severe adverse drug reactions or the probability of adverse drug reactions that would result in the discontinuation of therapy. (c) Practicality (Patient) This evaluation analyzes the effort involved in handling and administrating the treatment regimen by the patient. (d) Practicality (Physician) This aspect considers the amount of work (documentation, explanation, monitoring), personnel and equipment needs, time for physician/patient interaction, remuneration of therapeutic measures, invoicing difficulties/risk of recourse claims from the health insurance companies. (e) Cost/benefit Consideration of the costs of an induction therapy or a maintenance therapy. The evaluations of safety/tolerance in induction therapy or maintenance therapy as well as practicality for the physician or patient and the cost/benefit were performed using a scale ranging from poor (–) to good (++++). The gradation between these two extremes was made based on expert opinion and unsystematic literature search. A level of evidence was not given for these evaluations since no systematic literature review was performed. Evaluation of topical therapies Calcineurin inhibitors Table 1 Tabular summary Calcineurin inhibitors First approved in Germany  Pimecrolimus (Elidel®) 2002 (Atopic dermatitis, not approved for psoriasis vulgaris)  Tacrolimus (Protopic®) 2002 (Atopic dermatitis, not approved for psoriasis vulgaris) Recommended initial dosage Pimecrolimus cream: 1× to 2× dailyTacrolimus: 1× to 2× daily (application on the face: Begin with 0.03% salve; increase later dosage to 0.1% salve) Recommended maintenance dosage Individual therapeutic adjustment Expected beginning of clinical effect After approximately 2 weeks Response rate No data available Important contraindications Pregnancy and nursing (due to lack of experience) skin infections, immune suppression Important adverse drug reactions (ADRs) Burning sensation on skin, skin infections Important drug interactions None known Other Cave: Do not combine with phototherapy! Photoprotection! Corticosteroids Table 2 Tabular summary Corticosteroids First approved in Germany 1956 (Psoriasis vulgaris) Recommended control parameters None Recommended initial dosage One to two times daily Recommended maintenance dosage Gradual reduction following onset of effect Expected beginning of clinical effect After 1–2 weeks Response rate Betamethasone dipropionate, two times daily: marked improvement or clearance of the skin lesions in 46–56% patients after 4 weeks (LE 1) Important contraindications Skin infections, rosacea, perioral dermatitis Important ADRs Skin infections, perioral dermatitis, skin atrophy, hypertrichosis, striae Important drug interactions None Other – Coal tar Table 3 Tabular summary Coal tar First approved in Germany Listed active ingredient since 2000 (DAC on page 170), historical application, various tar-containing externals are licensed as drugs, application of tar as anti-psoriatic following publication by Goeckermann in 1925 Recommended control parameters After long-term application/application on large areas: if needed clinical controls for potential development of skin carcinoma Recommended initial dosage 5–20% salve preparations or gels for local therapy, 1× daily Recommended maintenance dosage No long-term application (max. 4 weeks, DAC 2000) Expected beginning of clinical effect After 4–8 weeks, efficacy improves in combination with UV application Response rate There is insufficient data available on the response rate as a monotherapy (LE 4) Important contraindications Pregnancy and nursing Important ADRs Color, odor, carcinogenic risk, phototoxicity—which is part of the desired effect Important drug interactions Not known with topical use Other DAC 2000 (on page 170), Hazardous Goods Directive Appendix 4 No. 13 Dithranol Table 4 Tabular summary Dithranol First approved in Germany Psoralon 1983 (Psoriasis vulgaris) Psoradexan 1994 (Psoriasis vulgaris) Micanol 1997 (Psoriasis vulgaris) Recommended control parameters Intensity of irritation Recommended initial dosage Begin with 0.5% preparation for long-term therapy or 1% for short-contact therapy, then increase if tolerated Recommended maintenance dosage Not recommended for maintenance therapy Expected beginning of clinical effect After 2–3 weeks Response rate Marked improvement or clearance of skin lesions in 30–50% of the patients (LE 2) Important contraindications Acute, erythrodermic forms of psoriasis vulgaris; pustular psoriasis Important ADRs Burning and reddening of the skin in >10% Important drug interactions – Other – Tazarotene Table 5 Tabular summary Tazarotene First approved in Germany 1997 (psoriasis vulgaris) Recommended control parameters Check development of skin irritations Recommended initial dosage Begin with one treatment daily of tazarotene gel 0.05% in the evening for approximately 1–2 weeks Recommended maintenance dosage If necessary continue for 1–2 weeks with tazarotene gel 0.1% Expected beginning of clinical effect After 1–2 weeks Response rate After 12 weeks therapy with 0.1% tazarotene gel improved findings of at least 50% in approximately 50% of the patients (LE 2) Important contraindications Pregnancy, nursing Important ADRs Pruritus, burning sensation of skin, erythema, irritation Important drug interactions Avoid concomitant use of preparations with irritating and drying properties Other – Vitamin D3 and analogues Table 6 Tabular summary Vitamin D3 and analogues First approved in Germany  Calcipotriol 1992 (Psoriasis vulgaris)  Tacalcitol 1994 (Psoriasis vulgaris)  Calcitriol 1999 (Psoriasis vulgaris)  Calcipotriol/Betamethasone 2002 (Psoriasis vulgaris) Recommended control parameters Monitor for skin irritations Recommended initial dosage Calcipotriol: 1× to 2× daily to affected locations, up to a maximum of 30% of the body surface Tacalcitol: 1× daily to affected locations, up to a maximum of 20% of the body surface Calcitriol: 2× daily to affected locations, up to a maximum of 35% of the body surface Recommended maintenance dosage Calcipotriol: 1× to 2× daily, up to 100 g/week for up to 1 year Tacalcitol: 1× daily for 8 weeks, for up to 18 months, on a maximum of 15% of the body surface with up to 3.5 g daily Calcitriol: insufficient experience with the application for more than 6 weeks Expected beginning of clinical effect After 1–2 weeks Response rate Between 30 and 50% of the patients demonstrated a marked improvement or clearance of the lesions after 4–6 weeks (LE 1) Important contraindications Diseases with abnormal calcium metabolism, severe liver and renal diseases Important ADRs Skin irritation (reddening, itching, burning) Important drug interactions Drugs which elevate the calcium levels, (e.g. thiazide diuretics), no concomitant application of topical salicylic acid preparations (inactivation) Other Exposure to UV light results in inactivation of the vitamin D3-analogues Phototherapy Table 7 Tabular summary Phototherapy First approved in Germany Clinical experience, depending on the modality for >50 years Recommended control parameters Regular skin inspection (UV erythema) Recommended initial dosage Individual dose depends on skin type; options: • UVB: 70% of minimum erythema dose (MED) • Oral PUVA (photochemotherapy): 75% of the minimum phototoxic dose (MPD) • Bath/cream PUVA: 20–30% of MPD Recommended maintenance dosage Increase according to degree of UV erythema Expected beginning of clinical effect After 1–2 weeks Response rate In >75% of the patients PASI, 75 after 4–6 weeks (LE 2) Important contraindications Photo-dermatoses/photosensitive diseases, skin malignancies, immunosuppressionOnly PUVA: pregnancy and nursing Important ADRs Erythema, itching, blistering, malignanciesOnly oral PUVA: nausea Important drug interactions Drugs causing phototoxicity or photoallergy Other Combination with topical preparations acts synergistically, PUVA may not be combined with cyclosporine Evaluation of systemic therapies Efalizumab Table 8 Tabular summary Efalizumab First approved in Germany September 2004 (psoriasis vulgaris) Recommended control parameters Prior to therapy exclusion of significant infections, complete blood count, liver values Recommended initial dosage 0.7 mg/kg body weight (BW) per week Recommended maintenance dosage 1 mg/kg BW per week Expected beginning of clinical effect After 4–8 weeks Response rate PASI 75 for approximately 30% of the patients after 12 weeks (LE 1) Important contraindications (limited selection) Chronic or acute infections, pregnancy, malignant diseases, immune deficiencies, no vaccinations before or during treatment Important ADRs (limited selection) Flu-like injection reactions, leukocytosis and lymphocytosis, rebound, exacerbation and arthralgia, thrombocytopenia Important drug interactions Not known Other Stop therapy due to the risk of exacerbation and rebound if a PASI reduction of 50% is not achieved after 12 weeks Table 9 Laboratory controls during treatment with efalizumab aHb (hemoglobin), HCT (hematocrit), erythrocytes, leukocytes, differential blood count, platelets bALT alanine aminotransferase, AST aspartate aminotransferase Etanercept Table 10 Tabular summary Etanercept First approved in Germany 2002 (Psoriasis arthritis )/2004 (psoriasis vulgaris) Recommended control parameters Prior to therapy exclusion of tuberculosis, complete blood count, liver and renal values, urinanalysis Recommended initial dosage Twice 25 mg per week or 2× 50 mg/week Recommended maintenance dosage Twice 25 mg per week Expected beginning of clinical effect After 4–8 weeks, at the latest after 12 weeks Response rate PASI 75 in 34% (2 × 25 mg) or 49% (2 × 50 mg) of the patients at the end of the induction phase (12 weeks) (LE 1) Important contraindications (limited selection) Infections, pregnancy, nursing, heart failure NYHA III–IV Important ADRs (limited selection) Local reactions, infections Important drug interactions (limited selection) Anakinra (IL-1 receptor antagonist) Other – Table 11 Laboratory controls during treatment with etanercept Infliximab Table 12 Tabular summary Infliximab First approved in Germany 2004 (Psoriasis arthritis)/2005 (psoriasis vulgaris) Recommended control parameters Prior to therapy exclusion of tuberculosis, during therapy: leukocyte and platelet counts, liver value controls, signs of clinical infection Recommended initial dosage 5 mg/kg BW at week 0, 2, 6 Recommended maintenance dosage 5 mg/kg BW in dosage intervals of 8 weeks Expected beginning of clinical effect After 1–2 weeks Response rate PASI 75 in ≥80% of the patients with moderate to severe psoriasis vulgaris (LE 1) Important contraindications (limited selection) Acute or chronic infections, tuberculosis, cardiac failure NYHA III–IV, pregnancy and nursing Important ADRs (limited selection) Infusion reactions, severe infections, progression of heart failure NYHA III–IV, very rare liver failure, autoimmune phenomena Important drug interactions (limited selection) Anakinra (IL-1 receptor antagonist) Other – Table 13 Laboratory controls during treatment with infliximab aHb, HCT, erythrocytes, leukocytes, differential blood count, platelets Cyclosporine Table 14 Tabular summary Cyclosporine First approved in Germany 1983 (Transplantation medicine) 1993 (Psoriasis vulgaris) Recommended control parameters Interview/examination: • Status of skin and mucous membranes • Signs of infection • Neurological, gastrointestinal symptoms • Blood pressure Laboratory controls: see Table 14 Recommended initial dosage 2.5–3 (max. 5) mg/kg BW Recommended maintenance dosage Interval therapy (between 8 and 16 weeks) with dosage reduction at the end of the induction therapy (e.g., 0.5 mg/kg BW every 14 days) or Continuous long-term therapy with dosage reduction (e.g. by 50 mg every 4 weeks after week 12) and a dosage increase by 50 mg with relapse Maximum total duration of therapy: 2 years Expected beginning of clinical effect After approximately 4 weeks Response rate Dose-dependent: after 8–16 weeks with 3 mg/kg BW; PASI 90 in approximately 30–50% of patient and PASI 75 in approximately 50–70% of patients (LE 1) Important contraindications (limited selection) Absolute: Reduced renal function, insufficiently controlled arterial hypertension, uncontrolled infections, relevant malignancies (current or previous, in particular hematologic diseases and cutaneous malignancies with the exception of basal cell carcinoma) Relative: Relevant hepatic dysfunction, pregnancy and nursing, concomitant use of substance which interacts with cyclosporine, concomitant UV-therapy or prior PUVA-pre-therapy with cumulative dosage >1000 J/cm², concomitant application of other immunosuppressives, retinoids or long-term prior-therapy with methotrexate (MTX) Important ADRs (limited selection) Renal failure, increase of blood pressure, liver failure, nausea, anorexia, vomiting, diarrhea, hypertrichosis, gingival hyperplasia, tremor, weariness, parasthesia, hyperlipidemia Important drug interactions (limited selection) Increase of the cyclosporine level (CYP3A inhibition) through: Allopurinol, calcium antagonists, amiodarone, antibiotics (macrolides, clarithromycin, josamycin, ponsinomycin, pristinamycin, doxycycline, gentamicin, tobramycin, ticarcillin, quinolones), ketoconazole, oral contraceptives, methylprednisolone (high dosages), ranitidine, cimetidine, grapefruit juice Decrease of the cyclosporine level (CYP3A induction) through: Carbamazepine, phenytoin, barbiturates, metamizole, St. John’s wort Possible reinforcement of nephrotoxic adverse drug reactions through: Aminoglycosides, amphotericin B, ciprofloxacin, acyclovir, non-steroidal antiphlogistics Specific interactions: Potassium-saving substances: increased risk of hyperpotassemia Reduced clearance of: Digoxin, colchicine, prednisolone, HMG-CoA reductase inhibitors (e.g. lovastatin), diclofenac Other Increased risk of lympho-proliferative diseases in transplant patients. Increased risk of squamous cell carcinoma in psoriasis patients following excessive phototherapy. Only moderately effective in and not approved for psoriatic arthritis Also used successfully in the therapy of chronic-inflammatory diseases in children Table 15 Laboratory controls during treatment with cyclosporine aErythrocytes, leukocytes, platelets bALT, AST, AP (alkaline phosphatase), γGT (gamma glutamyl transpeptidase), bilirubin cSodium, potassium dRecommended twice (fasting) and week-2 and 0 eOnly with indication (e.g. muscle cramps) Fumaric acid esters Table 16 Tabular summary Fumaric acid esters First approved in Germany 1995 (Psoriasis vulgaris) Recommended control parameters Serum creatinine, transaminases/γGT, complete blood count including differential blood count, urinanalysis Recommended initial dosage According to recommended dosage regimen see Table 17 Recommended maintenance dosage Individually adapted dosage Expected beginning of clinical effect After approximately 6 weeks Response rate PASI 75 in 50–70% of the patients at the end of the induction phase after 16 weeks (LE 2) Important contraindications (limited selection) Chronic diseases of the gastrointestinal tract and/or the kidneys and chronic diseases, which are accompanied by an impairment of the leukocyte count or functions, malignant diseases, pregnancy and nursing Important ADRs (limited selection) Gastrointestinal complaints, flush, lymphopenia, eosinophilia Important drug interactions None known Other – Table 17 Dosage regimen for Fumaderm therapy Fumaderm initial Fumaderm Week 1 1-0-0 Week 2 1-0-1 Week 3 1-1-1 Week 4 1-0-0 Week 5 1-0-1 Week 6 1-1-1 Week 7 2-1-1 Week 8 2-1-2 Week 9 2-2-2 Table 18 Laboratory controls during treatment with fumaric acid esters aLeukocytes, platelets, erythrocytes, differential blood count Methotrexate Table 19 Tabular summary Methotrexate First approved in Germany  Lantarel® 1991 (Psoriasis vulgaris)  Metex® 7.5/10 mg 1992 (Psoriasis vulgaris)  Metex® 2.5 mg 2004 (Psoriasis vulgaris) Recommended control parameters Complete blood count (Hb, HCT, differential blood count, platelets), renal function (serum creatinine, urea, urine sediment), liver values (serum transaminases), III-procollagen amino terminal propeptides Recommended initial dosage 5–15 mg per week Recommended maintenance dosage 5–22.5 mg per week depending on effect Expected beginning of clinical effect After 4–8 weeks Response rate PASI 75 in approximately 60% of the patients at the end of the induction phase of 16 weeks (LE 3) Important contraindications (limited selection) Absolute contraindications: Desire to have children (for both men and women), pregnancy and nursing, inadequate contraception, drug consumption, alcoholism, known sensitivity to active ingredient methotrexate (e.g. pulmonary toxicity), bone marrow dysfunction, severe liver disease, severe infections, immunodeficiency, active peptic ulcers, hematologic changes (leucopenia, thrombocytopenia, anemia), renal failure Relative contraindications: Kidney disorders, liver disorders, history of arsenic consumption , chronic congestive cardio-myopathy, adiposity, old age, diabetes mellitus, history of hepatitis, lack of patient compliance, ulcerative colitis, diarrhea, NSAID use, gastritis Important ADRs (limited selection) Liver fibrosis/cirrhosis, pneumonia/alveolitis, bone marrow depression, renal damage, alopecia (reversible), nausea, weariness, vomiting, elevated transaminases, infection, gastrointestinal ulcerations, nephrotoxicity Important drug interactions (limited selection) Cyclosporine, salicylates, sulfonamides, probenecide, penicillin, colchicin, NSAIDs (naproxene, ibuprofene, etc.), ethanol, co-trimoxazole, pyrimethamine, chloramphenicol, sulfonamides, prostaglandin synthesis inhibitors, cytostatics, probenecide, barbiturates, phenytoin, retinoids, sulfonamides, sulfonylurea, tetracyclines, co-trimoxazol, chloramphenicol, dipyridamole, retinoids, ethanol, leflunomide Other Consistent avoidance of alcohol, X-ray of the lungs prior to beginning therapy Table 20 Laboratory controls during treatment with MTX [137] aHb, HCT, erythrocytes, leukocytes, differential blood count, platelets bALT, AST, AP, γGT, albumin, bilirubin, lactate dehydrogenase (LDH) cLiver biopsy instead of a liver sonography in risk groups Retinoids Table 21 Tabular summary Acitretin First approved in Germany 1992 (Psoriasis vulgaris) Recommended control parameters Erythrocyte sedimentation rate (ESR), complete blood count, liver values, renal values, blood lipid values, pregnancy test, x-ray control of the bones in case of long-term therapy Recommended initial dosage 0.3–0.5 mg/kg BW per day for approximately 4 weeks, then 0.5–0.8 mg/kg BW Recommended maintenance dosage Individual dosaging dependent on the results and tolerance Expected beginning of clinical effect After 4–8 weeks Response rate Widely variable and dose-dependent, no definite statement possible, partial remission (PASI 75) in 25–75% of the patients (30–40 mg per day) (LE 3) in studies Important contraindications (limited selection) Renal and liver damage, desire to have children in female patients of child-bearing age, pregnancy, nursing, alcohol abuse, manifest diabetes mellitus, wearing of contact lenses, history of pancreatitis, hyperlipidemia requiring drug treatment Important ADRs (limited selection) Hypervitaminosis A (e.g., cheilitis, xerosis, nose-bleeding, alopecia, increased skin fragility) Important drug interactions (limited selection) Phenytoin, tetracyclines, methotrexate, alcohol, mini-pill Other Contraception up to 2 years after discontinuation in female patients of child-bearing age aDouble contraception is recommended (e.g., condom + pill; IUD/Nuva-Ring + pill; cave: no low-dosed progesterone preparations/mini-pills) during and up to 2 years after the end of therapy; effectiveness is reduced by acitretin Table 22 Laboratory controls during treatment with Acitretin aHb, HCT, leukocytes, platelets bPreferably assayed twice (week-2 and 0); HDL, high-density lipidoprotein Other therapies Climate/balneotherapy Table 23 Tabular summary Climate/balneotherapy First approved in Germany Clinical experience with balneotherapy has existed for more than 200 years Recommended control parameters Regular skin inspection Recommended initial dosage Therapy regimens vary depending on the institution/location Recommended maintenance dosage Therapy regimens vary depending on the institution/location Expected beginning of clinical effect Varies greatly Response rate Varies greatly (LE 4) Important contraindications Dependent on modality selected Important ADRs Dependent on modality selected Important drug interactions Not applicable Other Balneotherapy and climate therapy are frequently combined Psychosocial therapy Notes on the use of the Guidelines The presentation of the therapies deliberately focused on those aspects deemed particularly relevant by a panel of experts. Aspects which are not of specific importance for a certain intervention, but which are part of the physician’s general obligations to the patient, such as the investigation of intolerance and allergies toward certain drugs or the exclusion of contraindications, are not individually listed, but it is nevertheless taken for granted that these are part of the physician’s duty to provide care. It is recommended that each and every user carefully reads and follows the product information and compare it with the recommendations in the Guidelines on dosaging, contraindications, and drug interactions for completeness and currentness. Every dose or application is administered at the user’s own risk. The authors and the publishers kindly request that users inform them of any inaccuracies that they might notice. The users are requested to keep themselves constantly informed of any new findings subsequent to the publication of the Guidelines.

Support_Care_Cancer-3-1-2092408

Testicular cancer: a longitudinal pilot study on stress response symptoms and quality of life in couples before and after chemotherapy

Goals of work The current study was designed to longitudinally examine stress response symptoms (SRS) and quality of life (QoL) in couples confronted with disseminated testicular cancer. The objectives were to examine couples’ patterns of adjustment over time and possible differences in adjustment between the patient and his partner. Introduction Testicular cancer is a rare disease, although it is the most common tumor in men aged 20–35 years. About 500 new cases are diagnosed each year in The Netherlands [1]. Testicular cancer is distinguished into seminomas and nonseminomas, and each type accounts for about half of the total. Half of the men with nonseminoma are diagnosed with disseminated disease that is treated with chemotherapy [2]. Since the introduction of cisplatin-based chemotherapy in the late 1970s, up to 80% of the patients with disseminated disease can be cured [3–5]. Chemotherapy for testicular cancer has several acute physical side effects like nausea, fatigue, and neuropathy [6]. On a psychological level, patients receiving chemotherapy have been found to report anxiety, depression, and distress [7–10]. The partners of testicular cancer patients receiving chemotherapy may encounter adverse sequelae as well. Partners have to struggle with the fear of potentially losing their significant other. Partners are the primary source of information for family members and friends while their husbands are admitted to the hospital. In addition to this social task, partners often take on the caregiver role between chemotherapy cycles. Caregiving for cancer patients was found to negatively impact a caregiver’s physical and mental health [11–13]. The impact of testicular cancer on younger couples might differ from the impact on couples who face cancer at an older age. The young couples confronted with testicular cancer might not have been together for a substantial period of time. This can make the relationship more vulnerable to the stressors induced by major life events. Earlier studies have shown that relationships of shorter duration are more sensitive to disruption after a testicular cancer diagnosis [14, 15], although another study did not find such a connection [16]. Also, couples are confronted with possible treatment-related infertility and sexual difficulties at a time in life where partners are often focused on starting a family. Couples may face several individual and dyadic stressors after the diagnosis of testicular cancer, leading to the experience of distress and lowered quality of life (QoL). This is particularly true when chemotherapy is part of the treatment protocol, as it is a more demanding treatment modality than surgery alone in the case of testicular cancer. Several prospective studies focused on dyadic adjustment and functioning from cancer diagnosis in a variety of sites, up to 1 1/2 years later [17–20]. Although both patients and partners were reported to suffer from distress, levels of distress did not necessarily correspond within couples. Different patterns of distress and adjustment were found to be associated with gender and health status. Some studies reported that female partners were most vulnerable because they reported higher levels of distress than female and male patients and male partners [20–22]. One pilot study examined the course of distress and QoL in testicular cancer patients before, during, and after chemotherapy (complete data on ten patients). Patients reported highest levels of distress shortly before the start of chemotherapy. Distress and anxiety decreased over time [23]. Another recent prospective study on a large group of testicular cancer patients receiving chemotherapy focused on global QoL during the first 2 years after diagnosis. A considerable impairment in global QoL was found after 3 months. However, after 2 years 36% of patients reported improved functioning compared to baseline [24]. At the University Medical Center Groningen (UMCG) in The Netherlands, a tertiairy referral center for patients with testicular cancer, and at the University of Texas M. D. Anderson Cancer Center, USA considerable institutional research was done into epidemiology, short- and long-term medical, and sexual outcomes of testicular cancer [25–28]. These cross-sectional and retrospective studies were done on QoL of testicular survivors [29], their partners [30], and marital and sexual satisfaction [16, 31–33]. We chose to expand these findings by performing a prospective study, and the findings of Trask and Fossa by including partners in our prospective study. Patterns of stress response symptoms (SRS) and QoL of patients and partners were examined at three time points during the first year. SRS can occur after a range of traumatic events, including cancer. They involve intrusive thoughts and avoidance of thoughts and situations that remind them of the event, and are often studied in cancer patients [34]. Goals of the present study were: (1) to explore differences and relationships between patients and partners’ SRS and QoL and to relate those to relationship aspects (duration and presence of children); (2) to examine change over time in SRS and QoL in patients and partners; (3) to relate earlier levels of SRS and QoL in patients and partners to later levels; and (4) to examine differences in QoL of patients and partners with that of a reference group of men and women. Materials and methods Procedure and participants All patients diagnosed with a disseminated nonseminomatous testicular tumor who consecutively visited the UMCG in The Netherlands for treatment between April 2001 and March 2004 and who were married or cohabiting were approached for this study. Only patients who received chemotherapy and had a steady relationship (n=30) could be included in this study. Other inclusion criteria were age over 18 years at study entry, sufficient command of the Dutch language, no previous treatment for cancer, and absence of a psychiatric history. The study was introduced to the patients and their partners after orchiectomy (removal of the affected testicle) was performed. Couples received a questionnaire at the following three time points: after orchiectomy but before the start of chemotherapy (T1), immediately after completion of chemotherapy, which is approximately 3 months after T1 (T2), and 1 year after T1 (T3). The patients received four cycles of cisplatin, etoposide, and bleomycin with a 3-week interval between each cycle. Couples received a letter with information about the objectives of the study, an informed consent form, the questionnaires, and a prepaid return envelope. Thirty of the 70 patients diagnosed during the above-mentioned time period appeared to meet all of the inclusion criteria. They and their partners were approached to participate in the study. Of the eligible 30 couples, 21 participated (70%). Two couples did not complete all three measurement times, therefore, only 19 couples provided complete data. The study was approved by the Medical Ethics Committee of the UMCG. Measurements Sociodemographics: Data on the following sociodemographic variables were collected at T1: age, educational level, employment status, presence of children, and duration of the relationship. Highest educational level completed was measured on a seven-point scale: primary school [1], and lower vocational [2], lower secondary [3], middle secondary [4], high secondary [5], higher vocational degrees [6], and advanced university [7]. Employment status could be indicated as employed for wages, housekeeping, student, unemployed, unable to work, or retired. SRS were measured with the Impact of Event Scale (IES) [35, 36]. This scale (15 items) makes an inventory of the extent to which a subject is currently occupied with an event by measuring intrusion (intrusively experienced ideas, images, feelings, or bad dreams about the event; 7 items) and avoidance of unpleasant feelings or memories of the event (8 items), resulting in a total score of SRS. Patients and partners rated the frequency of SRS with respect to his cancer during the preceding 7 days. Examples of items include: “Any reminder brings back feelings about it” (intrusion) and “I try to banish it from my memory.” The IES is a valid instrument for measuring cancer-related SRS [37, 38]. With this questionnaire, information was obtained about the degree to which confrontation with testicular cancer was influencing the current daily life of the respondent. Higher scores indicate more SRS. The Dutch version of the IES indicates a total score of more than 26 as severe SRS, for which psychological help is recommended. Reliability of this scale was good for patients (Cronbach’s alpha for the different measurement times ranged from 0.82 to 0.89) and for partners (Cronbach’s alpha for the different measurement times ranged from 0.78 to 0.93). QoL was measured with three subscales of the RAND-36 [39]: physical functioning (ten items), social functioning (two items) and mental health (five items). To avoid statistical problems due to multiple comparisons, we chose these three subscales as a representation of overall QoL. The RAND-36 measures generic QoL. After recoding and transformation, scores on the subscales could range from 0 to 100. Higher scores indicate a better QoL. Reliability of these scales was good to very good for patients (Cronbach’s alpha for the different measurement times ranged from 0.67 to 0.90) and for partners (Cronbach’s alpha for the different measurement times ranged from 0.75 to 0.92). The Dutch manual for the RAND-36 provides reference scores. These comprised the mean scores of a group of 691 nonselected men and 372 nonselected women from a random representative sample of persons aged 18 years and older from the population register of a municipality in the north of The Netherlands (number of inhabitants=108,000). The mean age of the persons in the total random sample was 44.1 years (range 18–89 years) [39]. Statistical analyses The database consisted of matched pairs of patients and partners, making analyses on pair level possible. Paired t tests and chi-square test were performed to examine differences in sociodemographics between patients and partners. Wilcoxon signed rank test (because of small sample size) and correlations were computed to examine differences and relationships in SRS and QoL between patients and partners. Independent t tests were performed for having children (yes or no) with SRS and QoL. Pearson’s product moment correlations were calculated to examine the relationship between duration of the relationship and patients and partners’ SRS and QoL. Repeated measures analysis of variance (ANOVA) was used to examine change over time in functioning. Pearson’s product moment correlations were calculated to examine relationships between the functioning of patient and that of the partner and between measurement times. Strong correlation coefficients (>0.50) indicate consistency between measurement times and stable responses between earlier and later levels of functioning. Independent t tests were performed to compare QoL of patients and partners with those of a reference group of men and women. Results Sociodemographic and treatment-related variables Sociodemographic and treatment related variables are reported in Table 1. Patients were older than partners (t=−3.1, p<0.01). The mean time couples had been together was 5.7 years. Eighteen patients were employed for wages and 1 was a student, whereas 11 partners were employed for wages, 4 were home-keepers and the remaining 4 partners were students (chi-square=19.0, p=0.04). Of the 21 couples, 5 had children and 14 did not. After chemotherapy was completed (at T2), all patients were restaged. Patients without biochemical or radiological abnormalities were considered to have reached complete remission (n=8). In case of proven residual disease, a resection of residual retroperitoneal tumor mass (RRRTM) was performed. Eleven patients underwent this surgery and were considered to be in complete remission afterward. None of the 19 patients experienced a relapse during follow-up (up to T3). Table 1Descriptives of sociodemographic and treatment-related variables PatientPartnerMeanSDMeanSDAge (years)b31.66.628.97.6 Range19.9–43.519.8–44.7Duration relationship (years)5.76.3 Range1–22Education level3.61.44.01.6N%N%Employment statusb Employed for wages18951158 Student15421 Housewife421Children Yes526 No1474RRRTM + Yes1158 No842RRRTM Resection of residual retroperitoneal tumor massap<0.01bp<0.05 Patients’ and partners’ SRS and QoL At T1, SRS of patients were moderately strongly and negatively related to those of partners (r=−0.48, p<0.05). At T2, social functioning of patients was moderately strongly and positively related to that of partners (r=0.53, p<0.05). At the three measurement times all other correlations between patients’ SRS and QoL and those of partners were low to moderate, ranging from r=0.04 to 0.34, and not statistically significant. At T1, in two couples, both patient and partner experienced SRS above the cut-off point. At T2 and T3 there were no couples in which both patient and partner experienced SRS above the cut-off point. Wilcoxon tests showed only one significant difference between patients and partners’ functioning. At T2, patients reported a lower level of physical functioning than did partners (Z=−2.6, p<0.01). Independent t tests did not show a significant relationship between having children and SRS for patients at all measurement times, and for partners at T1 and T2. At T3 partners with children reported higher levels of SRS than partners who did not have children (t=2.6, p=0.018). The presence of children was unrelated to patients’ and partners’ QoL at all measurement times. The duration of the relationship was not significantly related to the patients’ or partner’s SRS or QoL at all measurement times and relationships were absent to very weak. The course of SRS and QoL over time in patients and partners Repeated measures ANOVA showed that the SRS of patients fluctuated according to a quadratic trend: The highest level was reported at T1 and after a decrease at T2, the level was increased again somewhat at T3. Earlier levels of SRS in patients were highly positively related to later levels (Table 2). At T1, five of the patients (26%) reported stress response levels above the cut-off point; at T2, two patients (11%); and at T3, three patients (16%). Two patients reported clinically elevated levels of stress response levels at all measurement times. For partners the level of SRS declined, via a linear trend. Partners’ reports of stress response levels were moderately strongly related between T1 and T2, strongly between T2 and T3, and the relationship between T1 and T3 was weak (Table 2). At T1, ten of the partners (53%) reported stress response levels above the cut-off point, at T2, six (32%), and at T3, two (10.5%). One partner reported clinically elevated levels of stress response levels at all measurement times. Table 2SRS and QoL of patients and partners T1T2T3Reference groupRepeated measures ANOVAT1–T2T2–T3T1–T3Mean (SD)Mean (SD)Mean (SD)Mean (SD)FprprprpStress response symptomsPatient18.1 (13.0)10.2 (10.7)12.3 (9.9)7.90.0120.650.0020.710.0010.700.001Partner25.8 (9.3)15.2 (12.6)12.1 (12.4)17.70.0010.41ns0.640.0030.16nsPhysical functioningPatient93.4 (10.8)70.5 (23.4)92.6 (13.3)84.5 (22.3)a,b,c21.40.0010.28ns0.38ns0.20nsPartner90.2 (19.0)90.8 (19.0)89.7 (19.9)80.7 (23.6)b,d0.19ns0.790.0010.800.0010.970.001Social functioningPatient77.6 (19.7)71.1 (22.8)90.8 (13.1)88.4 (19.6)d,e9.00.0080.610.0090.19ns0.17nsPartner72.1 (20.2)77.6 (20.7)f84.5 (16.5)86.1 (20.9)a12.90.0020.600.0070.37ns0.680.001Mental healthPatient68.2 (22.3)77.5 (11.9)81.9 (13.4)79.4 (17.3)d6.70.0190.550.0170.33ns0.23nsPartner69.1 (13.6)71.2 (15.6)74.3 (15.1)75.5 (18.9)2.6ns0.720.0010.640.0040.520.023ap<0.01 (independent t test T1 and reference group)bp<0.05 (independent t test T2 and reference group)cp<0.05 (independent t test T3 and reference group)dp<0.05 (independent t test T1 and reference group)ep<0.01 (independent t test T2 and reference group)fp<0.01 (Wilcoxon test patients and spouses) Lower physical functioning was reported by patients at T2 compared to T1, but physical functioning returned to baseline level at T3. Earlier levels of physical functioning in patients were not significantly related to later levels, and correlations were low. A decline in social functioning of patients was also found at T2 compared to T1, but higher social functioning than at baseline was found at T3. Social functioning of patients at T1 was positively and strongly related to functioning at T2, but the relationships between T2–T3 and T1–T3 were low to moderate. Mental health of patients improved over time. Mental health at T1 was positively and strongly related to levels at T2, and weakly to levels at T3. The relationship between mental health at T2 and T3 was moderately strong. For partners, no significant time effects were found for physical functioning and mental health. Earlier levels of physical functioning and mental health of partners were highly and positively related to later levels. Social functioning of partners improved over the year and was highly positively related between T1 and T2 and between T1 and T3. The relationship between T2 and T3 was moderately strong. (Table 2). QoL of patients and partners compared to that of a reference group of men and women At T1, patients (t=−3.4, p<0.01) and partners (t=−2.1, p<0.05) reported better physical functioning than the reference groups. Patients (t=2.3, p<0.05) and partners (t=2.9, p<0.01) reported worse social functioning than the reference groups at T1, and patients (t=2.2, p<0.05) reported worse mental health at T1. At T2, patients reported worse physical functioning than the reference group of men (t=2.6, p<0.02) but partners reported better physical functioning than the reference group of women (t=−2.2, p<0.05). Patients also reported worse social functioning than the reference men at T2 (t=3.3, p<0.01). No differences were found between the patients’ and partners’ mental health and the norm groups at T2. Only one significant difference was found at T3: Patients reported better physical functioning than men in the reference group (t=−2.6, p<0.02) (Table 2). Discussion The present study was the first to prospectively and longitudinally examine psychosocial functioning in both testicular cancer patients and their partners. We focused on SRS and QoL (physical functioning, social functioning, and mental health) after orchiectomy but before the start of chemotherapy, immediately after completion of chemotherapy, and 9 months later (1 year follow-up). SRS in couples were most salient before the start of chemotherapy. Twenty-six percent of patients reported clinically elevated SRS at this time, a number comparable to the 30% of patients with clinically elevated distress found in a recent study on testicular cancer patients [23]. A review of stress response syndromes in adult cancer populations showed that the incidence of clinically elevated SRS ranged from 3 to 4% in patients recently diagnosed with early stage cancer [34]. The majority of these studies used the same questionnaire for measuring SRS as we did. The same review identified younger age, greater proximity to diagnosis, more advanced disease, and greater treatment intensity as risk factors for a higher level of SRS. The much higher percentage found in our study at the first measurement time may be explained by the prevalence of these four risk factors in our patient group. Not only did patients appear distressed, but also almost twice as many of the partners compared to patients reported clinically elevated levels of SRS. Female sex was found to correlate with greater stress response symptomatology in cancer populations [34]. Of course, these spouses are not patients themselves, but they obviously face cancer-related fears and worries as well. Our results affirm earlier findings that female spouses of cancer patients are vulnerable to distress, and often report higher levels of distress compared to their male counterparts [20, 21, 40–42]. It was also found that partners who had children experienced more SRS a year after diagnosis compared to partners without children. This higher level of distress might also be a result of caregiving tasks that are often the domain of female partners. SRS in couples decreased after completion of chemotherapy. One year after diagnosis, stress response levels in patients rose slightly, but they continued to decrease in partners. These patterns suggest that the period before chemotherapy commences is most stressful, and that recovery seems to occur within a year. At the time of the first measurement the responses of the couples seem to be colored by the fact that they had recently learned the diagnosis, and the outcome of treatment remained uncertain. Concerns of couples about side effects of chemotherapy such as nausea, the possibility of immunodeficiency, and hair loss may contribute to distress. Partners also have caregiving tasks that can exacerbate their distress [22]. All couples in this study received positive news about the results of treatment, and 1 year after diagnosis none of the patients had experienced a relapse of disease. Couples seemed to have recovered from this major life event over the year, possibly as a consequence of the success of treatment reducing insecurity about outcome. An interesting finding was that before commencement of chemotherapy, the level of SRS of patient and partner was inversely related. When one spouse was reporting more SRS, the other reported less. This might be a psychological mechanism through which spouses want to protect one another from their own distress, a finding reported before in studies on couples facing cancer [40]. Another possible interpretation might be that patients who expressed very low levels of distress were using denial, avoidance, or other repressive psychological mechanisms while their partners carried the psychological burden of stress. A patient who expresses high distress may also allow his spouse to assume a stronger supportive role, and thus reduce her expression of stress. In line with the medical trajectory, physical functioning of patients was worst after completion of chemotherapy. As a consequence of chemotherapy, many patients still experience adverse side effects, including fatigue and a sense of physical exhaustion. It is surprising to note that physical functioning of patients was better than that of the reference group before start of chemotherapy and 1 year later. The first measurement occurred shortly after removal of the affected testicle, and patients may have experienced physical relief after the initiation of treatment. It may also be that age contributed to the difference found between the patient group and the reference group. Younger age is associated with better physical functioning [39], and the mean age of the patients in this study was somewhat more than 10 years lower than that of the reference group. There may have been other differences from the reference group as well, such as physical activity and SES. Partners’ physical functioning did not change over the year but they also experienced better physical functioning than the reference group before and after chemotherapy. In an earlier cross-sectional study we found that partners of men who survived testicular cancer between 1–20 years reported better physical functioning than a reference group of women, even years after diagnosis [30]. Partners of testicular cancer patients may have changed the evaluation of their own health in a positive way after witnessing the diagnosis and treatment of a life-threatening illness. In patients, social functioning was worst immediately after completion of chemotherapy, and in spouses before start of chemotherapy. Before start of chemotherapy, both partners reported worse social functioning compared to the reference groups. Before chemotherapy starts, couples are probably overwhelmed by the implications of a cancer diagnosis and focus more on the treatment to come rather than on being socially active as usual. As a consequence of chemotherapy, patients may still suffer from negative side effects like fatigue and impaired physical functioning, which in turn may affect their social functioning. Social functioning in patients and partners was positively related after completion of chemotherapy, meaning that when one partner experienced better social functioning the other also did. Couples had comparable social functioning to that of the reference groups 1 year after diagnosis; they seem to have returned to their usual social activities. This finding is in line with studies on testicular cancer survivors that show little or no change in social contacts and work activities [6, 43]. Mental health of patients improved over the year. It was poorer than that of men in the reference group but only before start of chemotherapy. Mental health of partners was comparable throughout the year and to that of a reference group of women, despite the high level of SRS they reported before start of chemotherapy. SRS apparently are a different expression of mental functioning and may be encapsulated or separated out. It appeared that functioning in couples facing testicular cancer was not similar in patients and partners. Testicular cancer patients showed u-shaped trajectories of SRS, physical functioning, and social functioning. This pattern seemingly followed the medical trajectory they had undergone and was found before in a large group of testicular cancer patients [24]. Mental health of patients improved over the year. Spouses reported a decline in SRS and an improvement in social functioning, but no change in physical functioning or mental health over the year. Differences between patients and partners were also noticeable in stability of functioning. In patients, earlier levels of SRS were strongly predictive of later levels while less individual stability was found with regard to physical and social functioning and to mental health. Partners reported less individual stability in SRS over time, although individual stability was found between T2 and T3. However, partners were individually highly consistent in their reports of physical functioning and somewhat less strong in mental health and social functioning. In addition, correlations between the functioning of patients and partners were moderate to very low. We also found that at the second and third measurement time, there were no couples in which both the patient and the partner reported clinically elevated levels of SRS. This finding confirms the lack of correspondence in functioning between patient and partner. These findings support recent studies that found different adjustment patterns for patients and partners, and a lack of correspondence in functioning [19, 20]. Research is needed to examine if these different reaction patterns to a cancer diagnosis affect the marital relationship. This study has some limitations. First, the possibility of including couples was limited because testicular cancer has a low incidence, and because of the young age of patients at diagnosis, part of this group will not yet have established a steady relationship. Second, no information was available on the functioning of couples who declined to participate. They may have been those who were functioning best or worst, thus biasing the results in either direction. Third, because of the limited sample size, variables that measure more in depth relationship aspects like marital satisfaction were not included in this study. In a previous retrospective study on testicular cancer, factors identified as important for couple adjustment like good communication, spousal support, and marital satisfaction all appeared to facilitate better functioning [32]. These moderating factors in adjustment of both patient and spouse deserve to be studied prospectively in the future. In summary, this study was the first longitudinal prospective exploration of functioning in couples facing testicular cancer during the first year after the diagnosis. Patients confronted with disseminated testicular cancer and their partners reacted differently to this stressor. Clinically elevated stress response levels were present in one third of patients and half of the partners before start of chemotherapy. Patients reported worst mental health before start of chemotherapy, and worst physical and social functioning immediately after completion of chemotherapy. Partners reported an improvement in social functioning, and no change in physical functioning and mental health over the year. QoL of patients and partners was comparable to that of the reference groups a year after start of chemotherapy, patients even reported better physical functioning at that time. These findings support earlier retrospective studies in testicular cancer survivors and their spouses that also reported few long-term effects in psychosocial functioning in the group overall, but that identified a small group that remains distressed [16, 31, 32]. Little correspondence was found in the functioning of the partners. Findings should be regarded as preliminary, as the study sample was limited in size. However, clinicians can be reassured that the effect of disseminated testicular cancer on the QoL of patients and their partners seems to be temporary, although a minority does seem to need clinical attention for SRS.

Knee_Surg_Sports_Traumatol_Arthrosc-4-1-2413121

Effects of biophysical stimulation in patients undergoing arthroscopic reconstruction of anterior cruciate ligament: prospective, randomized and double blind study

Pre-clinical studies have shown that treatment by pulsed electromagnetic fields (PEMFs) can limit the catabolic effects of pro-inflammatory cytokines on articular cartilage and favour the anabolic activity of the chondrocytes. Anterior cruciate ligament (ACL) reconstruction is usually performed by arthroscopic procedure that, even if minimally invasive, may elicit an inflammatory joint reaction detrimental to articular cartilage. In this study the effect of I-ONE PEMFs treatment in patients undergoing ACL reconstruction was investigated. The study end-points were (1) evaluation of patients’ functional recovery by International Knee Documentation Committee (IKDC) Form; (2) use of non-steroidal anti-inflammatory drugs (NSAIDs), necessary to control joint pain and inflammation. The study design was prospective, randomized and double blind. Sixty-nine patients were included in the study at baseline. Follow-up visits were scheduled at 30, 60 and 180 days, followed by 2-year follow-up interview. Patients were evaluated by IKDC Form and were asked to report on the use of NSAIDs. Patients were randomized to active or placebo treatments; active device generated a magnetic field of 1.5 mT at 75 Hz. Patients were instructed to use the stimulator (I-ONE) for 4 h per day for 60 days. All patients underwent ACL reconstruction with use of quadruple hamstrings semitendinosus and gracilis technique. At baseline there were no differences in the IKDC scores between the two groups. At follow-up visits the SF-36 Health Survey score showed a statistically significant faster recovery in the group of patients treated with I-ONE stimulator (P < 0.05). NSAIDs use was less frequent among active patients than controls (P < 0.05). Joint swelling resolution and return to normal range of motion occurred faster in the active treated group (P < 0.05) too. The 2-year follow-up did not shown statistically significant difference between the two groups. Furthermore for longitudinal analysis the generalized linear mixed effects model was applied to calculate the group × time interaction coefficient; this interaction showed a significant difference (P < 0.0001) between the active and placebo groups for all investigated variables: SF-36 Health Survey, IKDC Subjective Knee Evaluation and VAS. Twenty-nine patients (15 in the active group; 14 in the placebo group) underwent both ACL reconstruction and meniscectomy; when they were analysed separately the differences in SF-36 Health Survey scores between the two groups were larger then what observed in the whole study group (P < 0.05). The results of this study show that patient’s functional recovery occurs earlier in the active group. No side effects were observed and the treatment was well tolerated. The use of I-ONE should always be considered after ACL reconstruction, particularly in professional athletes, to shorten the recovery time, to limit joint inflammatory reaction and its catabolic effects on articular cartilage and ultimately for joint preservation. Introduction Articular cartilage performs mechanical functions absorbing the different loads applied to a joint in the course of daily activity [19]. Homeostasis and mechanical competence of cartilage are regulated by the activity of the chondrocytes that maintain the function and the integrity of the extracellular matrix, proteoglycans and collagen. In consideration of the scant repairability of the cartilage, even modest damages resulting from trauma or inflammation may be the starting point for cartilage degeneration leading over time to extensive lesions that deepen into the thickness of the cartilage itself, ultimately exposing the subchondral bone tissue [3, 17]. Joint injury may involve synovial tissue, cartilage and subchondral bone leading to joint inflammation, swelling and pain. Surgical interventions must certainly be included among the triggers of inflammatory reaction in a joint [12]. The development of arthroscopic procedures has undoubtedly limited joint damage associated to surgery for reconstruction of ligaments; nevertheless, it does not avoid the inflammatory response. Thus, while arthroscopic procedures make surgery less invasive, the inflammatory response at the joint remains and the release of pro-inflammatory cytokines in the synovial fluid is associated with an increase in the aggrecanase activities that lead to a degradation of the cartilage matrix, and also inhibit proteoglycan synthesis [11, 15, 18]. To prevent cartilage damage, current pharmacological therapies aim to control the catabolic effects of the pro-inflammatory cytokines and enhance anabolic activity, proteoglycan synthesis and proliferation of chondrocytes. Drugs that combine the above effects are called chondroprotectors; in this category should be included drugs with A2A adenosine receptor agonist activity, able to stimulate the physiological pathways that control inflammation and promote chondrocyte anabolic activities. Nevertheless, these drugs are in early stages of clinical testing [5]. Pre-clinical studies have shown that pulsed electromagnetic fields (PEMFs) in vitro favour the proliferation of chondrocytes [6, 16], stimulate proteoglycan synthesis [7] and demonstrate an A2A adenosine receptor agonist activity [20, 21]. Electromagnetic fields in vivo prevent degeneration of articular cartilage and down-regulate the synthesis and release of pro-inflammatory cytokines in the synovial fluid [2, 4, 8, 9]. These findings suggest that electromagnetic fields may be used to control joint inflammation and to stimulate cartilage anabolic activities, finally resulting in chondroprotection. A clinical study performed in patients undergoing arthroscopic treatment for cartilage lesions showed that biophysical treatment with PEMFs was well tolerated by the patients and led to a decrease in the use of non-steroidal anti-inflammatory drugs (NSAIDs) and to an early functional recovery; the positive effect of the treatment was maintained at a 3-year follow-up [22]. Arthroscopic reconstruction is the treatment of choice following anterior cruciate ligament (ACL) rupture; although minimally invasive, the procedure is associated with joint reaction involving the synovia and it is expected to lead to an increase of pro-inflammatory cytokines in the synovial fluid with catabolic effect on articular cartilage. In this study, we evaluated whether the treatment with PEMFs could be used to control joint inflammatory response in patients undergoing ACL reconstruction. The end points of the study were: (1) patients’ functional recovery evaluated by International Knee Documentation Committee (IKDC) form; (2) use of NSAIDs, necessary to control joint pain and inflammation. Materials and methods Patients and study design In 2004–2005, 84 patients undergoing ACL reconstruction were evaluated for inclusion in the study at five clinical centres. Of these, 69 gave their informed consent to participate in the study. The prospective randomized and double-blind study was approved by the local ethical committees. Inclusion criteria were the following: age between 18 and 45 years, ACL complete lesion following acute trauma or consequence of ligament chronic degeneration. All lesions were documented by MRI and confirmed during the intervention. The following were the exclusion criteria: osteonecrosis of the femoral condyle, rheumatoid arthritis, autoimmune disease, systemic disease and patients requiring meniscus repair. The patients were assigned to the active or placebo group according to the following randomization criteria: age (18–30 or 31–45), sex, smoking status, origin of ACL rupture (traumatic or degenerative). For randomization of patients, a computer-generated schedule was prepared by a biostatistician. In this process, a random number seed was entered into the computer to generate a list that assigned equal numbers of active and placebo stimulators. The minimum number of patients per group required was calculated by power analysis taking into account the results of a previous study [22]. Of the 69 patients included, two never started the therapy, two dropped out within 2 weeks of therapy, and five did not return at follow-up visits; a total of 60 patients were therefore available for subsequent analysis. The ACL rupture occurred during sports activity in 49 patients (24 active and 25 placebo), daily activity in eight patients (four active and four placebo) and traffic accident in three patients (three active). At the time of ACL reconstruction 29 patients underwent also meniscectomy: 15 in the active group and 14 in the placebo. Clinical evaluation The patients were evaluated by IKDC Form before the intervention and at 30, 60 and 180 days afterwards. The different parts of the questionnaire, IKDC Current Health Assessment Form (SF-36 Health Survey), IKDC Subjective Knee Evaluation Form and IKDC Knee Examination Form were analysed separately. As regards the scores of the questionnaires, for each subject we considered the changes at follow-up visits with respect to the values recorded at baseline, before surgery. Pain intensity was evaluated by visual analogue scale (VAS) of 10-cm length: 0 cm no pain, 10 cm maximum pain. The patients were allowed to use NSAIDs to control knee pain when present and had to report doing so. A 2-year follow-up telephone interview was conducted and the patients were asked: (a) if they had undergone further surgery at the knee, (b) if they had pain at the knee, (c) if they had functional limitation in daily activity, (d) if they returned to previous sport activity level. Surgical technique ACL arthroscopic repair was performed by quadruple hamstrings semitendinosus and gracilis technique. Tendons were harvested with the tendons stripper through a 2–3 cm vertical incision on the antero-medial tibial area. Diameter of the quadruple hamstrings semitendinosus and gracilis tendons was measured, while the tibial tunnel and same size femoral tunnel (30 mm length) were prepared. The graft was pulled up through the tibial tunnel with the knee at 90° of flexion and suspended on the external femoral cortex (Endobutton, Smith and Nephew, London, UK). Distally, the graft was fixed with an interference absorbable screw at the tibia at 10° of flexion. Rehabilitation All the patients underwent standard rehabilitation using passive knee flexion daily. Exercises started within the third post-operative day with isometric quadriceps contractions and then progressed to active closed-chain exercises by 4–6 weeks postoperatively. During the first 20 days patients were instructed to use two crutches and then progressive weight bearing until the end of the second month. Biophysical stimulation The patients were treated with active or placebo devices. The active stimulators (I-ONE; IGEA, Carpi, Italy) generated a magnetic field of peak intensity of 1.5 mT at a frequency of 75 Hz; no heat or vibration was felt by the patient during treatment (Fig. 1). Fig. 1Left I-ONE PEMFs generator. Right wave form of magnetic field, 1.5 mT peak value (top); electric field induced in a standard coil probe made of 50 turns (0.5 cm ∅) of copper wire (0.2 mm ∅), peak value 3 mV/cm (bottom) The patients were instructed to use the stimulator for 4 h per day, not necessarily consecutively, for 60 days. Treatment started within 7 days from the surgery. Each device contained a clock to monitor the hours of use. Statistical analysis The results were analysed with SPSS 13.0 (Statistical Packages for Social Sciences Inc, Chicago, IL, USA). Comparison among the continuous variables in the two groups was performed with Student’s heteroschedastic t test; comparison of continuous variables within each group during follow-up was performed with Student’s paired t test. Binomial and categorical variables were compared by contingency tables applying the chi-square test for 2 × 2 tables and the Cochran Mantel Haenszel test for larger size tables. Generalized linear mixed effects model was applied to the SF-36 Health Survey, IKDC Subjective Knee Evaluation and VAS data to test if a different trend between the two groups was present during follow-up by correcting for the following covariates: sex, age, weight, height, hours of treatment, smoking status, use of NSAIDs. In this analysis, a mathematical model is built which takes into account the trend over time of individual patients belonging to each group (Group × Time interaction) and determines if a statistical difference exists between the groups during the follow-up [10]. The minimum significance level for all the statistical tests was set at P < 0.05. Results At baseline, the two groups of study were homogeneous for age, weight, height, VAS, SF-36 Health Survey and IKDC Subjective Knee Evaluation score (Table 1). Table 1Characteristics of the groups at baselinePlacebo #29I-ONE #31PMeanSEMeanSEAge29.61.632.51.40.17Weight7227330.73Height175117420.59VAS2.40.33.20.50.27SF-36 Health Survey3723720.95IKDC Subjective Knee Evaluation Form4834730.90 Average daily treatment was the same in both groups: 3.92 ± 0.5 h/die versus 3.13 ± 0.3 h/die in the I-ONE group and the placebo group, respectively (P = n.s.). The average pain was modest and almost absent at 6 months’ follow-up: 0.7 ± 0.2 cm among placebo and 0.9 ± 0.2 cm among active. At 30 days, less patients in the active group used NSAIDs: 8% in the I-ONE group versus 27% in the placebo group (P < 0.05). The SF-36 Health Survey score decreased significantly at 30 days, in both groups (P < 0.0005). At 60 days the mean SF-36 Health Survey score in the I-ONE patients already exceeded the initial value (by 3.2 points), whereas in the patients of the placebo group SF-36 Health Survey score was slightly below the initial mean value (by −0.7 units). At 6 months a significant (P < 0.005) increase was observed for SF-36 Health Survey average values in both groups; the patients of the I-ONE group were above the initial values by 10.1 units, while the placebo group exceeds the baseline value by 7.2 units. The mean changes of SF-36 Health Survey score in the I-ONE group are systematically higher with respect to placebo during follow-up, P < 0.05 (Fig. 2). Fig. 2Mean changes of SF-36 Health Survey (±SE) versus baseline in the two groups (P < 0.05) The IKDC Subjective Knee Evaluation score increased over 6 months and did not show significant differences between the two groups at any follow-up visit. The IKDC Knee Examination Form outlined both groups including subjects with joint swelling before surgery (one in placebo and two in I-ONE group, P = n.s.) and at 30 days’ follow-up (five in placebo and six in I-ONE group, P = n.s.). On day 60, joint swelling was observed in the placebo group (two patients) only. Joint swelling was not observed any more at 6 months’ follow-up. Limitation in the passive range of motion of the knee was more frequent in the placebo group than in the I-ONE group (P < 0.05) (Fig. 3). Fig. 3Patients with limitation in passive range of motion in the two groups, P < 0.05 Finally, the generalized linear mixed effects analysis revealed a significantly different trend (group × time interaction, P < 0.0001) between the two groups for SF-36 Health Survey score, IKDC Subjective Knee Evaluation score and for VAS, showing a positive effect of I-ONE treatment. The estimate coefficients and significance of independent variables for three models are displayed in Table 2. Table 2Generalized linear mixed effects models in which the dependent variables considered are: SF-36 Health Survey score, IKDC Subjective Knee Evaluation score and VAS, respectivelyCoefficientStd. err.z testP < SF-36 Health Survey  Group0.3612.6900.130.893 Hours of treatment−0.0140.008−1.790.073 Sex−8.3484.810−1.740.083 Weight−0.4230.171−2.480.013 Height0.2200.2540.870.386 Age−0.2120.137−1.550.120 Smoking status0.7682.0750.370.711 Use of NSAIDs−4.6672.726−1.710.087 Time0.1250.0344.580.000 Group × Time0.0510.0143.670.0001 Constant37.94040.3020.940.347IKDC Subjective Knee Evaluation Group−1.4993.417−0.440.661 Hours of treatment−0.0070.011−0.620.533 Sex−19.3906.172−3.140.002 Weight−0.5780.216−2.680.007 Height0.1500.3220.470.642 Age−0.7540.185−4.080.000 Smoking status−2.2142.783−0.800.426 Use of NSAIDs−2.9413.805−0.770.440 Time0.1450.0512.890.000 Group × Time0.1670.0315.370.000 Constant94.56151.3821.840.066VAS  Group0.7990.4551.760.079 Hours of treatment−0.0000.001−0.220.824 Sex1.5830.8061.960.050 Weight0.0200.0290.670.505 Height0.0270.0430.640.524 Age0.0330.0221.50.133 Smoking status0.2460.3370.730.464 Use of NSAIDs−0.8250.434−1.900.058 Time−0.4410.2443.890.000 Group × Time−0.0090.002−4.000.000 Constant−5.5936.721−0.830.405The Group × Time interaction term describes the different trend between the groups At the 2-year follow-up interview 86% of the patients in the I-ONE group and 75% in the placebo group reported complete functional recovery, no knee pain and return to sport activity. ACL reconstruction and meniscectomy When the cohort of patients, undergoing both ACL reconstruction and meniscectomy, was analysed separately, the SF-36 Health Survey score confirmed the faster recovery trend among I-ONE treated patients compared to placebo, P < 0.05 (Fig. 4). At 6 months, SF-36 Health Survey average score increase was 11.4 in the I-ONE group (P < 0.005 vs. baseline) and 7.1 in placebo group (P = ns vs. baseline). Further, the average values of SF-36 Health Survey were significantly higher in the I-ONE group compared to the placebo (45.2 ± 1.5 vs. 37 ± 2.7, P < 0.05). Fig. 4Patients undergoing ACL and meniscectomy: mean changes of SF-36 Health Survey (±SE) versus baseline in the two groups (P < 0.05) The percent of patients with limitation in the passive range of motion was lower in the I-ONE group compared to the placebo one (34% I-ONE vs. 50% placebo at day 30 and 4% I-ONE vs. 17% placebo at day 60, P < 0.05). Discussion Arthroscopic surgery has gained a large success and led to a significant increase in its use: about 650,000 procedures are performed in the USA each year [14]. However, the access into the joint space is always associated to an inflammatory reaction that may jeopardize the benefits expected from surgery. Joint inflammation has a catabolic effect on extracellular matrix and inhibits chondrocyte activity; thus, all means capable of locally controlling the inflammation should be adopted to prevent the onset and limit the progression of cartilage damage. Furthermore, unlike bone tissue after damage, the cartilage will not completely recover its competence: once lost, the articular cartilage does not reform [13]. Many efforts are made to develop strategies able to control joint inflammation and to favour the anabolic activities of chondrocytes; these are challenging objectives, and up to now the pharmacological approaches based on the use of drugs, whether by systemic or by local route, have not yet been able to demonstrate a genuine chondroprotective effect in humans [19]. Pre-clinical studies have shown PEMFs to have a chondroprotective effect, mediated by the control of inflammation and by the stimulation of chondrocyte activity; thus, we hypothesized that after arthroscopic surgery PEMFs treatment can be used for articular cartilage protection and ultimately joint preservation. This prospective, randomized and double-blind study investigated whether and to what extent the employment of I-ONE, by controlling joint reaction to arthroscopy, could accelerate functional recovery in patients undergoing ACL reconstruction. The I-ONE treatment was well tolerated by the patients and no adverse side effects were observed. The results show that, at 30 days after surgery, in I-ONE group significantly fewer patients used NSAIDs to control pain, compared to patients in the placebo group; afterwards, the use of NSAIDs was not necessary in either group. When IKDC Subjective Knee Evaluation average scores were analysed, we found no statistically significant difference between the I-ONE and placebo group; this is in agreement with the findings of other authors who reported that this parameter does not correlate with the other clinical information collected using the SF-36 Health Survey form [1]. However, when the results of the two groups were analysed by generalized linear mixed effects model, which takes into account the trend of each patient in both groups and the effect of confounding factors, we could evidence a positive significant effect of I-ONE treatment also in the Subjective Knee Evaluation (Table 2). The SF-36 Health Survey average scores at baseline were the same in the I-ONE and placebo groups; however, the high standard deviation testify the large distribution of initial score values. To monitor patient’s recovery after ACL reconstruction, we considered the SF-36 Health Survey score changes with respect to baseline for each individual subject. At 2 and 6 months SF-36 Health Survey increase is undoubtedly higher in I-ONE group than in the placebo group. This result indicates a faster recovery in the treated patients. This positive effect of I-ONE treatment is confirmed by the generalized linear mixed effects analysis. Further, when the cohort of patients who underwent both ACL reconstruction and meniscectomy was analysed, we observed that the average increase of SF-36 Health Survey at 60 days in the I-ONE group was the same as that of placebo group at 6 months (6.0 vs. 7.1, P = n.s.). The IKDC Knee Examination Form showed how in the placebo group the resolution of joint swelling and the recovery of complete range of motion occur later compared to the I-ONE group; no significant difference in scoring was observed among centres. The study end-points were thus demonstrated: fewer patients in the I-ONE group required the use of NSAIDs and their functional recovery was faster. At 2-year follow-up no statistically significant difference was observed between two groups, although the percent of patients with complete recovery was slightly higher in the I-ONE group. In this study we applied a statistical analysis specifically developed for longitudinal studies that allows to calculate the group × time interaction. This test, that considers individual patient’s score at different time points and the possible influence of confounding factors, supports the positive effect of I-ONE treatment on the recovery of patients undergoing ACL reconstruction. Our data confirm the results reported by Zorzi et al. [22] in a group of patients treated with I-ONE following an arthroscopic treatment for cartilage lesions. To the authors’ knowledge, there are no other reports of use of biophysical stimulation after surgical procedures of the knee. Biophysical stimulation allows treating individual joints, permeating the whole cartilage surface and thickness, the synovia and the subchondral bone. The effectiveness of biophysical stimulation is not limited by considerations such as diffusion ability and concentration gradient, which are present and important in the dynamic of a pharmacological intervention; joint tissues are paramagnetic, they do not attenuate the biophysical signal and thus are all homogenously exposed to the treatment efficacy. Biophysical stimulation is an effective therapeutic intervention to control the detrimental consequences of the inflammation over articular cartilage in the absence of negative side effects. I-ONE should always be considered after ACL reconstruction, particularly in professional athletes, to shorten the recovery time, to limit joint inflammatory reaction and ultimately for joint preservation.

Int_J_Hematol-4-1-2330061

Identification of genes potentially involved in supporting hematopoietic stem cell activity of stromal cell line MC3T3-G2/PA6

Although coculture of hematopoietic stem cells (HSCs) with stromal cells is a useful system to study hematopoiesis in the niche, little is known regarding the precise cellular and molecular mechanisms of maintaining HSCs through cell–cell interactions. The murine preadipose stromal cell line MC3T3-G2/PA6 (PA6) has been demonstrated to support HSCs in vitro. In this study, microarray analysis was performed on PA6 cells and HSC-nonsupporting PA6 subclone cells to identify genes responsible for supporting HSC activity. Comparison of gene expression profiles revealed that only 144 genes were down-regulated by more than twofold in PA6 subclone cells. Of these down-regulated genes, we selected 11 candidate genes and evaluated for the maintenance of HSC function by overexpressing these genes in PA6 subclone cells. One unknown gene, 1110007F12Rik (also named as Tmem140), which is predicted to encode an integral membrane protein, demonstrated a partial restoration of the defect in HSC-supporting activity. Introduction Hematopoietic stem cells (HSCs) are capable of self-renewal and multilineage differentiation and generate all types of blood cells throughout the lifetime. A major challenge for a long time has been the establishment of culture conditions that can facilitate ex vivo expansion of HSCs. To date, a large number of attempts using combinations of multiple cytokines have resulted in limited success, with a several-fold increase in HSC numbers under the optimal culture conditions [1, 2]. A major problem is that culture conditions established thus far have induced HSC proliferation coupled with differentiation, leading to a progressive loss of in vivo long-term repopulating potential [3–8]. Therefore, a better understanding of mechanisms that regulate HSC self-renewal and differentiation is required to achieve ex vivo expansion of HSCs. In the adult bone marrow, HSCs are thought to reside in a specific microenvironment, referred to as the niche, which is composed of stromal cells that play a vital role in determining stem cell fate. Indeed, several stromal cell lines have been established not only from bone marrow but also from fetal liver and the aorta-gonad-mesonephros region and have been shown to maintain HSCs in vitro [9–16]. Coculture of HSCs with stromal cell lines is a useful system to study hematopoiesis in the niche. Although physical contact between HSCs and stromal cell lines has been demonstrated to be essential, little is known regarding the precise cellular and molecular mechanisms of maintaining HSCs by stromal cell lines. MC3T3-G2/PA6 (PA6) is a preadipose stromal cell line derived from newborn mouse calvaria that can support long-term hematopoiesis in vitro [17]. In addition, PA6 subclones incapable of supporting the proliferation of HSCs have been isolated [18]. It has also been shown that the hematopoiesis-supporting ability of PA6 cells is not solely conferred by their expression of c-kit ligand, stem cell factor (SCF). In this study, microarray analysis was performed to identify genes responsible for supporting HSC activity of PA6 cells. Eleven genes specifically down-regulated in the subclone cells were selected and characterized by overexpressing the corresponding cDNAs in the subclone cells. Materials and methods Mice C57BL/6 (B6-Ly5.2) mice were purchased from Charles River Laboratories Japan (Yokohama, Japan). C57BL/6 mice congenic for the Ly5 locus (B6-Ly5.1) were obtained from RIKEN BRC. B6-Ly5.1/Ly5.2 F1 mice were obtained from mating pairs of B6-Ly5.1 and B6-Ly5.2 mice. All animal experiments were approved by the Animal Experiment Committee of the RIKEN Tsukuba Institute. Stromal cell culture and lentiviral transduction The OP9 stromal cell line was obtained from RIKEN BRC Cell Bank. PA6 and OP9 cells were maintained in Minimum Essential Medium Eagle-α (MEM-α) (Sigma-Aldrich, St Louis, MO) containing 20% fetal bovine serum (FBS) (Sigma-Aldrich) at 37°C in a 5% CO2 atmosphere. FANTOM cDNA clones (kindly provided by Dr. Y. Hayashizaki and Dr. J. Kawai, RIKEN GSC) corresponding to the genes used in this study were subcloned into the lentiviral vector plasmid pCSII-EF-MCS-IRES2-Venus. Recombinant lentiviral vectors were produced as described previously [19]. PA6 subclone cells were transduced with lentiviral vectors expressing cDNAs at a multiplicity of infection of 200 and >90% of transduction efficiency was confirmed by fluorescence-activated cell sorting (FACS) analysis for Venus expression. Purification of CD34−KSL cells and coculture with stromal cells CD34−/lowc-Kit+Sca-1+lineage marker− (CD34−KSL) cells were purified as described previously with minor modifications [20]. Briefly, bone marrow cells isolated from 10- to 16-week-old B6-Ly5.2 mice were stained with a lineage marker antibody cocktail consisting of biotinylated anti-Gr-1, anti-Mac-1, anti-B220, anti-IgM, anti-CD4, anti-CD8, and anti-Ter119 antibodies (eBioscience, San Diego, CA). Lineage marker+ cells were depleted using streptavidin-coupled Dynabeads M-280 (Invitrogen, Carlsbad, CA). The remaining cells were stained with fluorescein isothiocyanate (FITC)-conjugated anti-CD34, phycoerythrin (PE)-conjugated anti-Sca-1, and allophycocyanin (APC)-conjugated anti-c-Kit antibodies (all from BD Biosciences, San Jose, CA). The biotinylated antibodies were developed with streptavidin-APC-Cy7 (BD Biosciences). FACS was performed with a FACSVantage SE (BD Biosciences). CD34−KSL cells were sorted into individual wells of a 96-well plate containing stromal cells (1 × 104 cells/well) irradiated with 1.5 Gy and were cocultured in 150 μL of MEM-α containing 20% FBS. Colony-forming cell assay After 10 days of coculture, CD34−KSL HSCs were collected and plated in a 12-well plate containing 0.6 mL of methylcellulose medium (MethoCult GF M3434) (StemCell Technologies, Vancouver, Canada) containing 50 ng/mL rmSCF, 10 ng/mL rmIL-3, 10 ng/mL rhIL-6, and 3 units/mL rhEPO. After 12 days of incubation, colonies were recovered, cytospun onto glass slides, and then subjected to Hemacolor (Merck KGaA, Darmstadt, Germany) staining for morphological examination. Colony-forming unit-granulocyte, erythrocyte, monocyte, megakaryocyte (CFU-GEMM), CFU-granulocyte, erythrocyte, monocyte (CFU-GEM), CFU-granulocyte, monocyte (CFU-GM), CFU-granulocyte (CFU-G), CFU-monocyte (CFU-M), and burst forming unit-erythrocyte (BFU-E) were scored using standard scoring criteria. The total colony number was expressed as colony-forming cell (CFC). Competitive repopulation assay The competitive repopulation assay was performed by using the congenic Ly5 mouse system as described previously [8]. CD34−KSL HSCs after coculture were mixed with 2 × 105 total bone marrow competitor cells from B6-Ly5.1 mice and transplanted into lethally (9.5 Gy) irradiated B6-Ly5.1 mice. At 12 to 16 weeks after transplantation, peripheral blood cells of the recipient mice were collected by retro-orbital bleeding. After lysis of red blood cells with ammonium chloride buffer, the remaining nucleated cells were stained with FITC-conjugated anti-Ly5.2, PE-conjugated anti-Ly5.1, biotinylated anti-Mac1, and biotinylated anti-Gr1 antibodies, followed by addition of streptavidin-PerCP. The cells were stained simultaneously with APC-conjugated anti-B220 antibody or a mixture of APC-conjugated anti-CD4 and anti-CD8 antibodies. FACS analysis was performed with a FACSCalibur. Donor chimerism was determined as the percentage of Ly5.2+ cells. When the percent chimerism was >1.0% for all myeloid, B-lymphoid, and T-lymphoid lineages, recipient mice were considered to be multilineage reconstituted. Microarray analysis Total RNA was isolated from stromal cells using the ISOGEN reagent (Nippon Gene, Tokyo, Japan) and purified with the RNeasy MinElute cleanup kit (Qiagen, Hilden, Germany). Biotin-labeled cRNA was prepared from 1 μg of the purified total RNA with a one-cycle cDNA synthesis kit and 3′-amplification reagents for IVT labeling (Affymetrix, Santa Clara, CA) and was hybridized to an Affymetrix Gene Chip Mouse Genome 430 2.0 array (Affymetrix), which contains approximately 45,000 probe sets for analyzing the expression levels of more than 34,000 mouse genes. After washing and staining with the antibody amplification procedure, the microarrays were scanned with an Affymetrix GeneChip Scanner 3000 7G. All these procedures were carried out according to the manufacturer’s instructions. The expression value (Signal) and detection call (Present (P), Absent (A), or Marginal (M)) for each probe set were calculated using GeneChip Operating Software version 1.4 (Affymetrix). The Signal values were normalized so that their mean in each experiment was 100 in order to adjust for minor differences between the experiments. The change value (Signal Log Ratio) and change call (Increase, Marginal Increase, No Change, Marginal Decrease, or Decrease) for each probe set were calculated by Comparison Analysis of the software. All experiments were performed in duplicate using two independent cell samples. To identify differentially expressed genes, we selected probe sets that showed a change call of Decrease and a Signal Log Ratio value of ≤–1 (more than twofold down-regulation) or a change call of Increase and a Signal Log Ratio value of ≥1 (more than twofold up-regulation) in each of the two independent experiments. Quantitative real-time polymerase chain reaction Total RNA was isolated from stromal cells using the ISOGEN reagent, and cDNA was synthesized from 1 μg of total RNA using SuperScript II reverse transcriptase (Invitrogen). Real-time polymerase chain reaction (PCR) amplification with two independent cell samples was carried out using the LightCycler FastStart DNA Master SYBR Green I kit (Roche, Penzberg, Germany) according to the manufacturer’s instructions. The primer sets used in this study are listed in Table 1. Data were normalized GAPDH mRNA levels. Gene-specific amplification was confirmed by determining the melting curves of the PCR products and by a single band of the expected size in agarose gel electrophoresis. Table 1Primers used in quantitative real-time PCRGenePrimer sequenceProduct size (bp)1110007F12RikForward5′-GCCCTGTGCCTGATGTTCTAC-3′111 Reverse5′-GCCCATGTCCTCCTTCCAC-3′2900064A13RikForward5′-GTTTGACCCTGTCCGAGTCG-3′205 Reverse5′-CGGGAGAACCATCATCATAACC-3′Ccl2Forward5′-TTAAAAACCTGGATCGGAACCAA-3′121 Reverse5′-GCATTAGCTTCAGATTTACGGGT-3′Ccl9Forward5′-TCAGATTGCTGCCTGTCCTAT-3′117 Reverse5′-GAACCCCCTCTTGCTGATAAAG-3′Cxcl5Forward5′-TGCGTTGTGTTTGCTTAACCG-3′107 Reverse5′-AGCTATGACTTCCACCGTAGG-3′IL-1rnForward5′-GCTCATTGCTGGGTACTTACAA-3′132 Reverse5′-CCAGACTTGGCACAAGACAGG-3′IL-6Forward5′-TAGTCCTTCCTACCCCAATTTCC-3′76 Reverse5′-TTGGTCCTTAGCCACTCCTTC-3′ Results and discussion Subclones of the PA6 cell line are defective in supporting HSCs Although PA6 subclones 2, 12, and 14 have been isolated in terms of inability to support long-term hematopoiesis in vitro, their HSC-supporting capacity has not been precisely assessed [18]. Therefore, we evaluated the maintenance of HSC function after coculture with PA6 subclones 2 and 12 (hereinafter referred to as S-2 and S-12, respectively) by in vitro CFC and in vivo competitive repopulation assays using CD34−KSL cells as highly purified HSCs [20]. Murine bone marrow-derived OP9 stromal cell line that can support HSCs was also used as a control [13]. As shown in Table 2, the CFC frequency significantly decreased after 10 days of coculture with PA6 S-2 and S-12 cells compared with PA6 cells, and lower levels of engraftment were observed in transplanted mice. These results indicate that PA6 subclone cells have a substantial defect in supporting HSCs even after short-term coculture. Table 2CFC frequency and competitive repopulation capacityStromal cellsCFC frequencyNo. of reconstituted mice (%)% ChimerismPA610.8 ± 7.516/9 (66.7)13.2 ± 19.9PA6 S-20.60 ± 0.97*4/10 (40.0)4.5 ± 11.2**PA6 S-123.2 ± 3.05*3/8 (37.5)3.4 ± 5.9**OP928.2 ± 6.615/9 (55.6)8.6 ± 11.5A total of 80 CD34−KSL HSCs (10 cells/well) were cocultured with stromal cells for 10 days and were then subjected to the CFC assay. The CFC frequency represents the colony number per ten input CD34−KSL HSCs. For the competitive repopulation assay, 30 CD34−KSL HSCs were cocultured with stromal cells for 10 days and then transplanted into lethally irradiated mice. At 12 weeks after transplantation, peripheral blood cells of the recipient mice were analyzed. Data represent the mean ± SD of two independent experiments*P < 0.01, **P > 0.2 versus PA6, Student’s t-test Identification of genes that were specifically down-regulated in PA6 subclone cells We hypothesized that genes responsible for supporting HSCs would be down-regulated in PA6 subclone cells. To identify genes specifically down-regulated in PA6 subclone cells, microarray analysis was performed on PA6 cells, PA6 S-2 and S-12 cells, and OP9 cells using an Affymetrix GeneChip array containing approximately 45,000 probe sets representing over 34,000 mouse genes. As expected, gene expression profiling showed a few differences in gene expression between PA6 cells and PA6 subclone cells. Compared to PA6 cells, only 144 genes were down-regulated by more than twofold in both PA6 subclone cells. Of these down-regulated genes, 41 genes were of unknown function, and 67 of 103 (65%) known genes encoded membrane and extracellular space proteins according to the Gene Ontology cellular component classification. On the other hand, 121 genes were up-regulated by more than twofold and 30 of 67 (45%) known genes encoded membrane and extracellular space proteins. The signal values and detection calls for this analysis are provided in supplemental data. Among the down-regulated genes, eight genes of known/predicted function including chemokine- and cytokine-related genes, which were expected to effect on the HSC-supporting activity, and three genes of unknown function, which were predicted to encode membrane proteins, were selected for further analysis (Table 3). Table 3List of candidate genes down-regulated in PA6 subclone cells compared with PA6 cellsGene symbolGene titleRefseq IDMicroarrayqPCRvs. PA6vs. OP9vs. PA6vs. OP9PA6 S-2PA6 S-12PA6 S-2PA6 S-12PA6 S-2PA6 S-12PA6 S-2PA6 S-121110007F12RikRIKEN cDNA 1110007F12 geneNM_1979860.300.460.110.190.290.230.160.131200009O22RikRIKEN cDNA 1200009O22 geneNM_0258170.300.212.791.48NDNDNDND2900064A13RikRIKEN cDNA 2900064A13 geneNM_1337490.410.371.231.160.290.330.480.55Ccl2Chemokine (C-C motif) ligand 2NM_0113330.090.130.040.060.160.070.090.04Ccl9Chemokine (C-C motif) ligand 9NM_0113380.140.190.030.040.100.090.040.03Cd53CD53 antigenNM_0076510.230.150.280.14NDNDNDNDCxcl5Chemokine (C-X-C motif) ligand 5NM_0091410.330.230.430.380.200.130.900.60HgfHepatocyte growth factorNM_0104270.410.410.850.88NDNDNDNDIl1rn(IL-1rn)Interleukin 1 receptor antagonistNM_0311670.090.120.260.31<0.030.04<1.251.45Il6(IL-6)Interleukin 6NM_0311680.380.442.072.620.300.162.421.31Ppap2bPhosphatidic acid phosphatase type 2BNM_0805550.220.220.090.09NDNDNDNDRelative expression levels in PA6 subclone cells (S-2 and S-14) compared with PA6 and OP9 cells were calculated using data from microarray and quantitative real-time PCR (qPCR) analyses. Data represent the average of fold changes from two independent experimentsND not determined Characterization of candidate genes To determine the requirement of candidate genes for the supportive activity of PA6 cells, PA6 S-2 cells were transduced with lentiviral vectors containing the corresponding cDNAs. In all experiments, the transduction efficiency was more than 90% as determined by FACS analysis for Venus-positive cells. CD34−KSL HSCs were cocultured with PA6 S-2 cells expressing candidate genes for 10 days and then subjected to the CFC assay. As shown in Fig. 1, the CFC frequency increased when cocultured with PA6 S-2 cells overexpressing the candidate genes except Hgf compared with PA6 S-2 cells. Morphological evaluation of the colonies revealed that the number of CFU-GEMM, which is the most primitive colony type observed in the CFC assay, increased except Ccl2 and Ccl9. Of note, although the overexpression of IL-6 increased the number of colonies to the same level as PA6 cells, a large proportion (∼80%) of colonies were CFU-M. Fig. 1Effect of over expression of candidate genes on the CFC frequency. A total of 80 CD34−KSL HSCs (10 cells/well) were cocultured with PA6 cells, PA6 S-2 cells, or PA6 S-2 cells expressing the indicated genes for 10 days and then subjected to the CFC assay. Individual colonies were scored according to their morphology. Data represent the average of two independent experiments We next performed the competitive repopulation assay to determine the effect of overexpressing seven candidate genes including Ccl9, IL-6, Ppap2b, and Cxcl5, and three unknown genes. As shown in Table 4, successful multilineage reconstitution was observed in all mice transplanted with HSCs cocultured with PA6 S-2 cells overexpressing 1200009O22Rik or IL-6 as well as with PA6 cells. However, comparison of the chimerism showed no significant differences from that of PA6 S-2 cells, suggesting that the overexpression of neither 1200009O22Rik nor IL-6 completely restored the defective HSC-supporting activity of PA6 S-2 cells. Microarray analysis showed that the expression of IL-6 was approximately twofold up-regulated in PA6 S-2 and S-12 cells when compared with OP9 cells; this finding was confirmed by quantitative real-time PCR analysis (Table 3). It has also been demonstrated that IL-6 alone or in combination with SCF is incapable of maintaining CD34−KSL HSCs [8]. Although IL-6 is involved in various steps of hematopoiesis and has been used for the ex vivo expansion of HSCs [21], IL-6 may not be crucial for the maintenance of HSCs governed by stromal cells. Table 4Effect of overexpression of candidate genes on the competitive repopulation capacityNo. of reconstituted mice (%)% ChimerismTotalMyeloidB-lymphoidT-lymphoidPA69/9 (100)22.1 ± 25.130.9 ± 30.925.1 ± 26.517.4 ± 23.0PA6 S-24/10 (40)7.3 ± 14.04.0 ± 7.410.8 ± 18.89.2 ± 17.81110007F12Rik7/10 (70)16.2 ± 14.8*21.8 ± 26.318.7 ± 16.915.5 ± 16.71200009O22Rik8/8 (100)8.2 ± 10.214.5 ± 12.29.6 ± 13.48.3 ± 11.32900064A13Rik2/5 (40)3.3 ± 4.610.5 ± 20.63.6 ± 5.41.0 ± 2.2Ccl92/5 (40)11.1 ± 15.014.2 ± 19.512.7 ± 17.611.4 ± 15.7IL-69/9 (100)8.5 ± 13.58.1 ± 7.111.6 ± 16.99.2 ± 16.3Ppap2b8/10 (80)9.7 ± 10.420.5 ± 26.79.2 ± 9.79.6 ± 14.6Cxcl53/5 (60)3.0 ± 4.012.8 ± 23.62.5 ± 3.51.9 ± 1.8Thirty CD34−KSL HSCs were cocultured with PA6 cells, PA6 S-2 cells, or PA6 S-2 cells expressing the indicated genes for 10 days and then transplanted into lethally irradiated mice. At 16 weeks after transplantation, peripheral blood cells of the recipient mice were analyzed. Data represent the mean ± SD*P = 0.18 versus PA6 S-2, Student’s t-test PA6 S-2 cells overexpressing 2900064A13Rik or CCl9 had no apparent effect on the HSC-supporting activity. In contrast, the overexpression of 1110007F12Rik, Ppap2b, or Cxcl5 resulted in a certain increase in the frequency of reconstituted mice. Although these genes as well as 1200009O22Rik and IL-6 could significantly increase the CFC frequency, the effect on the HSC-supporting activity assessed by the competitive repopulation assay was limited. This may be because 10 days of coculture was not long enough to detect a significant difference in HSC-supporting activity. It is of course well known that many growth factors and cytokines that can increase the CFC frequency have little influence on the maintenance and proliferation of HSCs. Of note, the overexpression of 1110007F12Rik showed a substantial increase in the chimerism, although the difference did not reach statistical significance (Table 4). To confirm the effect of 1110007F12Rik expression on the supportive activity of PA6 cells, we performed the following experiments. First, down-regulation of 1110007F12Rik expression in PA6 S-2 and S-12 cells compared to that in PA6 and OP9 cells was confirmed by quantitative real-time PCR analysis (Table 3). Next, PA6 S-2 cells were transduced with a myc-tagged 1110007F12Rik expression lentiviral vector and its expression was detected by Western blotting and immunofluorescence staining with anti-myc antibody. Coculture with PA6 S-2 cells overexpressing myc-tagged 1110007F12Rik resulted in an increase in the CFC frequency at a level similar to that shown in Fig. 1 (data not shown). The 1110007F12Rik cDNA encodes a predicted protein of 185 amino acids, which is rich in leucine (23%) and has no significant homology with any known protein sequences. 1110007F12Rik is also named as Tmem140 (transmembrane protein 140), which is predicted to be an integral membrane protein gene according to the Gene Ontology classification. Further studies are required to determine whether 1110007F12Rik is involved in the HSC-supporting activity of PA6 cells. Thus far, a large number of microarray analyses of stromal cell lines derived from a variety of tissue types have been reported. Comparison of gene expression profiles between HSC-supporting and HSC-nonsupporting stromal cell lines has resulted in the identification of a number of differentially expressed genes [22]. In the present study, we expected a few differences in gene expression by comparing HSC-supporting PA6 cells with HSC-nonsupporting PA6 subclone cells and indeed, we could narrow down the candidate genes involved in maintaining HSCs. Although one of the analyzed candidate genes, 1110007F12Rik, demonstrated a partial restoration of the defect in HSC-supporting activity of PA6 S-2 cells, it is possible that other down-regulated genes in PA6 subclone cells may also be required for the maintenance of HSCs. Continuing studies to identify genes responsible for HSC-supporting activity of PA6 cells would facilitate the understanding of the mechanisms that control stem cell fate through interaction with stromal cells. Electronic supplementary material Below is the link to the Supplemental data. Electronic supplementary material (XLS 388 kb)

Breast_Cancer_Res_Treat-3-1-2137942

Stromal mast cells in invasive breast cancer are a marker of favourable prognosis: a study of 4,444 cases

Purpose We have previously demonstrated in a pilot study of 348 invasive breast cancers that mast cell (MC) infiltrates within primary breast cancers are associated with a good prognosis. Our aim was to verify this finding in a larger cohort of invasive breast cancer patients and examine the relationship between the presence of MCs and other clinical and pathological features. Introduction Mast cells (MCs) are part of the innate immune system and are recruited to and activated in the microenvironment of a developing tumor. MCs originate from multipotential hemopoietic stem cells in the bone marrow, and express the proto-oncogene c-kit, a transmembrane type III tyrosine kinase receptor protein [1]. After leaving the vascular bed they differentiate and acquire functional maturity. MC infiltrates have been described in a variety of human cancers, including non-small-cell lung cancer [2–4], breast cancer [5], colorectal cancer [6], basal cell carcinoma [7] and pulmonary adenocarcinoma [8]. MCs are attracted to the tumor by tumor-derived chemo attractants where they either degranulate to release potential tumor cytotoxic compounds or become innocent bystanders depending on local tumor conditions [9]. There is controversy about the pro- and anti-tumorigenic effects of MCs in different cancers. The accumulation of MCs has been associated with enhanced growth and invasion of several human cancers [10]. On the other hand, MC infiltration has been associated with good prognosis in breast [11, 12] ovarian [13], lung [14] and colorectal [15] carcinomas. Murine models have shown that tumor incidence and growth are inversely correlated with the MC density in MC-deficient mice compared to normal [16]. Our previously published pilot report of 348 case series of breast carcinomas showed that the presence of stromal MCs correlated with a good prognosis (P = 0.0036) in invasive breast cancer [11].We present here a large tissue microarray (TMA) study of 4,444 cases of invasive breast carcinomas with clinical outcome data. This study validates the independent prognostic significance of stromal MCs in breast cancer. Materials and methods Patient selection A total of 4,620 archival samples from patients with invasive breast carcinoma referred to the British Columbia Cancer Agency between January 1986 and September 1992 were used for TMA construction. The Clinical Research Ethics Board of the University of British Columbia approved the study. Patients with in-situ disease, metastatic disease at presentation, and male breast cancer were excluded from analysis, thus bringing the final tally to 4,444 cases. This represents 34% of all patients diagnosed with breast cancer in the province of British Columbia during this time period. This large, well characterized cohort is derived from a consecutive series of patients who were referred to the BC Cancer Agency for consultation and had tumor samples sent to a central laboratory at the Vancouver General Hospital for estrogen receptor (ER) status. Consequently, for all of these patients we have available detailed demographic and outcome data, as well as formalin fixed paraffin embedded (FFPE) primary tumor samples for immunohistochemical analysis. Available clinical information includes age, histology, tumor grade, tumor size, lymph node status, type of local and adjuvant systemic therapy, and dates of first recurrence and death. Clinical and pathological variables were determined following well-established criteria. A portion of this cohort of patients was recently used in a population study validating the on-line breast cancer prognostic calculator ADJUVANT! Online [17]. TMA construction The Vancouver Hospital ER laboratory retained single archival tumor blocks from each case in this patient cohort. The material had been frozen prior to neutral buffered formalin fixation. All the paraffin sections were first stained with H&E and reviewed by a pathologist. Representative tumor areas were circled and matched with the donor blocks. 0.6-mm cores were punched out from the donor blocks and embedded 1 mm apart in 17 recipient blocks using Tissue Microarrayer (Beecher instruments Silver Springs, MD). These tissue arrays have been used to test the new ER SP1 antibody that improves the sensitivity for detecting ER by IHC [18]. Immunohistochemistry (IHC) Sections from TMA were cut at 4 μm and immunostained with antibody to KIT (CD117), a transmembrane tyrosine kinase acting as a type III receptor for MC growth factor. Slides were also concurrently stained for ER and Her2 using standard immunoperoxidase techniques. The antibodies and antigen retrieval methods are summarized in Table 1. Table 1Details of antibodies used for immunohistochemistryAntibodyIsotype CompanyAntigen retrievalConcentrationc-kitRabbit polyclonalDakoVentana1:100Her2 SP3Rabbit monoclonalLab visionSteam 30 min, 0.05 M TRIS buffer (pH 10)1:100ER SP1Rabbit monoclonalLab visionCitrate buffer (pH 6)1:250 Immunohistochemical image processing and scoring The stained slides were digitally scanned with a BLISS automated digital imaging microscope (Bacus Laboratories, Lombard, IL) which consists of a microscope with a scanning stage, video camera and software designed for scanning TMAs. A relational database was constructed using identification information and immunohistochemistry scores for each tissue core in the microarrays. An internet website was then constructed using this database and a WebSlideViewer Java applet provided by the manufacturer to view the microarray images and allow for an image zooming functionality. This website is publicly accessible through http://www.gpecimage.ubc.ca/tma/web/viewer.php. The slides were scored manually by two independent pathologists, blinded to the clinical outcome, as previously described [11]. Total number of stained MCs in each core was recorded. ER status was assessed using the rabbit monoclonal antibody (SP1) antibody [18]. The fractions of ER positive tumor nuclei were scored as 0 (<1%), 1 (1–25%), 2 (25–75%), and 3 (>75%). Her2 SP3 rabbit monoclonal from LabVision (NeoMarkers) was used to stain the TMA slides and were scored using Hercept test® (Dako Corporation, Carpinteria, CA) scoring system. Final Her2 score was derived using both IHC and Fluorescent in situ hybridization (FISH) assays. Cases with Her2 IHC Herceptest score = 3, were scored as positive. Those cases with Her2 IHC Herceptest score = 2 were re-evaluated using FISH assay, and only those cases with Her2 FISH amplification ratio ≥2.0 were scored as Her2 positive. C-Kit and ER scores were binarized for statistical analysis as follows: CD-117: 0 = no MCs; 1 = any MCs; ER: 0 ≤ 1% nuclei stained; 1 ≥ 1% nuclei stained. We excluded cases for which it was not possible to assign a score to the immunostaining (insufficient invasive tumor in the core, or missing core). Statistical analysis The raw scores were entered into an Excel database and each TMA core was assigned a unique core ID number. The spreadsheet was then processed utilizing TMA-Deconvoluter 1.06 software that had been adapted for TMA analysis [19]. A database was created by incorporation of all deconvoluted marker data into the clinicopathologic patient database. We applied a split-sample validation technique for our statistical analysis. Our 4,444 patient cohort was stratified into eight subgroups (Table 2) based on adjuvant treatment received and then randomized into equal sized training and validation sets. The two groups were balanced with respect to treatment received; but there are also no significant differences in clinical or pathological variables, including age, tumor size and grade, nodal status, and ER status. For this study, and future studies using this TMA, the primary investigator is given access to all clinical, outcome, and TMA data from the training set only. The training set is used to generate and refine hypotheses regarding the biomarker under study. Significant findings are then formally presented at a bimonthly joint scientific group meeting of the Genetic Pathology Evaluation Centre (GPEC, a collaborative group of scientists and pathologists) and Breast Cancer Outcomes Unit (BCOU, a group of oncologists and epidemiologists based at the BC Cancer Agency). Those findings considered to be of clinical and scientific interest are then re-tested on the validation set. A separate researcher who did not participate in the training set analysis performs the re-testing on the validation set. Our statistical approach is intended to minimize false positive results, particularly with subgroup analysis. Table 2Summary of clinical-pathological characteristics of the 4,444 breast cancer patientsTest setValidation setn% within group% within known valuesn% within group% within known valuesTotal2,2222,222Age at diagnosis (years)Median (range)60 (25–95)60 (23–91)<401707.77.71577.17.140–5445720.620.646821.121.155–6970231.631.680936.436.4≥7089340.240.278835.535.5GenderFemale2,21199.599.52,21099.599.5Male110.50.5100.50.5Nodal statusNegative1,27257.257.41,25656.556.8Positive94342.442.695743.143.2Unknown70.390.4Number of positive nodesMedian (range)2 (1–24)2 (1–28)1–358662.164.361464.266.24–924325.826.722923.924.7≥10828.79.0848.89.1Unknown323.4303.1ER Status at diagnosisNegative47221.221.847221.221.9Positive1,68976.078.21,68775.978.1Unknown612.7632.8Tumor size (cm)Median (range)2.0 (0.1–9.9)2.0 (0.1–9.9)0.1–1.026511.912.128512.813.01.1–2.086438.939.486438.939.52.1–5.094942.743.291541.241.8>5.01175.35.31265.75.8Unknown271.2321.4Tumor grade11255.66.01034.64.9287639.442.084438.040.231,08348.752.01,15351.954.9Unknown1386.21225.5 Statistical analysis was performed using SPSS 14.0. In univariate analysis, breast-cancer specific survival (BCSS) was estimated using Kaplan–Meier (KM) curves, and significant differences determined by Log rank tests. For BCSS, survival time was censored at the date of death if the cause was not breast cancer or at the date of last follow-up if the patient was still alive at the end of the study period. 6 patients with unknown cause of death were excluded from BCSS analysis. Cox proportional hazards models were used to calculate adjusted hazard ratios accounting for covariates. Kendall’s tau-b and the Mann-Whitney tests were used to measure the correlation of c-kit status to pathological variables. All statistical tests were two-sided. The Bonferroni correction for multiple comparisons was applied during validation set analysis, and the alpha value for each comparison is 0.05/n, where n is the total number of comparisons. X-tile analysis We also used X-tile software [20] to find the optimal cut-off point for the total number of MCs that will predict prognosis in breast cancer patients. X-tile program split the cohort randomly into a matched training and validation set as a method for selecting optimal cut-points. It than calculated a P value for every possible division of the cohort expression data. A two-dimensional graph with its corresponding survival curves was plotted where each colored pixel was proportional to is χ2 Value. The program automatically calculated the maximum χ2 value which served as a cut-point to separate the number of MCs that predicted prognosis. Results Out of 4,620 cases on the TMAs, we selected 4,444 breast cancers that showed invasive tumor in the cores. The clinico-pathologic characteristics of patients included in the study are depicted in Table 2. The total number of stained MCs was recorded as a continuous variable with counts ranging from 0 to 24 MCs per core. They were seen as 4–20 μm round to oval mononuclear cells with granular cytoplasm and single oval nucleus. The cytoplasmic granules were ganglion-, net-, or crystal-shaped (Fig. 1). Fig. 1TMA core showing stromal mast cells stained with c-kit (CD-117). Magnification, 20×. MCs are seen as brown, granular stained oval, spindle or polygonal cells Training set results Survival analysis A total of 2,222 patients were included in the training set analysis. After excluding cases which had insufficient invasive tumor, missing core or un-interpretable staining pattern, 1,801 cases were carried forward for the analysis. Out of these, MCs were present in 508 (28.2%) cases. The mean survival time of patients with presence of stromal MCs was 15.0 years (95% CI, 14.5–15.5) compared to 13.9 years (95% CI, 13.5–14.2) for those who did not have positively stained MCs in their tumor stroma. KM survival analysis (Fig. 2a) showed that the presence of stromal MCs was a favourable prognostic marker in the entire training set (BCSS @ 18.4 years, Log rank [Mantel Cox], P = 0.001). Fig. 2KM survival curve for all patients in training set (a) and validation set (b) with presence of stromal mast cells Correlation with other biomarkers There was positive correlation between MCs and ER (Kendall’s tau-b [τb], 0.034, P = 0.148), Bcl2 (τb = 0.077, P = 0.002), and Her2 (τb = 0.049, P = 0.052), and negative correlation between MCs and EGFR (τb = −0.029, P = 0.228) and CK5/6 (τb = −0.003, P = 0.906) in the training set analysis (Table 3). As these correlations were either not significant or extremely weak, they were not carried forward to the validation set for further analysis. Table 3Correlations between mast cells and other biomarkersTraining setMast cellsERKendall’s tau-b0.034Significance (2-tailed)0.148N1,788EGFRKendall’s tau-b−0.029Significance (2-tailed)0.228N1,646Her2Kendall’s tau-b0.049Significance (2-tailed)0.052N1,746CK5/6Kendall’s tau-b−0.003Significance (2-tailed)0.906N1,624Bcl2Kendall’s tau-b0.077Significance (2-tailed)0.002N1,616N—scorable for both markersAll scores binarized as detailed in the text Nodal status KM survival analysis showed no statistically significant difference in the survival between tumors with and without MCs in node-negative (BCSS @ 18.1 years, Log rank [Mantel Cox], P = 0.1199) and a significant difference in the node-positive group (BCSS @ 18.3 years, Log rank [Mantel Cox], P = 0.0140). Hence, this result was also not carried forward to the validation set. Multivariate analysis Cox proportional hazard model was used to carry out the multivariate analysis and included age, tumor grade, tumor size, nodal status, ER and Her2 as independent predictors of BCSS. All the above variables achieved statistical significance as shown in Table 4(a). Presence of MCs achieved statistical significance (P = 0.041) with a HR = 0.804, 95% CI 0.653–0.991. Table 4Cox proportional hazard regression analysis showing hazard ratios and P-values in patients with invasive breast carcinomaSignificanceHRBCSS 95% CI for HRLowerUpper(a) Training setMast cells0.0410.8040.6530.991ER0.0180.7770.6310.957Her20.0031.4391.1291.834Size of the lesion    2–5 cm vs. ≤2 cm0.0001.8841.5472.295    >5 cm vs. ≤2 cm0.0011.9311.3302.803Grade*    Grade 1, 2 vs. Grade 3 0.0011.3901.1391.696Nodal status0.0002.3801.9652.882Age    40–49 vs. <400.0000.5670.4200.765    50–65 vs. <400.0130.6890.5130.924    >65 vs. <400.0190.6770.4890.938(b) Validation setMast cells0.1280.8460.6831.049ER0.02890.7930 .6440 .976Her20.00181.4631.1521.859Size of the lesion    2−5 cm vs. ≤2 cm3.07 × 10−41.4391.1811.753    >5 cm vs. ≤2 cm1.66 × 10−62.2581.6183.151Grade*    Grade 1, 2 vs. Grade 3 1.96 × 10−61.6581.3462.042Nodal status7.06 × 10−192.4101.9842.927Age    40–49 vs. <400.4321.1580.8031.671    50–65 vs. <400.6341.0900.7651.552    >65 vs. <400.071.3940.9731.998* Grade 1—well differentiatedGrade 2—moderately differentiatedGrade 3—poorly differentiatedHR = adjusted hazard ratio, CI = confidence interval and BCSS = Breast Cancer Specific Survival Validation set results Survival analysis This group included the remaining 2,222 patients from the whole cohort. The mean age at diagnosis was 60 years and the median follow-up was 12.4 years. The median tumor size was 2.0 cm. 50% of patients had Grade 3 tumors, 43% were node positive, and 76% were ER positive. After excluding cases that had insufficient invasive tumor, missing core or un-interpretable staining pattern, 1,796 cases were carried forward for the analysis. Out of these, MCs were present in 494 (27.5%) cases. KM survival analysis (Fig. 2b) showed that the presence of stromal MCs was a favourable prognostic marker in the validation set (P = 0.006). Multivariate analysis Cox proportional hazard model was used to carry out the multivariate analysis and included age, tumor grade, tumor size, nodal status, ER and Her2 as independent predictors of BCSS. All the above variables achieved statistical significance as shown in Table 4(b). Presence of MCs did not achieve statistical significance (P = 0.128) but retained a similar HR = 0.846, 95% CI 0.683–1.049 as in the training set analysis. X-tile analysis results We assessed the association between patient outcome and number of positively stained MCs in the stroma of tumors using X-tile software. This software allowed us to define an optimal cut-off point that defined the number of MCs needed to predict good prognosis in the cancer patients. The program divided the population into low and high-risk groups based on the number of MCs detected. It converted continuous data into ordinal classes for statistical analysis. The X-tile plot showed that breast cancer patients with any number of stromal MCs had better prognosis compared to those who have no MC infiltration in their stroma. MCs ranging between 1 and 22 in the tumor stroma were present in 507 (27.61%) patients and were grouped together in the low population group of the training set. The high population group comprised of 1,329 (72.39%) patients in the training set. It reiterated the findings obtained by KM survival analysis that the presence of any number of stromal MCs predicts good prognosis in invasive breast cancer patients. Discussion The importance of the reciprocal relationship between tumor and stroma is being increasingly recognized [21], and is the role of stromal inflammatory cells like MCs, macrophages, fibroblasts and T cell subtypes in cancer initiation and progression. cDNA microarray analysis has shown that genes expressed by stromal cells correlate with differences in the biology of the tumors and are prognostic predictors in breast cancer [22]. There is an ongoing debate about possible detrimental or beneficial effects of MC accumulation in the stroma of solid tumors. In-vitro studies have shown that MC inhibits tumor growth [23]. This is in agreement with other studies showing increase in MC count in early stage of non-small cell lung cancer and thereby supporting their anti-tumor role. [3]. Increased islet/stromal MC-ratio and presence of tumor islet MCs were shown to be independent good prognostic indicators in non-small-cell lung cancer [24]. Recent studies in ovarian cancer [13] and colorectal cancer [15] have also confirmed their correlation with prognosis. MCs secrete factors like heparin, interleukin-8 (IL-8) and vascular endothelial growth factor that promote neovascularization; histamine which suppresses immune response; platelet-derived growth factor, nerve growth factor and stem-cell factors which are mitogenic; and proteases that promote metastasis [9]. On the other hand, MCs can inhibit tumor growth by releasing endogenous peroxidase that is cytotoxic to mammalian tumor cells [25], by natural cytotoxicity [21] and by recruiting neutrophils, eosinophils, lymphocytes and macrophages [26]. MCs can secrete protective substances without degranulation in the presence of tumor-derived blockers like oxidized polyamines and thus be detrimental to the tumor [27]. They can destroy tumor cell surface structures directly and indirectly in a fashion similar to the effect of arginase [28]. MCs can also inhibit tumor growth by secreting beneficial cytokines like IL-1, IL-4, IL-6, and tumor necrosis factor-α that induce apoptosis of endothelial cells [29]; and chondroitin sulphate which inhibits metastases [30]. Cathepsin G secreted by MC activates platelets, lymphocytes and macrophages, and is known to be cytotoxic to some mammalian cells [31]. Its inhibition by a tumor cell product was implicated in the progression of advanced squamous cell tumors [32] suggesting its cytotoxic capability against target cells. MCs secrete chymase, which stimulates apoptosis in different target cells [33] and inhibits angiogenesis by exerting cytotoxic effect on vascular smooth muscles [34]. MC tryptase stimulates inflammation and recruits fibroblasts leading to tumor fibrosis, and thereby limiting tumor growth and metastasis [35]. Tryptase, however, is also implicated in angiogenesis and its effect correlated with a better response to chemotherapy in ovarian cancer. This suggests that MC tryptase-mediated fibrosis and angiogenesis may be responsible for the good prognostic effect seen in advanced ovarian cancer [13]. In this TMA study, we showed that the presence of even one or more MCs in the tumor microenvironment was sufficient to exert a positive prognostic effect. Our finding supports the idea of assessing inflammatory cell infiltrates as prognostic markers in cancer. In our previously published pilot report, we had observed no correlation between the presence of stromal MCs with either B-cells (CD20-positive) or T-cells (CD3-positive) [11], and hence we did not stain for these inflammatory cells in this study. We also did not find a statistically significant prognostic effects of MCs in the node-negative group in the training set analysis. This differs from our earlier published report [11] on a 348 case series which concluded that the presence of MCs was a favourable prognostic factor in the node-negative patients (P = 0.018) but not the node-positive group (P = 0.384). This finding underscores the importance of validating clinically relevant findings on a larger series of patients and using the test-validation approach to arrive at a meaningful conclusion. A number of studies have elaborated on the role of immune cells in cancer etiopathogenesis. T regulatory (Treg) cells are known to be associated with poor outcome in ovarian cancer [36] and the initial results of a clinical trial aimed at specifically eliminating Treg cells has shown promising results [37]. Lymphoma-associated macrophages have been used to predict the outcome of follicular lymphoma [38] and reactive macrophages are used to predict breast, prostate, ovarian, and cervical cancer outcome [39]. Tumor-infiltrating neutrophils are used to predict outcome in patients with adenocarcinoma of the bronchioloalveolar carcinoma subtype [40] and gastric carcinoma [41]. In a recent report, it was concluded that the type, density, and location of immune cells within the tumor samples are a better predictor of patient survival than the histopathological methods currently used to stage colorectal cancer [42]. In conclusion, we confirm our earlier pilot study findings and confirm that stromal MCs correlate with a good prognosis in a large cohort of 4,444 invasive breast cancer patients with long-term follow-up. It highlights the critical role that the host stromal reaction, in particular the inflammatory cell infiltrate, plays in modulating cancer progression. MCs can be used as markers for risk stratification in invasive breast cancers.

Immunogenetics-3-1-1914236

The SPINK gene family and celiac disease susceptibility

The gene family of serine protease inhibitors of the Kazal type (SPINK) are functional and positional candidate genes for celiac disease (CD). Our aim was to assess the gut mucosal gene expression and genetic association of SPINK1, -2, -4, and -5 in the Dutch CD population. Gene expression was determined for all four SPINK genes by quantitative reverse-transcription polymerase chain reaction in duodenal biopsy samples from untreated (n = 15) and diet-treated patients (n = 31) and controls (n = 16). Genetic association of the four SPINK genes was tested within a total of 18 haplotype tagging SNPs, one coding SNP, 310 patients, and 180 controls. The SPINK4 study cohort was further expanded to include 479 CD cases and 540 controls. SPINK4 DNA sequence analysis was performed on six members of a multigeneration CD family to detect possible point mutations or deletions. SPINK4 showed differential gene expression, which was at its highest in untreated patients and dropped sharply upon commencement of a gluten-free diet. Genetic association tests for all four SPINK genes were negative, including SPINK4 in the extended case/control cohort. No SPINK4 mutations or deletions were observed in the multigeneration CD family with linkage to chromosome 9p21-13 nor was the coding SNP disease-specific. SPINK4 exhibits CD pathology-related differential gene expression, likely derived from altered goblet cell activity. All of the four SPINK genes tested do not contribute to the genetic risk for CD in the Dutch population. Introduction Celiac disease (CD) is a chronic inflammatory condition of the small intestine because of an immunological intolerance for the food protein gluten. Patients have to adhere to a life-long diet devoid of gluten to prevent the detrimental effects of a prolonged nutrient and mineral deficiency (Green and Jabri 2003). Susceptibility for CD is predominantly determined by genetic factors, and the complex inheritance patterns suggest the interaction of multiple genes (van Heel et al. 2005). It is well established that the adaptive immune response to gluten plays a pivotal role in the pathogenesis of CD. Th1 activation of CD4+ T cells follows gluten-peptide presentation by DQ2 or DQ8 molecules expressed on antigen-presenting cells (Sollid 2002). The HLA-DQA and -DQB gene variants coding for these molecules are the major genetic determinants for CD susceptibility (Koning et al. 2005). Recently, the importance of innate immunity in CD pathogenesis was also underscored by the observation of induced IL15 expression and NKT cell chemotaxis through the MICA and NKG2D molecules (Hue et al. 2004; Meresse et al. 2004). However, no genetic contribution of the cognate genes has been demonstrated. The notion of crosstalk between the adaptive and innate immune systems is not limited to CD and gets much attention in studies of the inflammatory process (Hoebe et al. 2004). This raises the question whether some aspects of innate immunity may contribute to the genetic susceptibility for CD. The innate immune system uses a wide array of defense mechanisms against the invasion of pathogens. These encompass the expression of pattern recognition receptors, release of antimicrobial molecules, and preservation of epithelial barrier and tissue integrity by, e.g., serine protease inhibitors (Kimbrell and Beutler 2001). One branch of the family of serine protease inhibitors is that of the Kazal type (SPINK) that originally consisted of four members in humans (SPINK1, SPINK2, SPINK4, and SPINK5). Recently, as part of a cluster of SPINK genes on chromosome 5q32 that already included SPINK1 and SPINK5, five new SPINK(-like) members were identified that were located more distally: SPINK5L2, SPINK6, SPINK5L3, SPINK7, and SPINK9, respectively (NCBI Map Viewer, build 36.1). However, these new members lack functional annotation and were therefore not included in this study. SPINK family members 1, 2, and 4 have a comparable size and structure coded for by 4 exons with a single Kazal type serine protease inhibitor domain. SPINK5, in contrast, contains 33 exons that encode 15 inhibitory domains. All four SPINK members are thought to be involved in the protection against proteolytic degradation of epithelial and mucosal tissues, although their major site of expression may differ. SPINK1 is expressed in the pancreas and the gastrointestinal tract, and mutations in this gene are reported in various forms of pancreatitis (Pfutzer and Whitcomb 2001). SPINK2 (located on 4q12) is expressed in the testis, epididymis, and seminal vesicle, where its antimicrobial function may be involved in protection of fertility (Rockett et al. 2004). SPINK4 was originally isolated from pig intestine (Agerberth et al. 1989) and is abundantly expressed in human and porcine goblet cells in the crypts of Lieberkühn but was also found in monocytes and in the central nervous system (Metsis et al. 1992; Norberg et al. 2003). SPINK5 is expressed in the thymus, vaginal epithelium, Bartolin’s glands, oral mucosa, tonsils, and the parathyroid glands (Magert et al. 1999). Mutations in SPINK5 are responsible for the Netherton syndrome, a lethal skin disorder characterized by ichthyosis, hair shaft defects, atopy, skin barrier defects, and recurrent bacterial infections (Bitoun et al. 2002). Mouse models of the Netherton syndrome have shown enhanced proteolysis of desmoglein 1 and filaggrin in SPINK5 mutants (Descargues et al. 2005; Hewett et al. 2005). Moreover, SPINK5 has also been associated with asthma and atopic dermatitis (Blumenthal 2005). Interestingly, both SPINK1 and SPINK5 are located on chromosome 5q32. This region contains the CELIAC2 susceptibility locus that emerged repeatedly in linkage studies (Babron et al. 2003). Despite the fact that this region is rich in candidate cytokine genes and intense mapping efforts were made, no closely associated genes were identified (Ryan et al. 2005). Likewise, SPINK4 is located on chromosome 9p13.3 and resides within a linkage region (9p21-13) where we previously identified a novel CD locus that segregated within a four-generation Dutch family (van Belzen et al. 2004). Taken together, the role of SPINK genes in epithelial and mucosal protection and the important genetic locations of SPINK1, SPINK4, and SPINK5 prompted us to subject the four conventional members of the SPINK family to gene expression and genetic association analyses to ascertain their possible role in CD pathogenesis. Materials and methods Patient material Duodenal biopsy samples were collected by endoscopy as part of a routine CD diagnostic procedure or to monitor the response to a gluten-free diet in previously diagnosed patients. All patients were classified using the Marsh nomenclature according to the UEGW criteria (Report of a working group of the United European Gastroenterology Week in Amsterdam 2001). Two biopsy samples taken in parallel to those used for histological examination were pooled and used for determination of gene expression. In total, 62 individuals were examined with quantitative reverse transcription polymerase chain reaction (qRT-PCR), of which 16 were normal controls, and 46 were CD patients (Fig. 1). The patient group consisted of 15 untreated cases with villous atrophy (MIII) and 31 patients treated with a gluten-free diet who were in various stages of mucosal recovery: MII (crypt hyperplasia; n = 11), MI (lymphocyte infiltration; n = 8), and M0 (complete remission; n = 12). The biopsies of all participating patients were reevaluated by an experienced pathologist, and only CD patients with a proven original Marsh III lesion were included in this study. The genetic association study on all four SPINK genes was initially conducted on a cohort of 310 independent CD patients and 180 independent age- and sex-matched controls, all of which were from Dutch Caucasian decent. In a second stage, exclusively focused on all SPINK4 SNPs, we added 360 controls to a total of 540. In case of three SPINK4 variants with suggestive P values, the power of the study was further enhanced by adding 169 extra CD cases to a total of 479. In parallel, we also examined the SPINK4 gene in a previously described four-generation Dutch CD family (van Belzen et al. 2004). Six family members and a CEPH control were subjected to DNA sequence analysis of the SPINK4 coding regions and splice sites boundaries. Additionally, we performed SPINK4 SNP haplotype analysis in this family. All patients and family members that volunteered for this study signed an informed consent. The study was approved by the Medical Ethics Committee of the University Medical Center Utrecht. Fig. 1Results of qRT-PCR of SPINK genes in normal controls (NC) and CD patients, either untreated (MIII) or on a gluten-free diet (MII–M0). The Marsh stages refer to the pathological conditions of the mucosa, characterized by atrophy of the villi (MIII); hyperplastic crypts between the villi (MIII–MII); and enhanced lymphocyte infiltration (MIII–MII–MI). Stage M0 indicates complete remission comparable to controls. The genes tested were as follows: SPINK1 (a); SPINK2 (b); SPINK4 (c); and SPINK5 (d). Measurements were made in triplicate, on pools of separately prepared cDNA samples. Expression data were normalized to the normal control pool.(e) Relative expression of all four SPINK genes with respect to SPINK2 in the healthy duodenal mucosa. Note the logarithmic scale here. The GUSB gene was used as an endogenous control in all tests. Errorbars indicate standard deviations Expression study The isolation of total RNA from biopsy samples and the analysis of gene expression by real-time qRT-PCR on an ABI Prism 7900HT was performed as described before (Wapenaar et al. 2004). We used the commercially available Assay-on-Demand test for SPINK1 (Hs00162154_m1), SPINK2 (Hs00221653_m1), SPINK4 (Hs00205508_m1), SPINK5 (Hs00199260), and the endogenous control gene GUSB (PDAR 4326320E; Applied Biosystems, Foster City, CA). All samples were tested in triplicate on pooled cDNA samples representing each Marsh class. The results were confirmed with cDNA from individual samples tested in duplicate. Relative levels of gene expression were obtained using the SDS2.1 software (Applied Biosystems). Genetic association and data analysis Haplotype tagging SNPs were selected for SPINK1, SPINK2, SPINK4, and SPINK5 based on HapMap (Phase I) data using Haploview (Barrett et al. 2005). For each haploblock containing SNPs in high linkage disequilibrium, one or more representative SNPs were selected that should capture the genetic variation within that block. For the four SPINK genes tested, this resulted in a set of 18 haplotype tagging SNPs and one coding SNP (see Table 1 and Fig. 2). SNP assays were obtained from Applied Biosystems and analyzed on an ABI Prism 7900HT. Hardy–Weinberg Equilibrium (HWE) was evaluated separately in cases and controls for all SNPs tested. Allele frequencies were compared between cases and controls, and P values were obtained by χ2 analysis. Table 1Allelic distribution of SPINK haplotype tagging SNPs in a Dutch CD case-control cohort Cases (n = 310)Controls (n = 180)  AlleleAllele countsAllele frequency (%)Allele countsAllele frequency (%)Gene nameSNP i.d.Position*Minor/majorMinorMajorMinorMajorMinorMajorMinorMajorχ2P valueSPINK1rs10515593147,178,993A/G11944921797626822.177.90.1660.6834SPINK1rs3777125147,184,010C/G22833640.459.615020042.957.10.5270.4681SPINK1rs4705204147,195,313C/A12542922.677.46027617.982.12.8130.0935SPINK1rs891992147,205,707G/A7748313.886.24730313.486.60.0190.8906SPINK2rs1001563057,520,070A/G21632839.760.313621638.661.40.1020.7489SPINK2rs78154257,528,232G/A22039036.163.912322135.864.20.0090.9237SPINK4rs56335333,206,428G/C12047220.379.76727919.480.60.1120.7375SPINK4rs56351233,206,484G/A505568.391.73631010.489.61.2430.2648SPINK4rs89167133,222,271G/T7054011.588.55230414.685.41.9980.1575SPINK4rs706107**33,230,225A/G11749919818126923.176.92.3580.1246SPINK4rs70610933,230,668A/G11848619.580.58227622.977.11.5490.2133SPINK4rs70611533,243,605G/C7753712.587.55230814.485.60.7160.3975SPINK5rs3756688147,422,972G/A23238037.962.113522137.962.10.0000.9968SPINK5rs4472254147,433,830A/C240376396114221439.960.10.0810.7757SPINK5rs4519913147,452,004G/A27933145.754.316819246.753.30.0790.7792SPINK5rs1422987147,466,552T/C495598.191.9273297.692.40.0700.7917SPINK5rs3815740147,471,386G/A385746.293.8213375.994.10.0470.8291SPINK5rs2052532147,476,500G/A20041432.667.411724132.767.30.0010.9723SPINK5rs3764930147,485,309G/A19941332.567.511724332.567.50.0000.9958*Basepair position according to NCBI build 35.1**Coding SNP, nonsynonymous change (Val7Ile)Fig. 2Genomic organization of the four SPINK genes. The upperhorizontalline indicates exon locations (verticalbars) and SNP positions (numberedasterisks). The SNPs are numbered for each gene consecutively as they appear in Table 1. SPINK4 SNP no. 4 represents the nonsynonymous (Val7Ile) coding SNP rs706107. The arrowpoints indicate the orientation of transcription. The lowerportion of the figure shows the pairwise linkage-disequilibrium structure between indicated SNPs given by D′ statistics based on the European population in the HapMap database (Phase II). Darkerred intensities indicate higher D′ values (numbers indicate D’ value, whereas SNP pairs without number have a D′ = 1) DNA sequence analysis DNA sequence analysis was performed on SPINK4 in six members of a four-generation Dutch CD family and one CEPH control (family, 1,331; individual, 2). Of these six family members, four were affected (index 02, 08, 32, and 41) and carried the disease-linked haplotype, and two were nonaffected (index 21 and 31) without this haplotype (Fig. 3). All coding sequences of the SPINK4 gene were PCR-amplified, including the intron–exon boundaries (for primers and protocols, see supplementary data Table 1). PCR products were examined on a 2% agarose gel and purified with a Millipore Vacuum Manifold (Billerica, MA). Samples were prepared with the BigDye terminator cycle sequencing ready kit (Applied Biosystems) according to the manufacturer’s protocol. PCR and sequencing amplification were performed on a GeneAmp PCR system 9700 (Perkin Elmer, Foster City, CA). Sequences were run on an ABI Prism 3730 analyzer (Applied Biosystems). Analysis and sequence alignment was carried out with Sequence Navigator (Applied Biosystems) and Vector NTI (InforMax, Massachusetts) software packages. Fig. 3Pedigree of the Dutch multigeneration CD family. Only affected descendents are depicted (10 out of 13 siblings in the second generation were affected). The grandparental SPINK4 haplotypes that are boxed and shaded are identical to the grandmaternal at-risk haplotype (noninformative). The SNPs are ordered (top-to-bottom) as they appear in Table 1. Genotype numbers 1, 2, 3, and 4 refer to A, C, G, and T alleles, respectively. Sequence analysis was performed on family members 02, 08, 21, 31, 32, and 41. Family member index numbers are indicated in bold Results SPINK gene expression in the CD mucosa The expression of all four conventional members of the human SPINK family was determined by real-time qRT-PCR on duodenal biopsy-derived cDNA pools from normal controls and CD patients, either untreated or in various stages of remission on a gluten-free diet. The results shown in Fig. 1 indicate that only SPINK4 (Fig. 1c) is differentially expressed, and that its transcriptional activity, which is at its highest in Marsh III (20-fold compared to controls), decreases sharply (fourfold) when patients improve and make a transition to Marsh II. To preclude that the results for SPINK4 might be biased by fortuitous differences in individual expression levels within the generated pools, we also examined the control and case samples each separately. This did not change the observed drop in SPINK4 expression during tissue recovery (see supplementary Fig. 1). Likewise, we performed the same analysis for the other three SPINK genes without affecting the profile already observed in the pools (results not shown). We also examined the relative expression of the four SPINK genes with respect to each other in the normal intestinal mucosa. This showed that both SPINK1 and SPINK4 have the highest expression, which is respectively 480-fold and 240-fold higher compared to SPINK2, whereas SPINK5 is in the same order of magnitude (fivefold) as SPINK2 (Fig. 1e). In conclusion, only the SPINK4 gene appears to be differentially regulated in the intestinal mucosa during recovery from the gluten-evoked CD lesion. This observation prompted us to examine whether SPINK4, or any of the other SPINK genes, could also be causally related to the CD pathogenesis. Genetic association analysis of SPINK genes We designed a haplotype tagging SNP strategy to capture all genetic variation in SPINK1, -2, -4, and -5. An overview of these four SPINK genes with their genomic organization, linkage-disequilibrium structure, and the position of the haplotype tagging SNPs used is depicted in Fig. 2. Initially, these haplotype tagging SNPs were tested in 310 CD cases and 180 controls (Table 1) and showed no significant association for any of the haplotype tagging SNPs in the four SPINK genes. Despite the initial negative result, we decided to pursue SPINK4 further because it is expressed in goblet cells (Metsis et al. 1992), displayed a CD pathology-related differential expression in the intestinal mucosa, and mapped within a CD linkage region (van Belzen et al. 2004). Initially, we expanded the control group with 360 samples to a total of 540 for all SPINK4 SNPs tested. As a result, the Val7Ile coding variant rs706107 and its flanking haplotype tagging SNPs rs891671 and rs706109 yielded suggestive but nonsignificant P values of 0.0595, 0.0510, and 0.1122, respectively (data not shown). To increase the power of the study even further, we subsequently added 169 CD cases to a total of 479. The effect on the P values of the three SNPs tested was such that they dropped below the significance threshold (see Table 2). From this, we conclude that the four SPINK genes tested do not contribute to the genetic susceptibility in the Dutch CD population. Table 2Allelic distribution of three selected SPINK4 SNPs in the extended Dutch CD case-control cohort Cases (n = 479)Controls (n = 540)  AlleleAllele countsAllele frequency (%)Allele countsAllele frequency (%)Gene nameSNP i.d.Position*Minor/majorMinorMajorMinorMajorMinorMajorMinorMajorχ2P valueSPINK4rs89167133,222,271G/T12681413.486.615890414.985.10.8890.3457SPINK4rs706107**33,230,225A/G20573721.878.224281422.977.10.3820.5365SPINK4rs70610933,230,668A/G21272422.677.424281622.977.10.0140.9053*Basepair position according to NCBI build 35.1**Coding SNP, nonsynonymous change (Val7Ile) SPINK4 sequence analysis in a multigeneration family We have previously described a four-generation CD family with an extraordinary high incidence of affected individuals (see Fig. 3). The disease segregated with a grandmaternal haplotype on chromosome 9p21-13 (van Belzen et al. 2004), a region that encompasses SPINK4. The apparent dominant inheritance pattern could be caused by a mutation that is rare in the general CD population but present with a high phenotypic penetration in this specific family. To assess if any functional variants of the SPINK4 gene were present in this family, we sequenced all its exons and intron–exon boundaries in six family members. However, we did not observe mutations in any of the samples tested (results not shown). Neither was the exon 1 coding SNP rs706107 specific for affected individuals as all seven individuals tested (including the CEPH control) carried the most frequent GG genotype (Fig. 3). To exclude the possibility of deletions in SPINK4 to be misinterpreted from the sequence data as homozygous genotypes, we also performed segregation analysis of the grandparental SPINK4 haplotypes within the entire family but observed no suspect inheritance pattern (Fig. 3). In conclusion, we have found no evidence that SPINK4 is a candidate gene for the chromosome 9p21-13 CD locus in the Dutch population in general or in the multigeneration Dutch CD family specifically. Discussion Chronic inflammatory conditions and autoimmune disorders are typically characterized by a deregulated adaptive and innate immune system. The innate defense consists of multiple components that include physical barriers, antimicrobial molecules, pattern recognition receptors, circulating phagocytes, and the complement system (Hoebe et al. 2004). A breach of the epithelial barrier and loss of microbial containment is often the first of a series of events that trigger or sustain chronic inflammatory diseases (Tlaskalova-Hogenova et al. 2004) as described, e.g., in Crohn’s disease, atopic eczema, asthma, and psoriasis (Schreiber et al. 2005). In CD, the gut–lumen separation is undermined by dietary gluten that evokes a combined innate and adaptive immune response (Londei et al. 2005). It is the joined action of gluten peptides, environmental factors, and genetic determinants that precipitates this enteropathy. The human leukocyte antigen locus is the major genetic contribution to the adaptive Th1 reaction (Koning et al. 2005). Recently, we identified MYO9B as a susceptibility gene in the Dutch population that possibly has an effect on epithelial barrier integrity (Monsuur et al. 2005). Several other studies have underscored the involvement of innate immunity in CD, however, without identification of underlying causative gene variants (Londei et al. 2005). Interestingly, it was also reported that the epithelial glycocalyx and the bacterial composition in the CD gut is distinct (Forsberg et al. 2004; Tjellstrom et al. 2005). In search of genes that may have a primary contribution to CD pathogenesis, we focused our attention to the SPINK family of serine protease inhibitors that play an important role in tissue preservation through the containment of uncontrolled proteolysis and bacterial growth. In this study, we demonstrated differential gene expression of mucosal SPINK4 in CD. Crypt hyperplasia is a feature of the Marsh III and Marsh II stages of CD, and the concomitant increase in the number of goblet cells may contribute to the increased SPINK4 expression. However, the observed sharp decrease in gene expression sets in during the MIII/MII transition, whereas crypt normalization is observed only later at the MII/MI recovery phase. This suggests that SPINK4 downregulation sets in soon after commencement of the gluten-free diet. This SPINK4 differential expression probably reflects altered goblet cell activity, but its functional significance and regulatory mechanism in CD pathology remains to be established. The combination of functional relevance and mapping to CD linkage intervals pointed to the SPINK family members as attractive functional and positional candidate genes. We have chosen a robust strategy for genetic association testing based on haplotype tagging SNPs and linkage-disequilibrium structure of the SPINK loci applied to a considerably sized Dutch case-control cohort. With our study design, we had 75% power to confirm association with SPINK1, -2, and -5 (relative risk 2.0; allele frequency 0.1–0.45; 95% confidence interval), whereas this was even 95% (RR 2.0) and 80% (RR 1.6) for SPINK4. These power estimates reflect a Type I error rate of 0.05, which is appropriate for testing a previously reported result. Initial detection of a new genetic association would require much more stringent criteria to assure reproducibility, and power would be correspondingly less. In parallel, we examined the extended Dutch CD family for variants and deletions in SPINK4. We hypothesized that a specific SPINK4 mutation, although rare in the general population, could have a dramatic impact on mucus composition, bacterial containment, and gluten sensitivity, thereby explaining the apparent dominant and high penetration inheritance pattern in our extended CD family. With both approaches, we were not able to establish a genetic involvement of the SPINK genes tested. However, we cannot completely rule out the possibility of a rare noncoding mutation in SPINK4 (outside the splice donor and acceptor regions) that might specifically segregate in this atypical CD family, characterized by an exceptional high prevalence of affected members. Despite this negative result in the Dutch CD population, we cannot formally rule out the possibility of genetic contribution of SPINK genes to CD in other European populations like the Italian in whom, unlike the Dutch (van Belzen et al. 2003), chromosome 5q linkage was established (Greco et al. 1998; Percopo et al. 2003). Genuine population heterogeneity has been reported before, e.g., between CARD15/NOD2 and Crohn’s disease (Lesage et al. 2002; Croucher et al. 2003) and between SPINK5 and asthma (Blumenthal 2005; Jongepier et al. 2005). The new SPINK members on chromosome 5q (SPINK5L2, SPINK6, SPINK5L3, SPINK7, and SPINK9) were not part of this study. Currently, no functional annotation is available for these genes that are located near SPINK1 and SPINK5 in a chromosomal region that appears to have been subjected to gene duplication during evolution. Therefore, we cannot exclude their possible involvement in CD or any other inflammatory disorder. Electronic Supplementary Material Below is the link to the electronic supplementary material. Supplementary Fig. S1 (GIF 14.0 kb) Supplementary Table 1 Primers and reaction conditions used for SPINK4 DNA sequence analysis. (PDF 80.0 KB)

Appl_Psychophysiol_Biofeedback-4-1-2259255

EEG Biofeedback as a Treatment for Substance Use Disorders: Review, Rating of Efficacy, and Recommendations for Further Research

Electroencephalographic (EEG) biofeedback has been employed in substance use disorder (SUD) over the last three decades. The SUD is a complex series of disorders with frequent comorbidities and EEG abnormalities of several types. EEG biofeedback has been employed in conjunction with other therapies and may be useful in enhancing certain outcomes of therapy. Based on published clinical studies and employing efficacy criteria adapted by the Association for Applied Psychophysiology and Biofeedback and the International Society for Neurofeedback and Research, alpha theta training—either alone for alcoholism or in combination with beta training for stimulant and mixed substance abuse and combined with residential treatment programs, is probably efficacious. Considerations of further research design taking these factors into account are discussed and descriptions of contemporary research are given. Introduction Substance use disorders (SUD) include disorders related to the taking of a drug of abuse (including alcohol), and represent the most common psychiatric conditions (APA 2000) resulting in serious impairments in cognition and behavior. Acute and chronic drug abuse results in significant alteration of the brain activity detectable with quantitative electroencephalography (qEEG) methods. The treatment of addictive disorders by electroencephalographic (EEG) biofeedback (or neurofeedback, as it is often called) was first popularized by the work of Eugene Peniston (Peniston and Kulkosky 1989, 1990, 1991) and became popularly known as the Peniston Protocol. This approach employed independent auditory feedback of two slow brain wave frequencies, alpha (8–13 Hz) and theta (4–8 Hz) in an eyes closed condition to produce a hypnagogic state. The patient was taught prior to neurofeedback to use what amounts to success imagery (beingsober, refusing offers of alcohol, living confidently, and happy) as they drifted down into an alpha-theta state. Repeated sessions reportedly resulted in long-term abstinence and changes in personality testing. Because the method seemed to work well for alcoholics, it has been tried in subjects with cannabis dependence and stimulant dependence—but with limited success until the work of Scott and Kaiser (Scott and Kaiser 1998; Scott et al. 2002, 2005). They described treating stimulant abusing subjects with attention-deficit type EEG biofeedback protocols, followed by the Peniston Protocol, with substantial improvement in program retention and long-term abstinence rates. This approach has become known widely as the Scott–Kaiser modification (of the Peniston Protocol). This “white paper” on EEG biofeedback for SUD will offer an assessment of efficacy according to the guidelines jointly established by the Association for Applied Psychophysiology and Biofeedback (AAPB) and the International Society for Neurofeedback and Research (ISNR). Assessing the efficacy of neurofeedback for SUD involves several considerations. The first of these involves difficulties assessing the efficacy of any treatment method for SUD. Outcome benchmarks (i.e., total abstinence, improved function and quality of life) and time points of outcome (i.e., one year, two years post treatment) are not clearly established. Outcome assessment for treatment of SUD in itself is a complex topic well beyond the scope of this article. Because different drugs of abuse are associated with different patterns of EEG abnormality, as will be discussed in detail in this article, it is difficult to assign broad-brush EEG biofeedback solutions to SUD as a whole. Any statements of efficacy will need to describe specific EEG biofeedback protocols for specific substances of abuse. Furthermore substance abuse is often mixed substance type and comorbid conditions are common and vary from subject to subject, as will also be borne out in this article. As of yet there are no gold standard medication or other treatments for the various types of SUD and efficacy of any SUD treatment method likely falls into the “possibly effective” to “probably effective” range according to the efficacy guidelines jointly established by the AAPB and ISNR. Finally, all of the studies of EEG biofeedback in SUD to date employ EEG biofeedback as an add on to cognitive behavioral or twelve step treatment regimes, so any statements of efficacy would have to acknowledge that EEG biofeedback is not a stand alone treatment for SUD. This article is divided into several sections. In the first section after “Introduction,” we review SUD prevalence and describe qEEG changes typical for the most widespread drugs of abuse (alcohol, marijuana, heroin, cocaine, and methamphetamine). The second section describes treatment studies employing EEG biofeedback in SUD. Studies that have used the Peniston Protocol are described first, along with critical commentaries of these studies. In the second part of this section, a description of the Scott–Kaiser modification is given, along with some discussion of a rationale for why this approach may be more successful with stimulant abusers. This section also describes some current research. The third section assesses efficacy of the Peniston Protocol and the Scott–Kaiser modification. The fourth section takes a look at the clinical implications of comorbidities in neurobiofeedback treatment of alcohol and drug abuse. The fifth section discusses the clinical implications of standard cognitive-behavioral therapies in SUD treatment and reviews the rationale for the application of qEEG-guided neurofeedback intervention in SUD in conjunction with these therapies. The final section summarizes findings in qEEG and neurofeedback in SUD and additionally proposes further directions for clinical research in this area. This article represents an update of earlier reviews (Trudeau 2000, 2005a, b) of EEG biofeedback for addictive disorders extended with a review on qEEG in SUD. This review is presented as one of a series of papers in both The Journal of Neurotherapy and The Journal of Applied Psychophysiology & Biofeedback describing and reviewing biofeedback applications for adult populations. No attempt will be made to review the fields of qEEG and neurobiofeedback generally (see current reviews by Hammond 2006; Kaiser 2006), or the field of addictive disorders generally, although some references will be made to specifics the authors feel are pertinent to a discussion of emerging concepts of qEEG as a sensitive tool for the brain function assessment in SUD, and EEG biofeedback as a treatment approach for SUD. SUD Prevalence and qEEG Changes Drug addiction can be described as a mental disorder with idiosyncratic behavioral, cognitive, and psychosocial features. The SUD commonly referred to as “drug addiction” is characterized by physiological dependence accompanied by the withdrawal syndrome on discontinuance of the drug use, psychological dependence with craving, the pathological motivational state that leads to the active drug-seeking behavior, and tolerance, expressed in the escalation of the dose needed to achieve a desired euphoric state. Drug addiction is a chronic, relapsing mental disease that results from the prolonged effects of drugs on the brain (Dackis and O’Brain 2001; Volkow et al. 2003, 2004). Drug addiction can take control of the brain and behavior by activating and reinforcing behavioral patterns that are excessively directed to compulsive drug use (Di Chiara 1999; Gerdeman et al. 2003). From the 11 classes of substances listed in the DSM-IV we will discuss in our review only alcohol, cannabis (marijuana), heroin, and such psychostimulants as cocaine and methamphetamine. Addiction leads to behavioral, cognitive, and social adverse outcomes that incur substantial costs to society. In 2002, it was estimated from the Substance Abuse and Mental Health Service Administration (SAMHSA 2004) that 22 million Americans have a substance abuse or dependence disorder, and 2 million of them were current cocaine users (Vocci and Ling 2005). In 2005, there were 2.4 million persons who were current cocaine users, which is more than in 2004 (SAMHSA 2006). The number of current crack users increased from 467,000 in 2004 to 682,000 in 2005. According to the 2004 revised National Survey on Drug Use and Health, nearly 12 million Americans have tried methamphetamine, and 583,000 of them are chronic methamphetamine users (SAMHSA 2004). In 2005, an estimated 22.2 million persons aged 12 or older were classified with substance dependence or abuse in the past year (9.1% of the population aged 12 or older). Of these, 3.3 million were classified with dependence on or abuse of both alcohol and illicit drugs, 3.6 million were dependent on or abused illicit drugs but not alcohol, and 15.4 million were dependent on or abused alcohol but not illicit drugs. There were 18.7 million persons classified with dependence on or abuse of alcohol in 2005 (7.7%). The specific illicit drugs that had the highest levels of past year dependence or abuse in 2005 were marijuana, followed by cocaine and pain relievers. Of the 6.8 million persons aged 12 or older classified with dependence on or abuse of illicit drugs, 4.1 million were dependent on or abused marijuana in 2005. This number represents 1.7% of the total population aged 12 or older and 59.9% of all those classified with illicit drug dependence or abuse. Marijuana was the most commonly used illicit drug (14.6 million past month users). In 2005, it was used by 74.2% of current illicit drug users. Among current illicit drug users, 54.5% used only marijuana, 19.6% used marijuana and another illicit drug, and the remaining 25.8% used only an illicit drug other than marijuana in the past month (SAMHSA 2006). Fatal poisoning, which include overdoses (ODs) on illicit drugs, alcohol, and medications, is the leading cause of injury death for individuals age 35–44 and the third leading cause of injury death overall, trailing motor vehicle accidents and firearm-related deaths (CDC 2004). Heroin-related ODs have increased at an alarming rate in portions of the US and other countries (Darke and Hall 2003; Landen et al. 2003), and OD has surpassed HIV infection as the primary cause of death for heroin users. Not surprisingly, heroin is frequently associated with opioid-related ODs, both as a single drug and in combination with other substances (CDC 2004). Many patients seeking treatment for addiction have multiple drug dependencies and psychiatric comorbidities (Volkow and Li 2005). Information from epidemiological surveys indicates that drug addiction is a common phenomenon and is associated with significant effects on both morbidity and mortality. Large individual and societal costs of drug abuse make research and treatment of drug addiction imperative (French et al. 2000; Mark et al. 2001). Recently through intensive clinical neurophysiological research and biological psychiatric studies many specific components of cognitive, emotional, and behavioral deficits typical for SUD have been identified and investigated. However, the practical values of these cognitive neuroscience and applied psychophysiology-based treatment (e.g., neurofeedback) findings depend on a further integration of these methodological approaches. qEEG in Substance Use Disorders EEG in Alcoholism EEG alterations have been described extensively in alcoholic patients (Porjesz and Begleiter 1998), but any attempt at drawing a common picture from qEEG data is difficult due to significant methodological differences, such as different definitions of frequency bands, different filtering methodology, number of channels, reference choice, etc. However, most reports of alcoholic patients agree in describing alterations mainly within the beta (Bauer 1997, 2001a; Costa and Bauer 1997; Rangaswamy et al. 2002, 2004) and/or alpha bands (Finn and Justus 1999). The qEEG and LORETA mapping studies of detoxified alcohol-dependent patients, as compared with normal controls, showed an increase in absolute and relative beta power and a decrease in alpha and delta/theta power (Saletu et al. 2002), which is in agreement with earlier reports of low-voltage fast EEG patterns, as often encountered by visual EEG inspection (Niedermeyer and Lopes da Silva 1982). As slow activities are considered to be inhibitory, alpha activity may be viewed as an expression of normal brain functioning and fast beta activities as excitatory, the low-voltage fast desynchronized patterns may be interpreted as hyperarousal of the central nervous system (CNS) (Saletu-Zyhlarz et al. 2004). The investigations by Bauer (2001a) and Winterer et al. (1998) showed a worse prognosis for the patient group with a more pronounced frontal CNS hyperarousal. It may be hypothesized that these hyperaroused relapsing patients require more CNS sedation than abstaining ones. The EEG maps of alcohol-dependent patients differ significantly from those of normal controls and patients suffering from other mental disorders and might be useful for diagnostic purposes (Pollock et al. 1992; Saletu et al. 2002; Saletu-Zyhlarz et al. 2004). Decreased power in slow bands in alcoholic patients may be an indicator of brain atrophy and chronic brain damage, while an increase in the beta band may be related to various factors such as medication use, family history of alcoholism, and/or hallucinations, suggesting a state of cortical hyperexcitability (Coutin-Churchman et al. 2006). Abnormalities in resting EEG are often associated with a predisposition to development of alcoholism. Subjects with a family history of alcoholism were found to have reduced relative and absolute alpha power in occipital and frontal regions and increased relative beta in both regions compared with subjects with a negative family history of alcoholism. These results suggest that resting EEG alpha abnormalities are associated with risk for alcoholism, although their etiological significance is unclear (Finn and Justus 1999). Alcohol-dependent individuals have different synchronization of brain activity than light drinkers as reflected by differences in resting EEG coherence (Kaplan et al. 1985, 1988; Michael et al. 1993; Winterer et al. 2003a) and power (e.g., Bauer 2001ab; Enoch et al. 2002; Rangaswamy et al. 2002; Saletu-Zyhlarz et al. 2004). Most differences in EEG coherence and power are found in the alpha and beta bands. Non-alcohol-dependent relatives of alcohol-dependent individuals also have EEG differences in alpha and beta coherence (Michael et al. 1993) and power (Bauer and Hesselbrock 2002; Finn and Justus 1999; Rangaswamy et al. 2002, 2004) as compared to subjects without alcohol-dependent relatives. This indicates that differences in functional brain activity as measured with qEEG in alcohol-dependent patients not only relate to the impact of long-term alcohol intake, but possibly also to genetic factors related to alcohol dependence. Both alcohol dependence (Schuckit and Smith 1996) and EEG patterns (Van Beijsterveldt and Van Baal 2002) are highly heritable. In addition, some genes coding for GABA receptors in the brain, which mediate the effects of alcohol, are related to certain EEG patterns (Porjesz et al. 2005; Winterer et al. 2003b). Moreover, some GABA-receptor genes that are related to EEG patterns are also associated with the risk to develop alcohol dependence. These associations again suggest that genetic factors play a major role in the EEG differences associated with alcohol dependence. The EEG coherence analysis is a technique that investigates the pairwise correlations of power spectra obtained from different electrodes. It measures the functional interaction between cortical areas in different frequency bands. A high level of coherence between two EEG signals indicates a co-activation of neuronal populations and provides information on functional coupling between these areas (Franken et al. 2004). De Bruin et al. (2004, 2006) investigated the pure effects of alcohol intake on synchronization of brain activity, while minimizing the confounding influence of genetic factors related to alcohol dependence. They showed that heavily drinking students with a negative family history had stronger EEG synchronization at theta and gamma frequencies than lightly drinking students with a negative family history. This study suggests that, in students, heavy alcohol intake has an impact on functional brain activity, even in the absence of genetic factors related to alcohol dependence. The findings of studies on the effects of alcohol dependence on EEG coherence can be summarized as follows: Kaplan et al. (1985) reported lower frontal alpha and slow-beta coherence in alcohol-dependent males and females. Michael et al. (1993) found higher central alpha and slow-beta coherence, but lower parietal alpha and slow-beta coherence in males with alcohol dependence. Winterer et al. (2003a, b) described higher left-temporal alpha and slow-beta coherence and higher slow-beta coherence at right-temporal and frontal electrode pairs in alcohol-dependent males and females. De Bruin et al (2006) showed that moderate-to-heavy alcohol consumption is associated with differences in synchronization of brain activity during rest and mental rehearsal. Heavy drinkers displayed a loss of hemispheric asymmetry of EEG synchronization in the alpha and slow-beta band. Moderately and heavily drinking males additionally showed lower fast-beta band synchronization. Therefore, qEEG alterations have been described extensively in alcoholics. Most EEG reports in alcoholic patients agree in describing alterations mainly within the beta and alpha bands. Patients with a more pronounced frontal hyperarousal have worse prognosis. Decreased power in slow bands in alcoholic patients may be an indicator of chronic brain damage, while increase in beta band may be related to various factors suggesting cortical hyperexcitability. Abnormalities in resting EEG are highly heritable traits and are often associated with a predisposition to alcoholism development. The studies on the effects of alcohol dependence on EEG coherence can be summarized as lower frontal alpha and slow-beta coherence in alcohol-dependent patients with some topographical coherence abnormality differences between alcohol-dependent males and females. EEG in Marijuana Abuse Several lines of evidence suggest that cannabis (marijuana, tetrahydrocannabinol—THC) may alter functionality of the prefrontal cortex and thereby elicit impairments across several domains of complex cognitive function (Egerton et al. 2006). Several studies in both humans and animals have shown that cannabinoid exposure results in alterations in prefrontal cortical activity (Block et al. 2002; O’Leary et al. 2002; Whitlow et al. 2002), providing evidence that cannabinoid administration may affect the functionality of this brain area. Despite the fact that a number of transient physiological, perceptual and cognitive effects are known to accompany acute chronic marijuana (THC) exposure in humans, persistent qEEG effects in humans resulting from continuing exposure to this drug have been difficult to demonstrate (Wert and Raulin 1986). In early reviews of EEG and ERP studies of acute and chronic THC exposure in humans (Struve et al. 1989, 1994), it was reported that significant associations between chronic exposure and clinically abnormal EEG patterns had not been demonstrated and that attempts to use visual EEG analyses to detect transient acute THC exposure induced EEG alterations failed to demonstrate consistent THC–EEG effects across studies. Quantitative methods of analyzing EEG spectra from single posterior scalp derivations began to be applied to studies of acute THC exposure. These early studies reported that acute THC exposure produced transient increases in either posterior alpha power, decreases in mean alpha frequency or increases in alpha synchrony (Fink et al. 1976; Struve et al. 1989; Tassinari et al. 1976; Volavka et al. 1971, 1973). These studies found that THC produced a transient dose-dependent rapid onset: (1) increase in relative power (amount, abundance) of alpha; (2) decrease in alpha frequency; and (3) decrease in relative power of beta as measured from posterior scalp electrodes. Later studies of Struve et al. (1998, 1999, 2003) demonstrated and replicated a significant association between chronic marijuana use and topographic qEEG patterns of persistent “alpha hyperfrontality” (i.e., elevations of alpha absolute power, relative power, and interhemispheric coherence over frontal cortex) as well as reductions of alpha mean frequency. These findings from chronic users are consistent with both non-topographic (Hockman et al. 1971; Tassinari et al. 1976; Volavka et al. 1973) and topographic (Lukas et al. 1995; Struve et al. 1994) transient EEG effects of acute THC administration. Therefore, chronic daily THC use was found to be associated with distinct topographic qEEG features. Compared with non-users, THC users had significant elevations of absolute and relative power, and interhemispheric coherence of alpha activity over the bilateral frontal cortex (referred to as “alpha hyperfrontality”). A second finding was that the voltage (not relative power or coherence) of all non-alpha frequency bands was significantly elevated in THC users, although the voltage increase was generalized and not frontally dominant. A third finding involved a widespread decrease in the relative power of delta and beta activity for cannabis users, particularly over the frontal cortical regions. A fourth finding was that interhemispheric coherence of theta and possibly delta activity was also significantly elevated over frontal cortex for marijuana users. Because most studies included daily THC users and non-users drawn from an inpatient psychiatric population, the effects of psychiatric diagnoses or medication were not controlled. Thus, qEEG studies on acute THC exposure reported a transient dose-dependent increase in relative power of alpha, decrease in alpha frequency, and decrease in relative power of beta at posterior EEG recording sites. Chronic marijuana abuse is known to result in a number of physiological, perceptual and cognitive effects, but persistent qEEG effects from continuing exposure to THC have been difficult to demonstrate. However, recent studies of Struve and his colleagues have demonstrated a significant association between chronic marijuana use and topographic qEEG patterns of persistent elevations of alpha absolute power, relative power, and interhemispheric coherence over frontal cortex, as well as reductions of alpha mean frequency. Another important qEEG finding was the elevated voltage of all non-alpha bands in THC users. A third qEEG finding involved a widespread decrease in the relative power of delta and beta activity over the frontal cortical regions in marijuana users. EEG in Heroin Addiction Only a few studies have investigated qEEG changes in heroin addicts. Qualitative changes were observed in more than 70% of heroin addicts in the early abstinence (acute withdrawal) period, and these included low-voltage background activity with diminution of alpha rhythm, an increase in beta activity, and a large amount of low-amplitude delta and theta waves in central regions (Olivennes et al. 1983; Polunina and Davydov 2004). Franken et al. (2004) found that abstinent heroin-dependent subjects have an enhanced fast beta power compared with healthy controls, and this finding is concordant with other EEG studies on alcohol and cocaine abusing subjects (Costa and Bauer 1997; Herning et al. 1994b; Rangaswamy et al. 2004; Roemer et al. 1995). Spectral power and event-related potentials (ERP) in heroin addicts strongly relate to abstinence length (Shufman et al. 1996, Bauer 2001a; Polunina and Davydov 2004). Most studies showed considerable or even complete normalization of EEG spectral power or magnitude of ERP components in heroin ex-addicts who maintained abstinence for at least 3 months (Bauer 2001b, 2002; Costa and Bauer 1997; Papageorgiou et al. 2001; Polunina and Davidov 2004; Shufman et al. 1996). Some quantitative changes were also reported in methadone-maintenance heroin addicts (Gritz et al. 1975), current heroin addicts, and subjects in heroin abstinence less than 80 days (Shufman et al. 1996). Gritz et al. (1975) demonstrated a significant slowing of occipital alpha rhythm peak frequency in 10 methadone-maintained patients and the same trend in 10 abstinent heroin-addicted subjects. In one study (Polunina and Davydov 2004), slowing of slow alpha (8–10 Hz) mean frequency was significantly related to the amount of heroin taken by these patients daily before withdrawal. The prolongation of ERP component latencies in heroin addicts was also reported (Papageorgiou et al. 2001), and these delays significantly correlated with years of heroin use, rather than with abstinence length in the study of Bauer (1997). Polunina and Davydov (2004) demonstrated frequency shifts in the fast alpha range at the frontal and central recording sites and a slowing of slow alpha mean frequency at the central, temporal, and occipital sites of recording in heroin abusers who used heroin for at least 18 months. In general, pronounced desynchronization is characteristic for acute heroin withdrawal, but as it was mentioned above, several studies (Bauer 2001a, 2002; Costa and Bauer 1997; Papageorgiou et al. 2001; Polunina and Davydov 2004; Shufman et al. 1996) showed that spectral power of EEG tends to normalize almost completely after several weeks of abstinence. The most consistent changes in EEG of heroin addicts were reported in alpha and beta frequencies, and included a deficit in alpha activity and an excess of fast beta activity in early heroin abstinence. The latter abnormality appears to reverse considerably when heroin intake is stopped for several months, and therefore it may be viewed as an acute withdrawal effect. The dynamics and characteristics of spectral power changes within the early opiate withdrawal suggest the participation of catecholamine imbalances, especially noradrenaline and perhaps to a lesser degree dopamine, which are widely recognized as a main cause of opiate physical dependency symptoms (Devoto et al. 2002; Maldonado 1997). Acute opiate administration has been shown to increase, while abstinence from chronic opiate use has been shown to decrease extracellular dopamine (DA) in the nucleus accumbens. In contrast, extracellular DA in the prefrontal cortex is not modified by acute opiate use, but is markedly increased during morphine and heroin abstinence syndrome (Devoto et al. 2002). Relationships between theta and beta frequencies shifts and neurotransmitter imbalances characteristic for heroin withdrawal remain unclear. Withdrawal state in heroin addicts is known to elicit a strong craving for drug, anxiety, nervousness, deficits in inhibitory control, dysphoric motivational state, and intrusive thoughts related to drugs (Franken 2003; Franken et al. 1999, 2004; Stormark et al. 2000). Research on functional connectivity in drug withdrawal states is restricted to a few studies on coherence of the EEG signal in abstinent heroin users (Franken et al. 2004; Fingelkurts et al. (2006a), active heroin abusers (Fingelkurts et al. 2006b), and in abstinent polysubstance abusers (Roemer et al. 1995). In a study on 22 opioid-dependent patients under acute opioid influence, Fingelkurts et al. (2006b) showed that longitudinal opioid exposure impairs cortical local and remote functional connectivity, and found that local connectivity increased, whereas the remote one decreased. These findings were interpreted as specific signs of independent processing in the cortex of chronic heroin addicts. It has been suggested that such independent processes may constitute the candidate mechanism for a well-documented pattern of impairment in addicts that expresses the lack of integration of different cognitive functions for effective problem solving and helps to explain the observed deficits in abstract concept formation, behavioral control, and problems in the regulation of affect and behavior. Specifically, Fingelkurts et al. (2006b) found that the number and strength of remote functional connections among different cortical areas estimated by the index of EEG synchrony was significantly higher in patients in acute heroin withdrawal than in healthy controls for most categories of functional connections. Although this result was observed in the alpha as well as in the beta frequency bands, it was most prominent for the beta range. In the same patient sub-sample under acute opioid influence the authors (Fingelkurts et al. 2006a) observed the opposite: a significant decrease in the number and strength of remote functional connections, when compared with healthy controls. Thus, the increase of remote synchronicity among cortical areas during the short-term withdrawal period may indicate the selective attentional focus on cues and memories related to drugs while ignoring neutral cues (Franken et al. 2000; Sokhadze et al. 2007). Generally this can explain a narrowing of the behavioral repertoire and compulsive drug seeking in abstinent addicted subjects (Vanderschuren and Everitt 2004). Therefore, the elevated synchrony within the beta frequency band in these studies (Fingelkurts et al. 2006a, b) may reflect a state of CNS activation toward reward-seeking behavior, with this being a prerequisite of relapse among opiate drug dependent patients (Bauer 2001a). qEEG changes in heroin addicts in the acute withdrawal period have been described as low-voltage background activity with a diminution of alpha rhythm, an increase in beta activity, and a large amount of low-amplitude delta and theta waves in central regions. In general, pronounced desynchronization is characteristic for acute heroin withdrawal, but the spectral power of EEG tends to normalize almost completely after several weeks of abstinence. The most consistent changes in EEG of heroin addicts were reported in the alpha and beta frequencies, and included a deficit in alpha activity and an excess of fast beta activity in early heroin abstinence. The excess of beta appears to reverse considerably when heroin intake is stopped for several months, and therefore it may be viewed as an acute withdrawal effect. Recent studies found that the number and strength of remote functional connections among different cortical areas estimated by the index of EEG synchrony for the beta range was significantly higher in patients in acute heroin withdrawal than in healthy controls for most categories of functional connections. EEG in Cocaine Addiction Qualitative and quantitative EEG measures are highly sensitive to the acute and chronic effects of neurointoxication produced by such psychostimulants as cocaine, as well as effects from withdrawal and long-term abstinence from cocaine use (Ehlers et al. 1989). However, some EEG characteristics observed in cocaine addicts are considered to be due to the toxic effects of this drug on the brain, whereas some EEG characteristics in cocaine addicts may also indicate a predisposition toward the development of SUD (Porjesz et al. 2005). Hans Berger (1937, cited by Gloor 1969; Herning et al. 1985) was the first to study the effects of cocaine on human EEG, reporting an increase in activity in the beta bandwidth. This was replicated in subsequent studies with a larger number of subjects (Alper 1999; Alper et al. 1990, 1998; Costa and Bauer 1997; Herning et al. 1985; Noldy et al. 1994; Prichep et al. 1996, 1999, 2002; Roemer et al. 1995). Beside beta effects, studies have reported an increase in delta activity (Herning et al. 1985) and frontal alpha activity (Herning et al. 1994b), while others have reported an increase in alpha wave EEG associated with bursts of cocaine-induced euphoria (Lukas 1991). More recently, researchers have begun analyzing qEEG profiles of cocaine-dependent patients using the spectral power of each primary bandwidth over the different topographic cortical areas. Excess alpha activity (Alper et al. 1990; Herning et al. 1994b; Lukas 1991; Prichep et al. 1996) and decreased delta activity (Alper et al. 1990; Noldy et al. 1994; Prichep et al. 1996; Roemer et al. 1995) have been reported, while others have reported increased beta power (Herning et al. 1985, 1994b; Noldy et al. 1994) in cocaine-dependent patients, recorded in eyes closed, resting conditions. The qEEG abnormalities, primarily found in anterior cortical regions, were shown to correlate with the amount of prior cocaine use (Herning et al. 1996a; Prichep et al. 1996; Roemer et al. 1995; Venneman et al. 2006). The qEEG has been used more often to characterize the effects of withdrawal in cocaine-dependent patients. Several studies reported that during protracted abstinence from cocaine qEEG effects are featured by long-lasting increases in alpha and beta bands together with reduced activity in delta and theta bands (Alper et al. 1990; Prichep et al. 1996; Roemer et al. 1995). Recently Reid et al. (2006) investigated qEEG profiles in cocaine-dependent patients in response to an acute, single-blind, self-administered dose of smoked cocaine base (50 mg) versus placebo. Cocaine produced a rapid increase in absolute theta, alpha, and beta power over the prefrontal cortex, lasting up to 25 min after administration of the drug. The increase in theta power was correlated with a positive subjective drug effect (“high”), and the increase in alpha power was correlated with nervousness. Cocaine also produced a similar increase in delta coherence over the prefrontal cortex, which was correlated with nervousness. Placebo resulted only in a slight increase in alpha power over the prefrontal cortex. These data demonstrate the involvement of the prefrontal cortex in the qEEG response to acute cocaine, and indicate that slow wave qEEG, delta and theta activity are involved in the processes related to experiencing rewarding properties of cocaine. Prichep et al. (1999, 2002) extended the idea of relating baseline EEG activity to outcome in cocaine-dependent patients in treatment programs. Subjects with cocaine dependence have persistent changes in brain function assessed with qEEG methods, present when evaluated at baseline, 5–14 days after last reported crack cocaine use, and persistent at one and six month follow-up evaluations (Alper 1999; Alper et al. 1990, 1998; Prichep et al. 1996, 2002; Venneman et al. 2006). Several recent studies employing qEEG techniques have already demonstrated an association between the amount of beta activity in the spontaneous EEG and relapse in cocaine abuse (Bauer 1997, 2001a). A decrease in the delta and theta bands of the EEG can be regarded as a specific sign of brain dysfunction. However, this sign, as well as other qEEG abnormal patterns, can be found in many different psychiatric disorders and none of them can be considered as pathognomonic of any specific mental or neurological disorder. EEG coherence in cocaine addiction was investigated in only one study (Roemer et al. 1995). The authors reported globally reduced interhemispheric coherence in the delta and theta bands, and frontally in the beta band. It should be noted that subjects in this study were cocaine-preferring polysubstance abusers during abstinence and these results can hardly be generalized to crack cocaine-only users or other categories of cocaine-dependent subjects not enrolled in any treatment. Therefore, acute effects of smoked crack cocaine have been shown to produce a rapid increase in absolute theta, alpha, and beta power over the prefrontal cortex, lasting up to half-an-hour after administration of the drug. The increase in theta power was reported to correlate with a positive subjective drug effect, while the increase in alpha power was reported to correlate with nervousness. qEEG measures are also sensitive to the acute and chronic effects of cocaine, as well as the effects from withdrawal and long-term abstinence from cocaine use. Some EEG characteristics observed in cocaine addicts are considered to be due to the neurotoxic effects, whereas some EEG characteristics in cocaine addicts may also indicate a predisposition toward the development of cocaine addiction. qEEG has been used more often to characterize the effects of withdrawal in cocaine-dependent patients. During protracted abstinence from cocaine qEEG effects are featured by long-lasting increases in alpha and beta bands together with reduced activity in delta and theta bands. Several recent studies employing qEEG techniques have demonstrated an association between the amount of beta activity in the spontaneous EEG and relapse in cocaine abuse. EEG in Methamphetamine Addiction Several studies have examined the neurobiological consequences of methamphetamine dependence using qEEG methods (e.g., Newton et al. 2003, 2004). It was found that methamphetamine dependent patients exhibited a significant power increase in the delta and theta bands as compared to non-drug-using controls (Newton et al. 2003). These results are in accordance with other neurocognitive studies (Kalechstein et al. 2003) suggesting that methamphetamine abuse is associated with psychomotor slowing and frontal executive deficits. Within the methamphetamine-dependent subjects, increased theta qEEG power was found to correlate with response time and was accompanied with reduced accuracy (Newton et al. 2004). To our knowledge, qEEG patterns associated with acute withdrawal and recent abstinence in methamphetamine dependence have not yet been sufficiently described. One study reported (Newton et al. 2003) that methamphetamine dependent volunteers with 4 days of abstinence had increased EEG power in the delta and theta but not in the alpha and beta bands. Within the methamphetamine dependent group, a majority of the conventional EEGs were abnormal (64%), compared to 18% in the non-methamphetamine using group. The qEEG may provide a sensitive neurophysiological outcome measure of methamphetamine abuse-related persistent alterations in neurocognitive functions (Newton et al. 2004). In a study by Simon et al. (2002), when performance of patients with SUD was compared to their matched non-using control groups, both methamphetamine and cocaine abusers were impaired on cognitive measures, but the type and degree of impairments were somewhat different. Some of these differences between methamphetamine and cocaine effects on cognitive functions and electrophysiological alterations can be explained by differential pharmacokinetics of these two drugs, as cocaine is rapidly metabolized with an elimination half-life of several hours, whereas methamphetamine is eliminated more slowly, with an elimination half-life averaging 12 h (Cook et al. 1993; Jeffcoat et al. 1989). Moreover, cocaine differs from methamphetamine in that cocaine inhibits the reuptake of dopamine, serotonin, and norepinephrine, whereas methamphetamine mobilizes and releases these monoamines from storage granules, thus producing rapid and large increases in synaptic concentrations (Simon et al. 2002, 2004). This might be responsible for the discrepancies in observed qEEG manifestations associated with chronic methamphetamine and cocaine abuse. Only a few studies have examined the qEEG consequences of methamphetamine dependence. They report that methamphetamine dependent patients exhibited a significant power increase in the delta and theta bands as compared to non-drug-using control. The qEEG patterns associated with acute withdrawal and recent abstinence in methamphetamine dependence have not yet been sufficiently described. One study reported that abstinent methamphetamine dependent patients had increased EEG power in the delta and theta but not in the alpha and beta bands. In general, qEEG studies of methamphetamine addiction are in accordance with other neurocognitive studies suggesting that methamphetamine abuse is associated with psychomotor slowing and frontal executive deficits. P300 Abnormalities in Cocaine, Methamphetamine, Heroin Addiction, and Alcoholism The P300 component of the ERP, occurring 300–600 ms post-stimulus, is the most widely used ERP in psychiatry and other clinical applications (Polich et al. 1994; Polich and Herbst 2000; Pritchard 1981, 1986; Pritchard et al. 2004). The amplitude of the P300 reflects the allocation of attentional resources, while the latency is considered to reflect stimulus evaluation and classification time (Katayama and Polich 1998; Polich and Herbst 2000). The P300 is usually obtained in an oddball paradigm, wherein two stimuli are presented in a random order, one of them frequent (standard) and another one rare (target) (Polich 1990). A modification of the oddball task has been used where a third, also rare stimulus (distracter), is presented along with standard and target stimuli. It was reported that these infrequent distracters elicit a frontocentral P300, so called P3a, whereas the rare targets elicit a parietal P300, so called P3b (Katayama and Polich 1996, 1998). The P3a is recorded at the anterior scalp locations and has been interpreted as reflecting frontal lobe activity (Gaeta et al. 2003; Knight 1984). Though the P300 response in general is thought to represent “context updating/closure,” in a three-stimuli oddball task the P3a is interpreted as “orienting,” and the P3b is viewed as an index of the ability to maintain sustained attention to target (Näätänen 1990). The anterior P3a indexes the contextual salience of the rare stimuli, whereas the posterior P3b is indexing task-relevance of the stimuli (Gaeta et al. 2003). A robust finding in ERP studies on alcoholism is that alcoholics as well as individuals at high risk to develop alcoholism have been shown to have a low P300 amplitude in various task paradigms (Cohen et al. 2002; Hada et al. 2000; Porjesz et al. 2005; Porjesz and Begleiter 1998). Kouri et al. (1996) examined the P300 component in patients who were dually dependent on cocaine and heroin. The results showed no P300 amplitude differences between the patients and healthy non-drug-dependent volunteers when patients presented for detoxification. However, after the course of detoxification, the P300 amplitude was significantly smaller in the cocaine- and heroin-dependent group than in the non-dependent control group. In a study by Bauer (2001b) the P300 did not differentiate among patients characterized by histories of either cocaine, or cocaine and alcohol, or heroin dependence. Across all the patient groups, the P300 was significantly reduced in amplitude relative to the P300 ERPs recorded from individuals with no history of alcohol or drug dependence. This study also demonstrated that continued abstinence from heroin and from cocaine and alcohol is also associated with a trend toward normalization of the P300. In a recent study of Papageorgiou et al. (2004) the P300 component was evaluated during the anticipatory period of a short memory task in 20 patients characterized by a past history of heroin dependence (6 months abstinence), in 18 current heroin users and in 20 matched healthy subjects. Abstinent heroin addicts exhibited a significant reduction of the P300 amplitude at the central frontal region, relative to the other two groups. The results of early work examining the effect of cannabis use and THC administration on visual and auditory ERPs have been inconclusive (Rodin et al. 1970; Roth et al. 1973). Later studies of Patrick et al. (1995, 1997) could not find P300 latency differences in audio and visual oddball tasks between THC users without psychiatric problems and controls. Although THC users displayed reduced auditory and visual P300 amplitudes in this study, when age differences between THC users and controls were removed, all significant P300 amplitude differences were removed as well. Acute and chronic use of cocaine exerts neuropharmacological effects on amplitude and latency of both anterior and posterior P300 ERP components (Biggins et al. 1997; Fein et al. 1996; Herning et al. 1994a; Kouri et al. 1996; Polich 1990). Longer P300 (P3b) latency without abnormalities in amplitude was reported in several studies on cocaine withdrawal (Herning et al. 1994a; Lukas 1993). Noldy and Carlen (1997) demonstrated effects of cocaine withdrawal on the latency of the P300 in an auditory oddball task. In cocaine-dependent patients, P3a amplitude decrements over frontal areas are persistent even after long periods of abstinence (Bauer 1997). The latency of the P3a was delayed and the amplitude was reduced to novel non-targets in cocaine and alcohol-dependent subjects compared to controls (Biggins et al. 1997; Hada et al. 2000) in auditory and visual three-stimuli oddball tasks. Several studies have investigated ERP changes associated with methamphetamine abuse and dependence. The P300 component of the auditory ERP was reported to show a prolonged latency in the oddball task in methamphetamine dependent subjects with a history of psychosis, compared to normal controls (Iwanami et al. 1994, 1998). In particular, the patients with methamphetamine dependence showed reduced P3a amplitude in the reading task and delayed P3b latency with normal P3b amplitude in the auditory oddball task. This was interpreted as indicating a prolonged central noradrenergic dysfunction due to earlier methamphetamine use. In most ERP studies the P300 did not differentiate among patients characterized by histories of either cocaine, or cocaine and alcohol, or heroin dependence. Across all the patient groups, the P300 was significantly reduced in amplitude relative to P300 ERPs recorded from individuals with no history of alcohol or drug dependence. The latency of the frontal and parietal P300 was reported to be delayed, and the amplitude was reduced to novel non-targets in cocaine and alcohol-dependent subjects compared to controls in auditory and visual three-stimuli oddball tasks. Continued abstinence from heroin, cocaine, and alcohol was shown to be associated with a trend toward P300 normalization. Several studies have investigated ERP changes associated with methamphetamine abuse and dependence. In general, chronic psychoactive substance abuse and drug dependence are associated with delayed and attenuated cognitive ERP in auditory and visual oddball tasks. qEEG and ERP Abnormalities in Addiction: Psychopharmacological Effects or Trait Markers? Whether qEEG alterations and P300 decrements found in most of SUD are only a coincident “marker” of vulnerability or make a direct etiologic contribution to risk for substance dependence is still unknown (Bauer and Hesselbrok 2001; Carlson et al. 2002; O’Connor et al. 1994; Polich et al. 1994; Porjesz and Begleiter 1998). The P300 reduction and abnormal qEEG patterns are seen in mental disorders that often are comorbid with substance abuse, such as conduct disorder (Bauer and Hesselbrock 1999, 2001), ADHD (Bauer 1997; O’Connor et al. 1994), and bipolar or major affective disorder (Friedman and Squires-Wheeler 1994). Reduced P300 amplitude related to prefrontal brain dysfunction may suggest that a deficit in inhibitory control is an underlying mechanism shared by different psychopathologies (Bauer and Hesselbrock 1999; Clark et al. 1999; Tarter et al. 2003). According to Bauer (2002), certain ERP and qEEG abnormalities and impaired functioning on complex cognitive tests in patients formerly dependent on cocaine might not be proximately caused by drug use per se but be more related to comorbid alcohol use or another psychiatric condition. Taken together, the findings converge on the conclusion that there exists an inherited predisposition for an externalizing psychopathology that includes ADHD, conduct disorder, and substance abuse. PTSD seems to heighten the risk for addiction as well. Thus, the reviewed findings support the hypothesis that addicted subjects may manifest a P300 amplitude reduction and qEEG abnormalities as a trait reflecting the CNS disinhibition, which may be a predisposing factor for addiction liability, resistance to drug habit extinction, and relapse vulnerability. Heritability and Neurotransmitter Considerations in Substance Use Disorders There has been a consistent drift in addiction research between the psychosocial, cognitive and behavioral aspects of addiction and the biological and genetic emphasis. In much of the present data relating to genetics and animal models (Blum et al. 2006; Porjesz et al. 2005; Ryabinin and Weitemier 2006; Samochowiec et al. 2006), studies suggest that a genetic predisposition for SUD is an accepted concept. Much of the genetic research addresses the influence of alleles thought responsible in coding for genes that express phenotypic neurotransmitter production and distribution; mainly involving endorphins, dopamine and serotonin. These neurotransmitters, dopamine in particular, are also suspect in other appetitive and mood disorders and psychopathologies, of particular note, Reward Deprivation Syndrome (RDS). RDS is described as a dysfunction in the Brain Reward Cascade and proposes that abnormal craving behavior is a consequence of defects in the DRD2 and D1, D3, D4 and D5 dopaminergic receptor genes (Blum et al. 2006). Blum and colleagues (1990, 1993, 1996) described this syndrome and identified the D2 dopamine receptor gene as a possible candidate for susceptibility to alcoholism in severe alcoholics (Blum et al. 1993) and proposed this gene’s association with dopamine production and distribution may produce a sevenfold increase in the likelihood of developing alcohol use problems (Uhl et al. 1993). This DRD2 dopamine receptor gene and polymorphisms within its genetic coding specific to addiction remain unclear due to its involvement in other disorders; including, obesity (Blum et al. 2006), Tourette’s syndrome (Comings et al. 1991) pathological aggression and violence, PTSD (Comings et al. 1996) and schizoid—avoidant disorder (Chen et al. 2005). SUD were classified as a subtype of RDS and treatment regimens for these disorders have been classified as inadequate (Blum et al. 2007) and research continues in developing possible genetic interventions that may produce dopamine and other neurotransmitter regulation in substance-induced rapid dopamine increase in limbic regions (Blum et al. 2007). It is clear that heritability plays an important role in addictive disorders, however, to what extent environment, perception and synaptic permanency and plasticity influence the course of genetic adaptation or maladaptive traits requires further investigation. Suggested neuroanatomical substrates involved in SUD implicate mesolimbic and diencephalon regions; including the substantia nigra, reticular formation, medial forebrain bundle, nucleus accumbens, septum pediculum, olfactory tubercule and hippocampus and suggest that any concentration of alcohol exposure to these regions would make alcohol use virtually unavoidable (Myers and Privette 1989). Studies of EEG Biofeedback in Substance Abuse Treatment The Peniston Protocol (Alpha-Theta Feedback) The early studies of Kamiya (e.g., Nowlis and Kamiya 1970) on self-regulation of alpha rhythm elicited substantial interest in the potential clinical applications of alpha biofeedback for SUD treatment. There were reported several uncontrolled case studies and conceptual reviews on alpha EEG training for alcohol (DeGood and Valle 1978; Denney et al. 1991; Jones and Holmes 1976; Passini et al. 1977; Tarbox 1983; Watson et al. 1978) and drug abuse treatment (Brinkman 1978; Goldberg et al. 1976, 1977; Lamontagne et al. 1977; Sim 1976), but the impact of alpha biofeedback training as a SUD therapy was not significant. The bulk of the literature to date regarding EEG biofeedback of addictive disorders is focused on alpha-theta biofeedback. The technique involves the simultaneous measurement of occipital alpha (8–13 Hz) and theta (4–8 Hz) and feedback by separate auditory tones for each frequency representing amplitudes greater than pre set thresholds. The subject is encouraged to relax and to increase the amount of time the signal is heard, that is to say, to increase the amount of time that the amplitude of each defined bandwidth exceeds the threshold. A variety of equipment and software has been used to acquire, process, and filter these signals, and there are differences in technique inherent with equipment and software. Alpha-theta feedback training was first employed and described by Elmer Green and colleagues (Green et al. 1974) at the Menninger Clinic. This method was based on Green’s observations of single lead EEG during meditative states in practiced meditators, during which increased theta amplitude was observed following an initial increased alpha amplitude, then a drop off of alpha amplitude (theta/alpha crossover). When the feedback of the alpha and theta signal was applied to subjects, states of profound relaxation and reverie were reported to occur. The method was seen as useful in augmenting psychotherapy and promoting individual insight. It could be seen as a use of brain wave signal feedback to enable a subject to maintain a particular state of consciousness similar to a meditative or hypnotic relaxed state over a 30- or 40-min feedback session. Goslinga (1975) gave the first description of the use of alpha-theta feedback in a SUD treatment program. This integrated program started in 1973 at the Topeka VA, and included group and individual therapies. Daily 20-min EEG biofeedback sessions (integrated with EMG biofeedback and temperature control biofeedback) were conducted over 6 weeks, resulting in free, loose associations, heightened sensitivity, and increased suggestibility. Patients discussed their insights and experiences associated with biofeedback in therapy groups several times a week, augmenting expressive psychotherapy. The first published clinical reports of efficacy of alpha-theta training at the Topeka VA were by Twemlow and Bowen (1976), who explored the impact of alpha-theta training on psychodynamic issues in 67 non-psychotic chronic male alcoholics in an inpatient treatment program. In this non-controlled study, they found that “religiousness” as a predictor of “self-actualization” may have increased as a result of imagery experienced in theta states. This was seen as positive to the program goal of augmenting Alcoholics Anonymous as a recovery philosophy. The high suggestibility of the method was acknowledged; “treatments such as brainwave training, which utilize abstract, ill understood techniques are potential repositories of magical projection and fantasy and would logically be more acceptable to alcoholics who are able to have ‘faith’ (devoutly or moderately religious)” (Twemlow and Bowen 1977). In another uncontrolled study at the Topeka VA, 21 alcoholics were reported to exhibit within and across session increases in raw theta amplitudes at occipital areas bilaterally measured by single lead EEG during the course of alpha-theta training, becoming more able to achieve deep states as manifested by EEG (Twemlow et al. 1977). These initial studies advanced the utility of biofeedback induced theta states in promoting insight and attitude change in alcoholics, with the assumptions that biofeedback-induced theta states are associated with heightened awareness and suggestibility, and that this heightened awareness and suggestibility would enhance recovery. Outcome data regarding abstinence were not reported. In the first reported randomized and controlled study of alcoholics treated with alpha-theta EEG biofeedback, Peniston and Kulkosky (1989) described positive outcome results. Their subjects were inpatients in a VA hospital treatment program, all males with established chronic alcoholism and multiple past failed treatments. Following a temperature biofeedback pre-training phase, Peniston’s experimental subjects (n = 10) completed 15 30-min sessions of eyes closed occipital alpha-theta biofeedback. Compared to a traditionally treated alcoholic control group (n = 10), and nonalcoholic controls (n = 10), alcoholics receiving brainwave biofeedback showed significant increases in percentages of EEG recorded in the alpha and theta rhythms, and increased alpha rhythm amplitudes (single lead measurements at international site O1). The experimentally treated subjects showed reductions in Beck Depression Inventory scores compared to the control groups. Control subjects who received standard treatment alone showed increased levels of circulating beta-endorphin, an index of stress, whereas the EEG biofeedback group did not. Thirteen-month follow-up data indicated significantly more sustained prevention of relapse in alcoholics who completed alpha-theta brainwave training as compared to the control alcoholics, defining successful relapse prevention as “not using alcohol for more than six contiguous days” during the follow-up period. In a further report on the same control and experimental subjects, Peniston and Kulkosky (1990) described substantial changes in personality test results in the experimental group as compared to the controls. The experimental group showed improvement in psychological adjustment on 13 scales of the Millon Clinical Multiaxial Inventory compared to the traditionally treated alcoholics who improved on only two scales and became worse on one scale. On the 16-PF personality inventory, the neurofeedback training group demonstrated improvement on seven scales, compared to only one scale among the traditional treatment group. This small n study employed controls and blind outcome evaluation, with actual outcome figures of 80% positive outcome versus 20% in the traditional treatment control condition at 4-year follow up. The protocol described by Peniston at the Fort Lyons VA cited above is similar to that initially employed by Twemlow and colleagues at the Topeka VA and Elmer Green at the Menninger Clinic, with two additions, i.e., (1) temperature training and (2) script. Peniston introduced temperature biofeedback training as a preconditioning relaxation exercise, along with an induction script to be read at the start of each session. This protocol (described as follows) has become known as the “Peniston Protocol” and has become the focus of research in subsequent studies. Subjects are first taught deep relaxation by skin temperature biofeedback for a minimum of five sessions that additionally incorporates autogenic phrases. Peniston also used the criteria of obtaining a temperature of 94° before moving on to EEG biofeedback. Participants then are instructed in EEG biofeedback and in an eyes closed and relaxed condition, receive auditory signals from an EEG apparatus using an international site O1 single electrode. A standard induction script employing suggestions to relax and “sink down” into reverie is read. When alpha (8–12 Hz) brainwaves exceed a preset threshold, a pleasant tone is heard, and by learning to voluntarily produce this tone, the subject becomes progressively relaxed. When theta brainwaves (4–8 Hz) are produced at a sufficiently high amplitude, a second tone is heard, and the subject becomes more relaxed and according to Peniston, enters a hypnagogic state of free reverie and high suggestibility. (Although theta increase and alpha decrease are thought by Peniston to be associated with a deeply relaxed state where hypnagogic reverie is present, this may simply represent drowsiness) (Niedermeyer 1999). Following the session, with the subject in a relaxed and suggestible state, a therapy session is conducted between the subject and therapist where the contents of the imagery experienced is explored and “abreactive” experiences are explored (Peniston and Kulkosky 1989, 1990, 1991). Saxby and Peniston (1995) reported on 14 chronically alcohol dependent and depressed outpatients using this same protocol of alpha-theta brainwave biofeedback. Following treatment, subjects showed substantial decreases in depression and psychopathology as measured by standard instruments. Twenty-one month follow-up data indicated sustained abstinence from alcohol confirmed by collateral report. These male and female outpatients received 20 40-min sessions of feedback. Bodenhamer-Davis and Calloway (2004) reported a clinical trial with 16 chemically dependent outpatients, 10 of whom were probationers classified as high risk for re-arrest. Subjects completed an average of 31 alpha-theta biofeedback sessions. Psychometrics demonstrated improvements in personality and mood. Follow-up at 74–98 months indicated 81.3% of the treatment subjects were abstinent. Re-arrest rates and probation revocations for the probation treatment group were lower than those for a probation comparison group (40% vs. 79%). Fahrion (1995) gave a preliminary report (n = 119) on a large randomized study of alpha-theta training for addiction in the Kansas Prison System using group-training equipment. A report of the completed study (n = 520) (Fahrion 2002) showed little difference between the two groups overall at 2-year outcome. But, when results were analyzed for age, race and drug of choice, neurofeedback emerged as a more efficacious treatment for younger and non-white and non-stimulant abusing participants. Interestingly, this protocol was not effective for cocaine abusers. (Stimulant abusers will be discussed later in this article under the Scott–Kaiser modification of the Peniston protocol.) The issue of alpha-theta biofeedback in culturally sensitive groups that have not responded to traditional modes of addiction treatment (such as confrontational group therapies) has been considered in an open case series reported by Kelly (1997). This three year follow-up study presented the treatment outcomes of 19 Dine’ (Navajo) clients. Four (21%) participants achieved “sustained full remission,” 12 (63%) achieved “sustained partial remission,” and 3 (16%) remained “dependent.” The majority of participants also showed a significant increase in “level of functioning”. Schneider et al. (1993) used slow cortical potential biofeedback to treat 10 unmedicated alcoholic patients in four neurofeedback sessions after hospitalization. Seven patients participated in a fifth session an average of 4 months later. Six out of these seven patients had not had a relapse at the follow-up. These results are similar to those reported for alpha theta training. Several other studies using the Peniston protocol and its modifications reported cases with positive clinical effects (Burkett et al. 2003, DeBeus et al. 2002; Fahrion et al. 1992; Finkelberg et al. 1996; Skok et al. 1997). These studies suggest that an applied psychophysiological approach based on an alpha-theta biofeedback protocol is a valuable alternative to conventional substance abuse treatment (Walters 1998). Nevertheless, most of these results were reported at the society meetings, and only few of these studies were published in mainstream peer-reviewed journals other than The Journal of Neurotherapy. A critical analysis of the Peniston Protocol is discussed at length in the previous reviews (Trudeau 2000, 2005a, b). Several controlled studies of the Peniston protocol for addictions, completed by Lowe (1999), Moore and Trudeau (1998), and Taub and Rosenfeld (1994), suggest that alpha-theta training for addictions may be non-specific in terms of effect when compared to suggestion, sham or controlled treatment, or meditational techniques. By contrast, Egner et al. (2002) showed that alpha-theta training results in an increase of theta/alpha ratios, as compared to a control condition. In an in depth critical analysis that examines inconsistencies reported in the original Peniston papers, Graap and Freides (1998) raise serious issues about the reporting of original samples and procedures in these studies. In their analyses, the results may have been due as much to the intense therapies accompanying the biofeedback as due to the biofeedback itself. The subjects may have been comorbid for a number of conditions, which were not clearly reported, particularly PTSD, which may have been the focus of the treatment. In his reply to these criticisms, Peniston (1998) acknowledges that it “remains unknown whether the temperature training, the visualizations, the ATBWNT (alpha-theta brain wave neurotherapy), the therapist, the placebo, or the Hawthorne effects are responsible for the beneficial results.” The criticism raised above by Graap and Friedes (1998) regarding Peniston’s papers could also be applied to earlier replication studies. Neither Peniston’s studies nor the replication studies provide sufficient detail regarding the specifics of the types of equipment used for alpha-theta feedback, including filtering methods for the EEG signal or other technical information, to permit exact reproduction of the feedback protocols with other equipment. Outcome criteria also vary in the replication studies, with varying measures of abstinence and improvement. An exception to these concerns is the report of Scott et al. (2005), which will be discussed later in greater detail. It should be noted that psychostimulant (cocaine, methamphetamine) addictions may require approaches and neurofeedback protocols other than alpha/theta training. Persons who are cocaine-dependent are cortically under-aroused during protracted abstinence (Roemer et al. 1995). qEEG changes, such as a decrease in high beta (18–26 Hz) power are typical for withdrawal from cocaine (Noldy et al. 1994). Cocaine abusers who are still taking this drug often show low amounts of delta and excess amounts of alpha and beta activity (Alper 1999; Prichep et al. 1999), whereas chronic methamphetamine abusers usually exhibit excessive delta and theta activity (Newton et al. 2003). Thus, cocaine and methamphetamine users may warrant a different EEG biofeedback protocol, at least at the beginning stages of neurofeedback therapy. The Scott–Kaiser Modification of the Peniston Protocol Scott and Kaiser (1998) describe combining a protocol for attentional training (beta and/or SMR augmentation with theta suppression) with the Peniston protocol (alpha-theta training) in a population of subjects with mixed substance abuse, rich in stimulant abusers. The beta protocol is similar to that used in ADHD (Kaiser and Othmer 2000) and was used until measures of attention normalized, and then the standard Peniston protocol without temperature training was applied (Scott et al. 2002). The study group is substantially different than that reported in either the Peniston or replication studies. The rationale is based in part on reports of substantial alteration of qEEG seen in stimulant abusers associated with early treatment failure (Prichep et al. 1996, 2002) likely associated with marked frontal neurotoxicity and alterations in dopamine receptor mechanisms (Alper 1999). Additionally, preexisting ADHD is associated with stimulant preference in adult substance abusers, and is independent of stimulant associated qEEG changes. These findings of chronic EEG abnormality and high incidence of preexisting ADHD in stimulant abusers suggest they may be less able to engage in the hypnagogic and auto-suggestive Peniston protocol (Trudeau et al. 1999). Furthermore, eyes-closed alpha feedback as a starting protocol may be deleterious in stimulant abusers because the most common EEG abnormality in crack cocaine addicts is excess frontal alpha (Prichep et al. 2002). In their initial report, Scott and Kaiser (1998) described substantial improvement in measures of attention and also of personality (similar to those reported by Peniston and Kulkosky 1990). Their experimental subjects underwent an average of 13 SMR-beta (12–18 Hz) neurofeedback training sessions followed by 30 alpha-theta sessions during the first 45 days of treatment. Treatment retention was significantly better in the EEG biofeedback group and was associated with the initial SMR-beta training. A subsequent published paper (Scott et al. 2005) reported on an expanded series of 121 inpatient drug program subjects randomized to condition, followed up at 1 year. Subjects were tested and controlled for the presence of attentional and cognitive deficits, personality states and traits. The experimental group showed normalization of attentional variables following the SMR-Beta portion of the neurofeedback, while the control group showed no improvement. Experimental subjects demonstrated significant changes (p < .05) beyond the control subjects on 5 of the 10 scales of the MMPI-2. Subjects in the experimental group were also more likely to stay in treatment longer and more likely to complete treatment as compared to the control group. Finally, the one-year sustained abstinence levels were significantly higher for the experimental group as compared to the control group. The approach of beta training in conjunction with alpha-theta training has been applied successfully in a treatment program aimed at homeless crack cocaine abusers in Houston, as reported by Burkett et al. (2003), with impressive results. Two hundred and seventy (270) male addicts received 30 sessions of a protocol similar to the Scott Kaiser modification. One-year follow-up evaluations of 94 treatment completers indicated that 95.7% of subjects were maintaining a regular residence; 93.6% were employed/in school or training, and 88.3% had no subsequent arrests. Self-report depression scores dropped by 50% and self-report anxiety scores by 66%. Furthermore, 53.2% reported no alcohol or drug use 12 months after biofeedback, and 23.4% used drugs or alcohol only one to three times after their stay. This was a substantial improvement from the expected 30% or less expected recovery in this group. The remaining 23.4% reported using drugs or alcohol more than 20 times over the year. Urinalysis results corroborated self-reports of drug use. The treatment program saw substantial changes in length of stay and completion. After the introduction of the neurofeedback to the mission regimen, length of stay tripled, beginning at 30 days on average and culminating at 100 days after the addition of neurotherapy. In a later study the authors reported follow-up results on 87 subjects after completion of neurofeedback training (Burkett et al. 2005). The follow-up measures of drug screens, length of residence, and self-reported depression scores showed significant improvement. It should be noted that this study had limitations, because neurofeedback was positioned only as an adjunct therapy to all other faith-based treatments for crack cocaine abusing homeless persons enrolled in this residential shelter mission and was an uncontrolled study. Yet the improvement in program retention is impressive and may well be related to the improved outcome. Continuing Research Self-Perception and Experimental Schemata in the Addicted Brain Rex Cannon, Joel Lubar, and Deborah Baldwin of the Brain Research and Neuropsychology Laboratory at University of Tennessee at Knoxville are performing research with three goals in mind: First, to attempt to reconcile and integrate data from all disciplines involved in addiction research in order to develop a novel approach for neurophysiological study pertaining to SUD and conceivably determine and describe EEG source generators that are instrumental in the processes of self-perception and experiential schemata utilizing a recently developed assessment instrument. Second, to utilize this information to develop an integrative treatment model for addictive disorders based on this research, involving novel group processing methods and spatial specific neurophysiological operant learning (LORETA Neurofeedback) (Cannon et al. 2006, 2007; Congedo 2003; Congedo et al. 2004), and finally, third, to utilize both the assessment and neurophysiological data for development of statistical models for possible diagnostic and predictive purposes and to provide a means for a neurophysiological measure of treatment efficacy. Research indicates that substance abusers have elevated beta activity in an EEG resting state as compared with normative groups (Rangaswany et al. 2002) and elevated alpha activity after administering a mood altering substance (Cohen et al. 1993; Kaplan et al. 1985). It is suggested that many of the neurophysiological markers may provide information about the state of the individual prior to the development of an addictive disorder and that these brain functions are under genetic control (Porjesz et al. 2002, 2005; Tapert 2004). Kaplan et al. (1985) reported lower frontal alpha and slow-beta coherence in alcohol-dependent males and females. Michael et al. (1993) found higher central alpha and slow-beta coherence, but lower parietal alpha and slow-beta coherence in males with alcohol dependence; contrarily, other findings suggest that morphine, alcohol and marijuana show increased alpha 2 power in the spectral EEG and relate this to the euphoric state produced by the drug (Lukas 1989, 1995). Winterer et al. (2003a, b) described higher left-temporal alpha and slow-beta coherence and higher slow-beta coherence at right-temporal and frontal electrode pairs in alcohol-dependent males and females. De Bruin et al. (2004) showed that moderate-to-heavy alcohol consumption is associated with differences in synchronization of brain activity during rest and mental rehearsal. Heavy drinkers displayed a loss of hemispheric asymmetry of EEG synchronization in the alpha and low-beta band. Moderately and heavily drinking males additionally showed lower fast-beta band synchronization. Decision-making processes and the ability to form a resistance to drugs, i.e., the ability to say no, involve numerous brain regions; including, the insular, somatosensory, orbitofrontal, anterior cingulate and dorsolateral prefrontal cortices, as well as the amygdala, hippocampus and thalamic nuclei (Bechara 2005). This research considers the integration of the features of addicted persons as reported in earlier studies, case reports and theoretical concepts as vital in understanding behavioral manifestations of the suspected neural pathways that are premised to be involved in the development of SUD. Some of the fundamental descriptions of addicted individuals portray them as passive with dependent strivings, emotionally immature, abounding with fears of responsibility or independent action and ultimately, infantile inadequate personalities (Coodley 1961), as well as emotionally, socially, and educationally underdeveloped (Meyerstein 1964), and immature and regressive (Dorsey 1961; Gerard and Kornetsky 1955; Hill 1962). These individuals are reported to struggle with affirming positive thoughts of self-esteem, tendencies to undervalue themselves and be self deprecating, and exhibit difficulty adjusting to others and these tendencies are veiled by overt behavioral patterns, including, physical or verbal abuse. Individuals with SUD present with a vast number of paradoxical characteristics; including an overwhelming sense of inadequacy disguised by an apparent overwhelming sense of confidence. Similarly, an apparent abundance of anger and aggression utilized as a disguise for a paralyzing sense of fear, more specifically, fear of people, economic insecurity, rejection, and alienation, which paradoxically are exacerbated by the continued use of the substance. One of the more profound idiosyncratic characteristics of this population is the tendency to ruminate and associate past events, perceptions and the associated emotions with both present and future. The perception of experience is often clouded by the personalization of events (real or imagined) and reinforced with a deliberate, ambiguous effort to avoid reconciling this confound, which reinforces an uninhibited association of all current interactions and situations with past events. Opposite to what often is implied, these features may not originate from the consequences of substance abuse, but from earlier periods in development (Vos 1989), and in the perspective of this research these features and others have an etiology in specific neurophysiological regions that are the direct result of dendritic pruning that occurs in early development that continues on into adolescence and, unless intervention or awareness of these schemata are achieved, they remain problematic into adulthood. To date, studies identifying such schematic source generators and their relationship with SUD using qEEG and standardized low-resolution electromagnetic tomography (sLORETA) are scant. This research is designed to assess the neural activation patterns relative to schemata regarding the self in recovering addicts and identify possible generators in the cortex as compared to controls. In this research, it is hypothesized that there is dendritic pruning early in developmental phases that contribute to frequency specific activity in neuronal populations in the ventromedial portions of the prefrontal cortex and limbic regions. Furthermore, it is proposed that these neural pathways hinder the integration of affect, cognition, reward and decision-making processes and adversely influence the perception of self and self in relation to experience and the development of adaptive schemata and personality characteristics. Integration of Cognitive Neuroscience Approaches in Assessment of Functional Outcomes of Neurofeedback and Behavioral Therapy Based Interventions in SUD Sokhadze et al. (2007a) in their conceptual review proposed an integrated approach to assessment and treatment utilizing cognitive neuroscience methods (e.g., qEEG, ERP), conventional psychotherapeutic treatment, and neurofeedback therapy to assess the recovery of cognitive and emotional functions affected by chronic psychostimulant drug abuse co-occurring with PTSD. Cognitive neuroscience methodologies used for assessment of the outcome effects of psychotherapy and neurofeedback interventions for comorbid disorders have significant potential for additionally identifying neurophysiological and clinical markers of treatment progress (Sokhadze 2005). These outcome markers may provide useful information for planning bio-behavioral interventions in this form of dual diagnosis. Stotts and colleagues (2006) at the University of Texas at Houston, in collaboration with researchers at Rice University, used motivational interviewing (MI) with personalized feedback, particularly employing the ERP markers of deficiencies in selective attention task produced by cocaine abuse in crack addicts. In a randomized, controlled pilot study these authors (Stotts et al. 2006; Sokhadze et al. 2004, 2005) evaluated the feasibility and preliminary efficacy of a brief MI intervention using EEG/ERP graphical feedback for crack cocaine abusers. Treatment-seeking cocaine abusers (N = 31) were randomly assigned to a two-session MI intervention or a general educational drug conseiling (control) condition. All participants received EEG assessments based on dense-array ERP tests in a selective attention task at intake and post-treatment. Results indicated that the MI intervention was feasible and the subjective impact of the EEG/ERP feedback was positive. Significant group differences in percentage of cocaine positive urine screens across the study were found, favoring the MI group; 84.9% for the control group and 62.6% in the MI group. In a current study at the University of Louisville, Tato Sokhadze and his colleagues are utilizing dense-array qEEG/ERP variables and measures of behavioral performance on mental tasks (reaction time, accuracy) to explore the cognitive functions in patients with cocaine abuse/dependence diagnosis, and the recovery of these functions during bio-behavioral intervention based on an integrated neurofeedback approach (NFB, Scott–Kaiser protocol) and motivational enhancement therapy (MET) in an outpatient population. The purpose of this research is also to characterize changes in cognitive functioning associated with the success rate of three arms for cocaine addiction treatment (MET, NFB, combined MET + NFB). Prior, during, and subsequent to the above bio-behavioral therapies, individual differences in qEEG and dense-array ERP are being assessed during cognitive tasks containing drug-related and generally affective cues, and during cognitive tasks aimed to test cortical inhibitory capacity, selective attention, response error processing, and cortical functional connectivity. Preliminary data from this study were presented at the 2007 annual meeting of ISNR (Sokhadze et al. 2007b) and are being prepared for the publication. Efficacy of Alpha Theta Training The Guidelines for Evaluation of Clinical Efficacy of Psychophysiological Interventions (LaVaque et al. 2002), which have been accepted by AAPB and ISNR, specify five types of classification for the effectiveness of biofeedback procedures, ranging from “Not empirically supported” to “Efficacious and Specific.” The requirements for each classification level are summarized in brief below. A more complete description may be found in LaVaque et al. (2002). Criteria for Levels of Evidence of Efficacy Level 1:Not empirically supported. This classification is assigned to those treatments that have only been described and supported by anecdotal reports and/or case studies in non-peer reviewed journals. Level 2:Possibly efficacious. This classification is considered appropriate for those treatments that have been investigated in at least one study that had sufficient statistical power, well-identified outcome measures, but lacked randomized assignment to a control condition internal to the study. Level 3:Probably efficacious. Treatment approaches that have been evaluated and shown to produce beneficial effects in multiple observational studies, clinical studies, wait list control studies, and within-subject and between-subject replication studies merit this classification. Level 4:Efficacious. In order to be considered “efficacious,” a treatment must meet the following criteria:In a comparison with a no-treatment control group, alternative treatment group, or sham (placebo) control utilizing randomized assignment, the investigational treatment is shown to be statistically significantly superior to the control condition or the investigational treatment is equivalent to a treatment of established efficacy in a study with sufficient power to detect moderate differences;The studies have been conducted with a population treated for a specific problem, from whom inclusion criteria are delineated in a reliable, operationally defined manner;The study used valid and clearly specified outcome measures related to the problem being treated;The data are subjected to appropriate data analysis;The diagnostic and treatment variables and procedures are clearly defined in a manner that permits replication of the study by independent researchers, andThe superiority or equivalence of the investigational treatment have been shown in at least two independent studies” (LaVaque et al. 2002, p. 280). Level 5:Efficacious and Specific. To meet the criteria for this classification, the treatment needs to be demonstrated to be statistically superior to a credible sham therapy, pill, or bona fide treatment in at least two independent studies. Using these criteria and based on the studies reported to date alpha-theta training can be classified as Level 3—probably efficacious—when combined with an inpatient rehabilitative treatment modality in subjects with long standing alcohol dependency. This classification is based on the original randomized and controlled study of the Peniston Protocol (Peniston and Kulkosky 1989, 1990, 1991) and multiple observational and uncontrolled studies that preceeded (Twemlow and Bowen 1977, Twemlow et al. 1977) and followed these studies (DeBeus et al. 2002; Burkett et al. 2003; Fahrion et al. 1992; Finkelberg et al. 1996; Skok et al. 1997; Bodenhamer-Davis and Calloway 2004; Saxby and Peniston 1995; Fahrion 1995). Using these criteria and based on reported studies to date the Scott–Kaiser modification of the Peniston Protocol can also be classified as probably efficacious (Level 3) when combined with residential rehabilitation modalities in stimulant abusers. This rating is based on one controlled study of 121 subjects in which Peniston’s outcomes of both psychometric improvement and abstinence improvement were replicated (Scott et al. 2005) and one observational study of 71 subjects (Burkett et al. 2003). Alpha-theta training protocols do not completely meet the criteria for the Level 4, “efficacious” classification. Although there are sufficient studies that show statistically significant superiority of randomly assigned treatment groups to no-treatment control groups, and studies have been conducted with populations treated for a specific problem, from whom inclusion criteria are delineated in a reliable, operationally defined manner, and the studies cited use valid and clearly specified outcome measures related to the problem being treated with data subjected to appropriate data analysis, there remains the shortcoming cited by Graap and Freides (1998) for the initial reports of Peniston and Kulkosky (1989, 1990, 1991). We recall the qualifying limitations of LaVaque et al. (2002), who stated that “the diagnostic and treatment variables and procedures are not clearly defined in a manner that permits replication of the study by independent researchers” (p. 280). However, the Scott et al. (2005) report does appear to clearly delineate treatment variables and procedures. One other independent study showing the superiority of modified alpha-theta training to control condition would meet the stated criteria for a Level 4 “efficaceous” classification. To be considered Level 5 (“efficacious and specific”) modified alpha-theta training would need to be shown to be superior to sham or bona fide treatment. It has not been demonstrated that the Peniston type alpha-theta feedback is more efficacious than sham treatment (Trudeau 2000, 2005a, b; Lowe 1999; Moore and Trudeau 1998) or alternative treatment that involves meditation (Taub and Rosenfeld 1994). Clinical Considerations: Comorbidities of SUD and Implications for Individualized (qEEG-Guided) Neurofeedback There are several conditions commonly associated with addictive disorders that have known neurophysiological aberrations. The co-occurrence of alcohol and other SUD with other psychiatric disorders has been widely recognized. Co-occurrence of SUD and other psychiatric diagnosis (e.g., PTSD, antisocial personality disorder, ADHD, unipolar depression etc.) is highly prevalent (Drake and Walach 2000; Evans and Sullivan 1995; Grant et al. 2004; Jacobsen et al. 2001). Persons with co-occurring other mental disorders and SUD have a more persistent illness course and are more refractive to treatment than those without dual diagnoses (Brown et al. 1995; O’Brien et al. 2004; Schubiner et al. 2000; Swartz and Lurigio 1999). Depression occurs in approximately 30% of chronic alcoholics (Regier et al. 1990). In treatment settings, these depressed patients can present particular challenges to the clinician, as they may not respond as well to treatment as other patients, may have greater relapse, attrition, and readmission rates, and may manifest symptoms that are more severe, chronic, and refractory in nature (Sheehan 1993). Independent of other psychiatric comorbidity, ADHD alone significantly increases the risk for SUD (Biederman et al. 1995). Associated social and behavioral problems may make individuals with comorbid SUD and ADHD treatment resistant (Wilens et al. 1998). In males ages 16–23, the presence of childhood ADHD and conduct disorder is associated with non-alcohol SUD (Gittleman et al. 1985; Manuzza et al. 1989). In summary childhood ADHD associated with conduct disorder in males is an antecedent for adult non-alcohol SUD and anti-social personality disorder (Wender 1995). The incidence of ADHD in clinical SUD populations has been studied and may be as high as 50% for adults (Downey et al. 1997) and adolescents (Horner and Scheibe 1997). Adult residual ADHD is especially associated with cocaine abuse and other stimulant abuse (Levin and Kleber 1995). Monastra et al. (2005) in a white paper review of ADHD, cite positive treatment outcomes of just under 80% in treatment of ADHD with neurofeedback. Rates of PTSD occurring in persons primarily identified with or in treatment for substance abuse vary from 43% (Breslau et al. 1991) up to 59% (Triffleman et al. 1999). In a general population study, Cottler et al. (1992) reported that cocaine abusers were three times more likely to meet diagnostic criteria for PTSD compared to individuals without a SUD. Kalechstein et al. (2000) found that methamphetamine-dependent individuals are at greater risk to experience particular psychiatric symptoms. There was reported a significant dependence-by-gender effect, with methamphetamine-dependent females reporting significantly more overall posttraumatic stress symptomatology compared to females reporting no dependence, whereas males significantly differed only with respect to depression. Peniston and Kulkosky (1991) reported effective treatment of PTSD using a protocol similar to the one they employed for alcoholics. Hughes and John (1999) review the applicability of qEEG findings in SUD. They note that in numerous qEEG studies there is a consensus of increased beta relative power in alcoholism and increased alpha in cannabis and crack cocaine users. They conclude that the evidence provided by studies to date is insufficient to recommend qEEG as a routine clinical assessment tool in SUD, although it may be useful in differential diagnosis in difficult cases. A number of specific qEEG abnormalities have been described as specific to suspected neurotoxicities associated with chronic stimulant abuse. These studies (Alper et al. 1990; Noldy et al. 1994; Prichep et al. 1996; Roemer et al. 1995; Trudeau et al. 1999) based on reasonably uniform abstinence times and employing different EEG technology and analytical approaches, have produced remarkably similar findings of alpha relative amplitude excess with delta relative amplitude deficit that is striking. Excess alpha amplitude with slowing of alpha frequency associated with chronic cannabis abuse has been reported (Struve et al. 1998). As noted, Scott and Kaiser (1998) describe combining a protocol for attentional training (beta reward) with alpha-theta training in a population of subjects whose primary drugs of abuse were stimulants and who had features of ADHD. It may make good sense clinically to consider specific neurotherapy treatment of these disorders either in place of or preceding alpha-theta therapy, similar to the Scott–Kaiser approach. Second, applicable neurotherapy approaches are attractive alternative therapies for coexisting or underlying conditions in SUD clients who have high-risk behaviors for medication treatment, such as overdosing, abuse, or poor compliance. While there are no published systematic studies of neurotherapy treatment of co-occurring depression, TBI, ADHD, PTSD, or drug neurotoxicity on the course and outcome of addictive disorders, several recent reports of neurotherapy for addictions based on qEEG findings, which in turn may be related to comorbidities, have been presented. Basically, this technique involves the use of qEEG to identify patterns of EEG that deviate from standardized norms, and individualized EEG biofeedback protocols to correct them (Romano-Micha 2003). DeBeus et al. (2002) are presently conducting a randomized controlled study of neurotherapy for SUD that examines the difference between a qEEG-based treatment, a research-based (Scott–Peniston) treatment, and a wait-list control for chemically dependent outpatients. Preliminary results are promising. While historically, alpha-theta training has been the accepted approach in treating chemical dependency, this study suggests qEEG-based training is a viable alternative, demonstrating similar outcomes for personality change and abstinence rates. Future directions include determination of those likely to benefit from one of the particular treatments or a combination of the two and analysis of long-term abstinence rates. Gurnee (2004) has presented data on a series of 100 sequential participants with SUD who were treated by qEEG-based neurotherapy, with marked heterogeneity of qEEG subtypes and corresponding symptom complexes. In this clinically derived scheme, qEEGs that deviate from normative databases, mainly with excess alpha amplitude, are associated more often with depression and ADD. Those with deficient alpha amplitude are associated with anxiety, insomnia, and alcohol/drug abuse. Beta excess amplitude is associated with anxiety, insomnia, and alcohol/drug abuse. Central abnormalities are interpreted as mesial frontal dysfunction and are associated with anxiety, rumination, and obsessive compulsive symptoms. The therapeutic approach is to base neurotherapy on correcting identified qEEG abnormalities, i.e., train beta excess amplitude down when present, while monitoring symptoms. Tentative findings suggest that qEEG variables may be used to predict those alcoholics and drug abusers most at risk for relapse. Winterer et al. (1998) were able to predict relapse among chronic alcoholics with 83–85% success, significantly outperforming prediction from clinical variables. Although they found more desynchronized (less alpha and theta and more beta activity) over frontal areas in alcoholics in general, those individuals who relapsed displayed even more of this activity. Bauer (2001) obtained EEG data on alcohol, cocaine or opioid dependent patients after 1–5 months of sobriety. Those who had relapsed by 6 months later were also characterized by increased beta (19.5–39.8 Hz) activity relative to those maintaining abstinence. Relative beta power was superior to severity of the alcoholism, depression level, antisocial personality disorder, childhood conduct problems, family history, or age as predictors, and was unaffected by the substance of abuse. The EEG differences between relapse-prone and abstinence-prone groups were found to be related to the interaction of two premorbid factors: childhood conduct disorder and paternal alcoholism. These findings receive further support from Bauer (1993) and from Prichep et al. (1996) who also found that beta activity was predictive of treatment failure. They found two clusters among cocaine addicts: One had more severe damage (alpha) and tended to remain in treatment. Those with less severe alpha excess and more beta activity tended to leave treatment. They also discovered that dropouts could not be determined from the presence of anxiety or depression or demographic variables. Treatment of patients with substance abuse disorder by neurofeedback may become more complicated when patients present various psychiatric conditions. When addiction is comorbid with ADHD it is suggested that SMR (or beta increase, theta decrease) training should be conducted to address the related ADHD disorder first (Biederman et al. 1997). Applicability of neurofeedback methods to treat anxiety and affective disorders is reviewed by Hammond (2006). Peniston and Kulkosky (1990) describe personality normalization in alcoholics treated with EEG biofeedback. Alpha-theta feedback has also been reported as efficacious in alcoholics with depressive symptoms (Saxby and Peniston 1995). There are only a few case studies on the efficacy of neurofeedback for treating generalized anxiety disorder (Vanathy et al. 1998) and PTSD (Huang-Storms et al. 2006; Graap et al. 1997). Alpha-theta feedback has been described as efficacious in post-combat PTSD (Peniston and Kulkosky 1991; Peniston et al. 1993). However, additional research needs to be completed to determine the clinical outcome and efficacy of bio-behavioral treatment based on brain wave self-regulation in addiction disorders that are comorbid with various anxiety disorders and PTSD. Clinical Considerations: Cognitive-Behavioral and Neurofeedback Treatment in Substance Use Disorders Because of its chronic nature, long-term treatment for SUD is usually necessary (Crits-Christoph et al. 1997, 1999). Effective agonist and antagonist pharmacotherapies as well as symptomatic treatments exist for opioid dependence, but neither agonists nor antagonists have been approved as uniquely effective for treatment of stimulant abuse or dependence (Grabowski et al. 2004). There is no current evidence supporting the clinical use of carbamazepine (Tegretol), antidepressants, dopamine agonists (drugs commonly used to treat Parkinson’s and Restless Leg Syndrome), disulfiram (Antabuse), mazindol (an experimental anorectic), phenytoin (Dilantin), nimodipine (Nimotop), lithium and other pharmacological agents in the treatment of cocaine dependence (de Lima et al. 2002; Venneman et al. 2006). Because no proven effective pharmacological interventions are available for cocaine addiction or for methamphetamine addiction, treatment of stimulant addiction has to rely on existing cognitive-behavioral therapies (CBT) or CBT combined with other biobehavioral approaches (Van den Brink and van Ree 2003). According to Volkow et al. (2004) successful strategies for behavioral treatment in drug addiction may include (1) interventions aimed to decrease the reward value of the drug and simultaneously increase values of natural reinforcement; (2) approaches aimed to change stereotype conditioned drug-seeking behaviors; and (3) methods to train and strengthen frontal inhibitory control. Because stressful events can result in relapse to drug taking behavior (Koob and Le Moal 2001), an adjunct treatment strategy is to interfere with the neurobiological responses to stress (Goeders 2003; Koob and Le Moal 2001). Treatment of comorbid mental conditions may also require the concurrent treatment of drug addiction. In some cases, however, comorbid drug addiction may result from attempts to alleviate the psychiatric disorder through self-medication (i.e., co-occurring cocaine use and ADHD and/or heroin addiction co-occurring with PTSD). In other cases, severity of a psychiatric disorder symptom may increase as a result of drug abuse (Volkow et al. 2003, 2004). In patients with drug abuse arising from an attempt to self-medicate (Khantzian 1985, 1997), treatment of the comorbid mental disorder may help prevent abuse. For instance, treatment of the preexisting condition of ADHD may prevent cocaine abuse (Biederman et al. 1995, 1997). In some cases though the persistent qEEG abnormalities associated with chronic SUD may happen to be independent from ADHD clinical status (Trudeau et al. 1999). The co-occurrence of ADHD and SUD has received considerable attention in the recent clinical and scientific literature (Davids et al. 2005). These two disorders are often linked to one another. Because the core symptoms of ADHD may be mimicked by the effects of psychoactive drugs, it is difficult to diagnose one disorder in the presence of the other (Davids et al. 2005). ADHD has been found to be associated with an earlier onset of SUD (Horner and Scheibe 1997). It is generally assumed that untreated ADHD is a risk factor for SUD development (Biederman et al. 1997, 1998; Manuzza et al. 1998; Trudeau 2005a, b). In a case of comorbidity in which the use of drugs antecedes a mental disease (e.g., substance-induced anxiety disorder, DSM-IV-TR, APA 2000) or is not driven by self-medication strategies, the simultaneous treatment of both psychiatric conditions may be required. In this situation, treatment could be guided by the two following concepts: (1) Behavioral interventions to activate and strengthen circuits involved in inhibitory control, such as bio-behavioral self-regulation training, may increase successful abstinence from drug taking. (2) Considering the important role of cognitive and emotional processes involved in the predisposition for drug abuse, the development of non-pharmacological interventions (e.g., CBT, stress management, neurofeedback) is a feasible strategy. Directions for Further Research Specific patterns of qEEG abnormality associated with specific substance use toxicity such as those found in stimulant abuse or alcohol abuse or with comorbidities such as ADHD (Chabot and Serfontein 1996), PTSD (Huang-Storms et al. 2006) or TBI (Thatcher et al. 1989) suggest underlying brain pathologies that might be amenable to EEG biofeedback that is tailored to the person. These approaches would likely be individualized rather than protocol based, and would be used independently or in conjunction with classic alpha-theta training. By way of example, these could include protocols specific to the qEEG abnormality, such as frontal delta reward to correct the frontal delta deficit in cocaine abuse that Alper (1999) hypothesizes may be related to cocaine sensitization and changes in dopamine transmission. To our knowledge this has never been studied and is clearly a research (not a clinical) recommendation. The qEEG patterns and abnormalities depend significantly on whether the subject is still currently using, the chronicity of use, and the current stage of withdrawal or protracted abstinence. A neurofeedback protocol selected for an individual client with SUD should be directly related to the level of current substance use or abstinence, especially in such classes of drugs as heroin, where the withdrawal syndrome results in substantial physiological manifestations including transient qEEG changes. Even though there are no reported systematic studies of EEG biofeedback treatment of commonly occurring comorbidities of SUD, it makes sense that clinical EEG biofeedback treatment study protocols consider the presence of ADHD, TBI, depression, and drug-associated neurotoxicity. This approach may improve outcome, especially in conventional treatment resistant participants. Unfortunately, only a few large-scale studies of neurofeedback in addictive disorders have been reported in the literature. Most, if not all of the recommendations previously made regarding further research (Trudeau 2000, 2005b) have yet to be implemented. These recommendations are summarized as follows.Studies require external, systematic replicability of brain wave feedback methods and results in diverse populations that include various control and alternative treatment conditions wherein the groups are matched on key dimensions.Details need to be given regarding the equipment that was used and the associated technical specifications (e.g., details about amplification, filtering, spectral extraction, windowing, and other pertinent information) needed by neurofeedback specialists for replication and comparison.The essential components and durations for brain wave feedback required for therapeutic advantage need to be stated, including double-blinded studies that control for all other possible therapeutic effects.Open clinical trials that investigate efficacy of the types of protocols used for ADHD, PTSD, depression, and TBI remediation with SUD subjects comorbid for those conditions need to be reported.Open clinical trials that assess the efficacy of EEG biofeedback in addressing the specific qEEG changes of chronic alcohol, heroin, cannabis and stimulant abuse need to be reported.The physiological and psychological processes of the therapeutic effects of EEG biofeedback, including studies of qEEG and ERP changes, need to be investigated and reported.Studies need to adhere to clearly defined outcome measures that have established reliability and validity. Other important recommendations for future development of the field are listed below:The availability of an increased number of channels for EEG and ERP recording (e.g., higher spatial sampling rate) makes it possible to better localize the source of brain activity. More focused research of this type seems warranted.There are several specific functional diagnostic tools from the cognitive neuroscience arsenal that are very specific for testing addictive disorders. Those that may be especially valuable include cue reactivity tests using qEEG and ERP measures. Cue reactivity is a very sensate test of motivational relevance of drug-related items (Carter and Tiffany 1999) that can be detected using EEG methods.In addition to using more traditional neurocognitive tests (TOVA, IVA+, etc.) that are commonly included in neurofeedback research (e.g., in particular in studies on effectiveness of neurotherapy in ADHD treatment) there may be value in incorporating standardized tests with EEG/ERP recording to assess executive functions in addicts. Tests that warrant mention are the Continuous Performance Test (Go-NoGo task), Stroop test, Eriksen flanker test, etc. Some of these tests are sufficiently sensitive for assessing recovery of cortical inhibition function commonly known to be impaired in patients with SUD.Testing emotional reactivity and responsiveness in addiction is another important domain where qEEG and ERP methods may help to obtain more effective evaluation of the affective state of recovering addicts. In future neurofeedback treatment for SUD attempts should be made to integrate neurotherapy with other well known behavioral interventions for drug abuse, such as cognitive-behavioral therapy (CBT) and motivation enhancement therapy (MET, Miller and Rollnick 2002). As a population, drug addicts are very difficult to treat, characterized by a low motivation to change their drug habit and a reluctance to enter inpatient treatment. CBT and MET are powerful psychotherapeutic interventions that can help to bring about rapid commitment to change addictive behaviors. These behavioral therapies are especially useful for enhancing compliance with drug-dependent individuals and facilitating their neurofeedback treatment engagement. Neurofeedback may be among the most promising biofeedback modalities for the treatment of adolescents with addictive disorders because of the neuroplasticity potential of the adolescent brain. While there is little work available on the prevention and treatment of SUD in adolescents utilizing neurotherapy, there is no reason to suspect that the approaches used in adults would not be applicable in SUD adolescents (Trudeau 2005b). The EEG biofeedback treatment of ADHD may be important in prevention for children and adolescents at risk for developing SUD. It may be possible that EEG biofeedback therapy of childhood ADHD may result in a decrease in later life SUD (Wilens et al. 1998). This remains speculative, as there have been no reported studies of the effects of neurofeedback treatment on prevention of SUD to date. There are several important applications of the neurofeedback protocols for enhancement of cognitive performance in healthy subjects (reviewed in Vernon 2005). This promising new line of neurofeedback-based cognitive neuroscience research (Barnea et al. 2005; Egner and Gruzelier 2001, 2003, 2004a, b; Egner et al. 2004; Vernon et al. 2003) has significant potential to elucidate neurobiological mechanisms explaining how neurofeedback training may alter and enhance cognition and behavioral performance in patients with SUD as well. Drugs of abuse can impair cognitive, emotional and motivational processes. More qEEG and cognitive ERP research is needed to characterize the chronic and residual effects of drugs on attention, emotion, memory, and overall behavioral performance. More research is needed also to relate cognitive functionality measures to clinical outcome (e.g., relapse rate, drug screens, psychiatric status, etc.). Such qEEG/ERP studies may facilitate the translation of clinical neurophysiology research data into routine practical tools for assessment of functional recovery both in alcoholism and addiction treatment clinics. We believe that administration of some of above described qEEG assessments at the pre-treatment baseline might provide useful predictors of clinical outcome and relapse risk. Incorporation of cognitive tests with EEG and ERP (e.g., P300) measures into cognitive-behavioral and neurofeedback based interventions may have significant potential for identifying whether certain qEEG/ERP measures can be used as psychophysiological markers of treatment progress (and/or relapse vulnerability), and also may provide useful information in planning cognitive-behavioral and neurotherapy treatment when substance abuse is comorbid with a mental disorder. With the advances made in the last several years, it is hoped that continued interest will be generated to further study brainwave biofeedback treatment of addictive disorders. Effectiveness in certain “hard to treat” populations (conventional treatment resistant alcoholics, crack cocaine addicts, cognitively impaired substance abusers) is promising. The prospect of an effective medication free, neurophysiologic, and self-actualizing treatment for a substance based, brain impaired, and self-defeating disorder such as SUD is attractive.

Psychopharmacologia-4-1-2362139

Expression of amphetamine sensitization is associated with recruitment of a reactive neuronal population in the nucleus accumbens core

Rationale Repeated exposure to psychostimulant drugs causes a long-lasting increase in the psychomotor and reinforcing effects of these drugs and an array of neuroadaptations. One such alteration is a hypersensitivity of striatal activity such that a low dose of amphetamine in sensitized animals produces dorsal striatal activation patterns similar to acute treatment with a high dose of amphetamine. Introduction Repeated exposure to psychostimulant drugs causes a long-lasting enhancement of certain behavioral responses to the drug, such as psychomotor activity and stereotypy, and behaviors related to incentive motivation, a process termed behavioral sensitization (Stewart and Badiani 1993). Behavioral sensitization is known to be associated with long-lasting functional changes within limbic corticostriatal systems (Pierce and Kalivas 1997; Robinson and Kolb 2004; Vanderschuren and Kalivas 2000). These systems comprise functionally and anatomically heterogeneous areas with a fine-grained specificity of anatomical projections connecting the divisions within the dorsal striatum, ventral striatum, and prefrontal cortex (Groenewegen et al. 1997; Voorn et al. 2004). This anatomical and functional heterogeneity is of potential importance to the roles of these areas in sensitization. Within the dorsal striatum, subareas termed patches (or striosomes) show more reactivity to amphetamine than the surrounding matrix areas in sensitized animals (Canales and Graybiel 2000; Vanderschuren et al. 2002). Our previous studies demonstrated that this pattern of neuronal reactivity is also seen in acutely challenged animals with the important difference that sensitized animals show preferential activation in patches at much lower doses of amphetamine than those required to produce this type of differentiation in activation in drug-naive animals (Vanderschuren et al. 2002). An imbalance in patch–matrix activation has been suggested to underlie stereotyped behavior (Canales and Graybiel 2000). Because our drug treatment regimen caused robust locomotor sensitization, which is incompatible with profound stereotypy, we hypothesized that hyperreactivity of patch compartments is not sufficient to produce stereotypy (Vanderschuren et al. 2002). To extend our previous findings, we ran new experiments using the same doses and regimen of amphetamine administration previously used and measured locomotor activity, stereotypy, grooming, and rearing to establish whether our drug treatment regimen, which causes hyperreactivity of dorsal striatal patches, produces stereotypy. The ventral striatum, specifically the nucleus accumbens (Acb), and the prefrontal cortex (PFC) are both involved in behavioral sensitization (Pierce and Kalivas 1997; Vanderschuren and Kalivas 2000), an involvement which has received particular attention because of the important role that both areas play in appetitive and consummatory properties of both natural and drug rewards (Everitt and Wolf 2002; Robbins and Everitt 2002; Salamone et al. 2003; Volkow and Li 2004). There are functional differences between subregions within both the Acb, i.e., core and shell, and within subregions of the PFC (Cardinal et al. 2002; Robbins and Everitt 2002). However, there is presently inconclusive evidence on the respective roles that the subregions play during the expression of behavioral sensitization to psychostimulants. Studies using neurochemistry, lesions, cellular activity markers, and study of morphological changes have suggested exclusive roles for the core (Cadoni et al. 2000; Li et al. 2004; Phillips et al. 2003) or the shell (Filip and Siwanowicz 2001; Hsieh et al. 2002; Pierce and Kalivas 1995; Todtenkopf et al. 2002a). Other studies, including previous work from our own laboratory, suggest a lack of sensitization of accumbens activity all together (Ostrander et al. 2003; Vanderschuren et al. 2002). The medial PFC, and particularly the prelimbic area, has been implicated in induction of psychostimulant sensitization (Tzschentke and Schmidt 1998, 2000), although conflicting results have been found for cocaine- vs. amphetamine-induced sensitization (Tzschentke and Schmidt 2000). The dorsomedial prefrontal cortex has been shown to be involved in the expression of sensitization (Pierce et al. 1998). However, the roles of the orbital and lateral areas in psychostimulant sensitization remain to be investigated. Thus, clarification of the specific roles of the Acb and PFC subregions is needed. To study the activation of the Acb and PFC in detail during the expression of behavioral sensitization, we examined levels of c-fos-like proteins (henceforth c-fos) in detail in the subregions of the Acb and PFC of the rat after an amphetamine challenge in behaviorally sensitized rats. Materials and methods Animals and drug treatments All experiments were approved by the Animal Ethics Committee of the Vrije Universiteit and were conducted in agreement with Dutch laws (Wet op de Dierproeven 1996) and European regulations (Guideline 86/609/EEC). A total of 48 male Wistar rats weighing 180–200 g upon arrival in the laboratory (as in, e.g., De Vries et al. 1996; Vanderschuren et al. 1999a, b, 2002) were housed in Macrolon cages in groups of two animals per cage under controlled laboratory conditions (lights on 0700 to 1900 hours). Food and water were available ad libitum. Drug treatment started after an acclimatization period of at least 1 week. Animals were briefly handled during the 2 days before all injections. In the acute amphetamine experiments, animals were injected with either saline or 1, 2.5, or 5 mg/kg d-amphetamine sulfate (O.P.G., Utrecht, The Netherlands; n = 4 per dose). Sensitization regimens were according to a protocol previously established to produce locomotor sensitization in our laboratory (De Vries et al. 1996; Vanderschuren et al. 1999a, b), and doses were the same as used previously in a c-fos study in our laboratory (Vanderschuren et al. 2002). Animals received once daily injections for five consecutive days of 2.5 mg/kg d-amphetamine sulfate or saline in the home cage (pretreatment phase). Two weeks post treatment, half of the animals from each pretreatment group were given challenge injections of 1 mg/kg d-amphetamine sulfate while the other half was injected with saline. This gave a total of four experimental groups, n = 8 per group: amphetamine-pretreated, amphetamine-challenged (AA); amphetamine-pretreated, saline-challenged (AS); saline-pretreated, amphetamine-challenged (SA); and saline-pretreated, saline-challenged (SS). Locomotor activity quantification and behavioral scoring All injections for the acute experiments and challenge injections for the sensitization experiments took place in our locomotor activity setup. On the challenge injection day, animals were first placed in the Perspex cages (length × width × height = 40 × 40 × 35 cm) in which locomotor activity was measured and allowed to acclimate for 2 h. After that period, challenge injections were administered and horizontal activity was measured in 10-min blocks for 90 min using a video tracking system (EthoVision, Noldus Information Technology B.V., Wageningen, The Netherlands), which determined the position of the animal five times per second. Behavior of the animals was also videotaped and scored afterwards for three mutually exclusive categories: grooming (rubbing two paws over head and/or body), rearing (both front paws off of the ground but not grooming), or stereotypical behavior (repeated movements without horizontal movement, e.g., head shaking). Behavior was scored for 5 min of every 15 min, giving a total of seven measurements per animal, by an observer unaware of the treatment of the animals using a time-sampling program written in PC Basic. Time spent performing each behavior was expressed as the percentage of total time for each 5-min block. Replication scoring several months after initial scoring produced results identical to initial observations, demonstrating the reliability of our scoring procedures. c-fos Immunocytochemistry At 90 min after the challenge injection, animals were decapitated, the brains were snap-frozen in isopentane and stored at −80°C until use. Sections of 20 μm were cut on a cryostat and mounted onto coated slides (SuperFrost Plus) which were dried and stored at −80°C until use. For visualization of c-fos, sections were defrosted and fixed in a 4% paraformaldehyde solution in phosphate-buffered saline (PBS; 0.1 M, pH 7.4). Sections were washed with Tris-buffered saline (TBS, 0.1 M, pH 7.4) then incubated with primary antibody against c-fos (1:1,800, Oncogene Research, Burlington, MA, USA) in TBS with 0.5% Triton-X and 0.5% bovine serum albumin (TBS-TX-BSA) overnight at 4°C. After rinsing with TBS, endogenous peroxidase activity was removed by incubation of sections in a 1% hydrogen peroxide solution for 15 min. Sections were rinsed with TBS, then incubated in biotinylated goat antirabbit antibody (1:100, Dako, Denmark) in TBS-TX-BSA for 1 h, washed in TBS and incubated in avidin–biotin complex with horseradish peroxidase (HRP) (1:100, Vector Laboratories, Burlingame, CA, USA) for 1 h. Sections were rinsed in Tris–HCl then incubated in 3′3-diaminobenzidine (DAB; Sigma Chemical, 0.05% DAB in Tris–HCl) and rinsed in Tris–HCl. Sections containing the prefrontal cortex were incubated with Hoechst 33258 (1:2,000; Molecular Probes, Eugene, OR, USA), a fluorescent nuclear stain used to visualize cytoarchitecture. Sections were dried and finally coverslipped with Merckoglas (Merck, Darmstadt, Germany). Histological quantification Quantification of c-fos immunopositive nuclei was performed using an MCID Elite imaging system (Imaging Research, Ontario, Canada). Images of the nucleus accumbens in the c-fos DAB immunostained sections were digitized using an objective magnification of ×10 on a Leica DM/RBE photomicroscope with a Xillix MicroImager digital camera (1,280 × 1,024 pixels). Digitized images were combined so that the core and the shell areas were included, using the MCID tiling tool. Three (in some cases, two) sections per rat were chosen for quantification at the rostral–caudal levels in which inputs from the prefrontal cortex, thalamus, and amygdala have been particularly well characterized (Wright and Groenewegen 1995). The prefrontal cortex was digitized in the same fashion with the exception that color digital images were acquired using a Sony HAD camera (Sony DXC 950v, 640 × 512 pixels) of both the DAB staining and the epifluorescence of the Hoechst 33258 staining. The core and shell areas of the nucleus accumbens were delineated on the basis of atlas drawings from sections stained for calbindin (Jongen-Relo et al. 1993). The prefrontal cortex was delineated into prelimbic, infralimbic, orbital, and lateral areas on the basis of cytoarchitectonic criteria visible in the Hoechst 33258 staining. The c-fos immunopositive nuclei in the nucleus accumbens were segregated from background staining levels using several point operators and spatial filters combined in an algorithm designed to detect local changes in the relative optical density (ROD). Briefly, images underwent histogram equalization and smoothing (low-pass filter, kernel size 7 × 7). The unfiltered image was subtracted from the smoothed image, followed by a series of steps to optimize the processed image and make it a suitable measuring template for detecting objects the size and shape of c-fos immunopositive nuclei. This algorithm was preferred over ROD thresholding because it does not involve an observer-dependent operation. The number of nuclei counted was corrected with a factor indicating approximate size of a c-fos immunopositive nucleus, thus preventing two groups of stained pixels touching one another in the image being mistakenly counted as one nucleus. The results of all counting were expressed as the number of nuclei per surface area (mm2). Integrated ROD for each segmented immunopositive nucleus was determined. Segregation of c-fos positive nuclei in the color-digitized images was performed in a similar fashion. Subsequently, we set out to compare c-fos positive density (i.e., the number of cells per surface area) in a manner that accounts for labeling intensity: dark, light, or midrange. First, histograms of c-fos nuclei ROD values (for black and white images) or intensity values (for color images) were constructed for each brain area for each treatment group and qualitatively compared. These histograms were used to determine the value of the 33rd and 66th percentile optical density within the SS group. Based on these values, all nuclei from all animals for each area were binned as “light” (ROD values under the 33rd percentile of the SS group), “midrange” (ROD values between the 33rd and 66th percentile of the SS group) or “dark” (ROD values above the 66th percentile of the SS group). The number of immunopositive nuclei per bin was counted per rat, and the group averages were determined from the rat averages. For the prefrontal cortex, the same technique was used, but as these areas were digitized in color, intensity was used to bin rather than optical density. An increase in the number of nuclei in the “dark” bin would signify a rightward shift in the histograms, indicating that the increased cellular activity measured was primarily the result of more c-fos expression in the same group of neurons. An increase in the number of nuclei in the “midrange” bin would signify an upward shift in the histograms, indicating that the increase in the total number of nuclei measured was the result of the addition of a new group of nuclei to the cellular response (Fig. 1). Fig. 1Theoretical conceptualization of shifts in the ROD histograms. The total number of c-fos positive nuclei per mm2 counted is represented as the area under the curves to the right of the detection level, indicated as an arrow on the x-axis. Curve 1 indicates the control group; vertical lines indicate the ROD used to separate the neurons into light, midrange, and dark. An increase in the treatment group compared to the control group in the total number of c-fos positive nuclei per mm2 could be the result of an increase in the frequency of c-fos positive nuclei, indicated by an increase in frequency in curve 2, causing an increase in the number of c-fos positive nuclei in the midrange. This would indicate that a new group of neurons is being recruited in the c-fos response, represented by the difference between curves 1 and 2 in frequency. Alternatively, the same number of neurons could be active, but expressing more c-fos protein. This would cause a rightward shift in the curve (curve 3) and allowing more c-fos positive nuclei to come above the detection level and causing more c-fos positive nuclei to be measured in the dark range. A similar line of reasoning can be followed for other options, for instance a leftward shift in the case of reduced levels of c-fos protein (not illustrated) The dorsal striatum of the AA group was qualitatively inspected by two observers, both blind to the experimental conditions. The distribution pattern of c-fos positive nuclei was described and compared with that in a series of closely adjacent sections from the same animals stained immunocytochemically for the μ-opioid receptor to visualize striatal patches (Vanderschuren et al. 2002). Statistics For the quantification of c-fos immunoreactivity, the experimental groups were compared for effects of pretreatment and challenge (saline vs. amphetamine) and for interactions between these effects using a two-way ANOVA test followed by a Tukey post hoc test. For locomotor activity and behavioral scores, a repeated-measures ANOVA was conducted using time as within-subjects factor and followed by a Tukey post hoc test. Results Behavioral results For all behavioral measures, time was included as a within-subjects measure in a repeated-measures ANOVA followed by a post hoc test when significant time × pretreatment or time × challenge interactions were found. For brevity, only the most relevant of the results of these post hoc test results are described in this section and other results are presented in the figures. Behavioral measures: responses to acute amphetamine Locomotor activity was dose-dependently altered in animals treated acutely with amphetamine (Fig. 2a; main effect dose F(3,10) = 43.914; p < 0.001). Saline-treated animals showed generally low activity levels, averaging a total of 1,335 ± 514 cm traveled during the 90-min period. The groups treated with 1 mg/kg (14,045 ± 1,836 cm) and 5 mg/kg amphetamine (14,906 ± 2,321 cm) showed comparable levels of locomotor activity, whereas the highest locomotor response to amphetamine was observed in the 2.5 mg/kg group (25,543 ± 1,704 cm). All amphetamine-treated groups differed from saline in the post hoc test. Acute amphetamine caused a significant increase in rearing (F(3,12) = 7.971; p < 0.005; Fig. 2b), although not dose-dependently, as all doses differed significantly from the saline-treated group in post hoc testing, but there were no significant differences among the amphetamine-treated groups (average percent time spent rearing, saline = 2.1 ± 0.75; 1 mg/kg = 15.2 ± 2.7; 2.5 mg/kg = 20.4 ± 3.5; 5 mg/kg = 16.8 ± 7.7). No significant effect of amphetamine on grooming was seen (main effect of dose: F(3,12) = 3.098, n.s.; Fig. 2b). Stereotypy was exclusively observed in the group treated with 5 mg/kg of amphetamine (average percent time spent in stereotypical behavior, 5 mg/kg = 28.4 ± 16.5; all other groups average = 0 ± 0; Fig. 2b). A clear main effect of dose was seen (F(3,12) = 11.724; p < 0.005). The 5-mg/kg group differed significantly from all other groups in the post hoc test. Fig. 2Locomotor activity (a) and behavioral observations (b) for animals treated acutely with saline or 1, 2.5, or 5 mg/kg amphetamine. The highest levels of locomotor activity were seen in the 2.5-mg/kg treated group, which differed significantly from all other groups in the post hoc tests. Stereotypical behavior was observed in the animals treated with 5 mg/kg amphetamine, which differed significantly from all other groups, but no stereotypy was seen in any of the other treatment groups. No significant differences were seen in grooming behavior (b). All amphetamine-treated groups showed significantly more rearing than the saline-treated group (b), but the amphetamine treatment groups did not differ from one another. Bars indicate group averages, error bars represent SEM, n = 4 per group. Asterisks indicate significant difference in post hoc testing (p < 0.05) Behavioral measures: responses to challenge after repeated amphetamine Pretreatment with amphetamine caused a clear-cut augmentation of the locomotor response to amphetamine, as illustrated in Fig. 3a. Overall across the 90-min period, the AA group of amphetamine-pretreated rats that were challenged with amphetamine on the test day showed a 70% increase in activity compared to the SA group, which was pretreated with saline and challenged with amphetamine (total traveled distance during the 90-min test period = 20,432 ± 3,403 cm by AA compared to 11,924 ± 2,149 cm by SA). Both groups that were challenged with saline on the test day (AS and SS, pretreated with amphetamine and saline, respectively) showed considerably less locomotor activity, showing approximately 10% of the activity displayed by the AA group (total traveled distance during 90 min, SS = 1,770 ± 493 cm; AS = 2,196 ± 623 cm). These differences were reflected in significant main effects of pretreatment (F(1,20) = 6.15; p < 0.05) and challenge (F(1,20) = 62.07; p < 0.001), and a significant pretreatment × challenge interaction (F(1,20) = 5.029; p < 0.05). Fig. 3Locomotor activity (a) and behavioral observations (b) in amphetamine-pretreated and amphetamine-challenged animals and control groups. Bars represent group average of locomotor activity (a) or percentage of time spent displaying each mutually exclusive category of behavior (grooming, rearing or stereotypical behavior; b). No bars are present representing stereotypical behavior because this behavior was not observed in any animals included in this experiment. Error bars indicate SEM, n = 6 per group. AA amphetamine-pretreated, amphetamine-challenged; SA saline-pretreated, amphetamine-challenged; AS amphetamine-pretreated, saline-challenged; SS saline-pretreated, saline-challenged. Asterisks indicate significant difference in post hoc testing (p < 0.05) As shown in Fig. 3b, the evaluation of the percentage of time spent rearing revealed an increase in rearing in the AA group compared to all other groups (average over 90 min, AA = 14.3 ± 2.9%, AS = 1.84 ± 0.64%, SA = 7.09 ± 1.2%, SS = 3.34 ± 0.92%). This was confirmed by a significant main effect of challenge (F(1,20) = 27.52; p < 0.001) and pretreatment × challenge interaction (F(1,20) = 7.96; p < 0.05). No main effect of pretreatment was observed for this parameter (F(1,20) = 3.41; n.s.). Percentages of time spent grooming were generally low with average scores under 4% in all groups (average over 90 min, AA = 1.66 ± 0.64, AS = 3.54 ± 1.97, SA = 1.78 ± 0.86, SS = 3.97 ± 0.87; Fig. 3b). Despite the somewhat higher averages in the saline-challenged groups compared to the amphetamine-challenged groups, no significant effects of pretreatment, challenge, or interaction between the two were seen on grooming (pretreatment: F(1,20) = 0.64, n.s.; challenge: F(1,20) = 3.42, n.s.; pretreatment × challenge: F(1,20) = 0.21, n.s.). No stereotyped behavior was observed in any animal in any of the groups included in the repeated amphetamine experiments during any of the time periods scored (Fig. 3b). Cellular reactivity results In the nucleus accumbens and prefrontal cortex, the quantification of c-fos positive nuclei was tested for statistical significance by two-way ANOVAs using pretreatment and challenge as factors, followed by a Tukey HSD post hoc. For brevity, results of these post hoc tests are presented in the figures and in Table 1. Table 1Mean±SEM density (number of cells per mm2) of c-fos immunoreactive nuclei in light, dark, and midranges of cellular staining intensity Treatment groupAAASSASSNucleus accumbensCoreLight21.4 ± 6.917.6 ± 4.217.9 ± 3.221.2 ± 4.2Midrange§34.2 ± 2.8&19.0 ± 3.2#20.5 ± 3.722.1 ± 5.1Dark¶/****94.5 ± 14.1&/+35.6 ± 4.6#61.2 ± 12.424.1 ± 3.6#ShellLight28.8 ± 8.726.9 ± 7.328.8 ± 5.938.3 ± 6.4Midrange42.3 ± 3.241.7 ± 9.234.5 ± 6.640.3 ± 6.6Dark*96.1 ± 17.163.6 ± 9.286.5 ± 20.744.5 ± 7.6Prefrontal cortexPrelimbicLight48.2 ± 5.034.9 ± 6.739.3 ± 5.432.2 ± 3.8Midrange**50.0 ± 6.931.0 ± 6.046.4 ± 7.027.7 ± 6.3Dark***74.3 ± 12.432.0 ± 6.7%82.28 ± 18.8+/&28.7 ± 7.9%InfralimbicLight37.6 ± 3.833.1 ± 5.830.4 ± 5.629.9 ± 5.1Midrange46.7 ± 8.334.9 ± 7.835.0 ± 7.928.3 ± 5.4Dark***78.1 ± 13.7+40.7 ± 7.170.3 ± 14.530.2 ± 6.9#OrbitalLight27.8 ± 3.624.7 ± 6.123.0 ± 4.321.0 ± 4.4Midrange**61.6 ± 8.5+32.6 ± 7.745.0 ± 8.430.7 ± 5.9#Dark****129.7 ± 21.7&/+32.6 ± 8.6#120.4 ± 24.2+26.3 ± 5.7#/%LateralLight*22.8 ± 2.314.8 ± 10.018.1 ± 2.814.4 ± 3.3Midrange*39.9 ± 5.219.3 ± 5.127.4 ± 7.618.4 ± 3.7Dark****69.5 ± 11.6&/+18.3 ± 4.4#65.1 ± 15.6+16.7 ± 4.6#/%Results of ANOVA testing are shown in the column indicating ROD range as follows: paragraph mark significant main effect of pretreatment, asterisk significant main effect of challenge, section mark significant pretreatment × challenge interaction; one symbol: p < 0.05, two symbols: p < 0.01, three symbols: p < 0.005, four symbols: p < 0.001. Results of post hoc testing are indicated in the cells containing the value of the average number of nuclei as follows: plus sign significantly different from SS, percent sign significantly different from SA, ampersand significantly different from AS, number sign significantly different from AA. Sensitization alters distribution of reactivity in dorsal striatum In the dorsal striatum, visual inspection of the sections from animals both pretreated and challenged with amphetamine (AA group) showed overall more c-fos positive nuclei compared to the AS group, and a differential distribution throughout the area (Fig. 4). c-fos positive nuclei were more abundant medially than laterally. Within the general distribution, heterogeneity of reactivity was seen conforming to patterns previously observed in patches in sensitized animals. Comparison of c-fos staining pattern with μ-opioid stained patches in sections from another series from the same animals confirmed the location of high concentrations of nuclei in μ-opioid stained patches. Fig. 4Digital micrographs of c-fos immunopositive nuclei in the dorsal striatum of representative animals from groups pretreated with amphetamine and challenged with either saline (a, group AS) or amphetamine (b, group AA). Scale bars indicate 100 μm, arrows point to c-fos immunopositive nuclei. Note the higher number of c-fos positive nuclei in the AA animal compared to the SA animal, and the inhomogeneous distribution of nuclei in the AA animal Cellular reactivity sensitizes in nucleus accumbens core, but not accumbens shell or prefrontal cortex Within the nucleus accumbens, the core and shell subdivisions showed different c-fos immunoreactivity response patterns to an amphetamine challenge after the sensitizing regimen (Fig. 5a–d and Fig. 6). The core showed a clear effect of sensitization with a number of c-fos positive nuclei in the AA group 48% higher than that of the SA group and more than 100% compared to both saline-challenged groups (Fig. 6). Significant main effects of pretreatment (F(1,28) = 10.59; p < 0.005) and challenge (F(1,28) = 44.75; p < 0.001) and a significant pretreatment × challenge interaction (F(1,28) = 7.55; p < 0.05) confirmed this observation. In the nucleus accumbens shell, more c-fos positive nuclei were counted in the amphetamine than saline-challenged groups, but no sensitization effect was seen (Fig. 6). A significant main effect of challenge (F(1,28) = 4.41; p < 0.05), but not of pretreatment (F(1,28) = 0.57; n.s.), was observed, and no pretreatment × challenge interaction (F(1,28) = 0.04; n.s.) was present in the ANOVA of the nucleus accumbens shell data. Fig. 5Digital micrographs of c-fos immunopositive nuclei and representations of nuclei quantified in the nucleus accumbens from representative animals treated acutely with amphetamine (SA group; a, c, and e) and challenged with amphetamine after amphetamine pretreatment (AA group; b, d, and f). The low magnification overviews of nucleus accumbens in a (SA animal) and b (AA animal) show differences in the number of c-fos immunopositive nuclei between the two experimental groups. Corresponding detail images of the nucleus accumbens core, shown in c (detail from b) and d (detail from b), are medial to anterior commissure. Nuclei counted for quantification in c and d are illustrated in e and f, respectively. Red squares indicate immunoreactive nuclei classified as “dark,” green squares as “midrange,” and blue squares as “light” (see text for classification procedures). Broad white arrows in d and f indicate a “dark” immunopositive nucleus and its representation in the counted nuclei; gray arrows represent “midrange”; narrow white arrows indicate “light.” Scale bar in a indicates 300 μm for a and b; scale bar in c represents 100 μm for c and d. cp caudate putamen, ac anterior commissure, core nucleus accumbens core, shell nucleus accumbens shellFig. 6Mean number of c-fos immunopositive nuclei in nucleus accumbens core and shell and prelimbic, infralimbic, orbital, and lateral prefrontal cortex. Two-way ANOVA analysis of accumbens data showed significant main effects of pretreatment and challenge and a pretreatment × challenge interaction in the nucleus accumbens core. A significant effect of challenge was observed in the nucleus accumbens shell and for all PFC areas measured. Bars represent the number of c-fos immunopositive nuclei per mm2, error bars indicate SEM, n = 8 per group. Asterisk indicates significant differences in post hoc testing. See Fig. 3 for abbreviations of treatment groups In the prefrontal cortex, an increase in c-fos positive nuclei in amphetamine-challenged groups compared to saline-challenged groups was observed in all four subregions measured. No differences were seen between amphetamine- and saline-pretreated animals in the number of c-fos positive nuclei after an amphetamine challenge (Fig. 6). Significant main effects of challenge were observed in all areas (PL: F(1,28) = 13.87, p < 0.001; IL: F(1,28) = 6.94, p < 0.05; orbital PFC: F(1,28) = 25.96, p < 0.001; lateral PFC: F(1,28) = 20.55, p < 0.001). No significant effects of pretreatment (PL: F(1,28) = 0.14, n.s.; IL: F(1,28) = 1.20, n.s.; orbital PFC: F(1,28) = 0.945, n.s.; lateral PFC: 0.79, n.s.) or pretreatment × challenge interactions (PL: F(1,28) = 0.08, n.s.; IL: F(1,28) = 0.037, n.s.; orbital PFC: F(1,28) = 0.039; lateral PFC: F(1,28) = 0.67) were observed. Frequency distributions of relative optical densities across groups show sensitization of midrange in accumbens core When comparing the ROD of the c-fos positive nuclei of each area across groups, significantly higher (=darker) averages were seen in amphetamine- compared to saline-challenged groups in the core (optical density 0.1216 ± 0.0026 in saline-challenged, 0.1372 ± 0.0040 in amphetamine-challenged; F(1,28) = 10.815, p < 0.005), the shell (optical density 0.1208 ± 0.0023 in saline-challenged, 0.1307 ± 0.0035 in amphetamine-challenged; F(1,28) = 5.513, p < 0.05), the infralimbic area (intensity 0.7173 ± 0.0057 in saline-challenged, 0.6985 ± 0.0660 in amphetamine-challenged; F(1,28) = 4.368, p < 0.05), the orbital prefrontal area (intensity 0.7181 ± 0.0048 in saline-challenged, 0.6921 ± 0.0073 in amphetamine-challenged; F(1,28) = 8.258, p < 0.001), and the lateral prefrontal area (intensity 0.7208 ± 0.0045 in saline-challenged, 0.6992 ± 0.0067 in amphetamine-challenged; F(1,28) = 6.753, p < 0.05). Significant differences were seen between the AA and SS groups and between the SA and SS group in the core in post hoc testing. No significant differences were seen in the shell in post hoc testing. The distributional patterns of the cellular ROD suggested differences in the distributions of ROD across the treatment groups. To further analyze the optical densities and compare the frequencies of the various optical densities between groups, histograms of the optical densities of all cells per area, per group were constructed (Figs. 5e–f and 7a and b) and used to divide immunostained nuclei into light, midrange, and dark optical density (OD) ranges based on the 33rd and 66th percentile values in the SS group (Table 1), as described in the “Materials and methods” section. Two-way ANOVAs for pretreatment × challenge were conducted, followed by Tukey HSD post hoc tests. For brevity, the results of the post hoc tests are indicated in Table 1. Fig. 7Representations of distributions of optical densities in the nucleus accumbens. In a and b, histograms of relative optical densities of individual c-fos immunopositive nuclei in nucleus accumbens core (a) and shell (b) are depicted. T1 and T2 lines represent the 33rd and 66th percentile values, respectively, for the SS group. In the nucleus accumbens core, the AA group shows more neurons in the midrange segment (between T1 and T2), as confirmed by the analysis of number of cells per segment (c). A significant pretreatment × challenge interaction is present in the midrange of the nucleus accumbens core. The amphetamine-challenged groups both contain more neurons in the dark segment of both core (c) and shell (d) compared to saline-challenged groups, as evidenced by a significant effect of challenge in the ANOVA test. No significant differences were observed in post hoc testing in the shell; asterisks indicate significant differences in post hoc testing for core in c (p < 0.05). Line colors and patterns or bar fillings indicating groups are as shown in a and c. For c and d, bars represent the number of c-fos positive nuclei in each OD segment per mm2, error bars indicate SEM, n = 8 per group. See Fig. 2 for abbreviations of treatment groups In both the core and the shell, amphetamine-challenged animals showed higher numbers of c-fos positive nuclei than saline-pretreated animals in the dark range (core challenge: F(1,28) = 23.928, p < 0.001; shell challenge: F(1,28) = 6.421, p < 0.05). A significant effect of pretreatment was seen in the core in this range (F(1,28) = 5.202, p < 0.05), and no effect was seen in the shell (F(1,28) = 0.949, n.s.). No pretreatment × challenge interactions were observed in the core (pretreatment × challenge: F(1,28) = 1.227, n.s.; Fig. 7c and Table 1) or shell (pretreatment × challenge: F(1,28) = 0.103, n.s.; Fig. 7d and Table 1) in this range. In the midrange nuclei, the core area showed a distinct pattern of distribution compared to all other areas measured as a sensitization effect was confirmed by a significant pretreatment × challenge interaction (F(1,28) = 4.901, p < 0.05; Fig. 7c), while no significant main effects were observed (pretreatment: F(1,28) = 1.985, n.s.; challenge: F(1,28) = 3.195, n.s.). No differences between groups were observed in the midrange nuclei in the shell (pretreatment: F(1,28) = 0.474, n.s.; challenge: F(1,28) = 0.146, n.s.; pretreatment × challenge: F(1,28) = 0.229, n.s.; Fig. 7d). No differences between groups were seen in the light range nuclei in the core (Fig. 7c; pretreatment: F(1,28) = 0.000, n.s.; challenge: F(1,28) = 0.002, n.s.; pretreatment × challenge: F(1,28) = 0.532, n.s.) or the shell (Fig. 7d; pretreatment: F(1,28) = 0.624, n.s.; challenge: F(1,28) = 0.280, n.s.; pretreatment × challenge: F(1,28) = 0.632, n.s.). Within the prefrontal cortex, the areas measured showed a relatively homogenous pattern of distribution of c-fos immunoreactivity across the OD ranges (Table 1). In the dark range, all areas showed significantly more c-fos positive nuclei in the amphetamine-treated groups compared to the saline-treated groups (main effect challenge: PL: F(1,28) = 14.974, p < 0.005; IL: F(1,28) = 11.563, p < 0.005; orbital: F(1,28) = 31.370, p < 0.001; lateral: F(1,28) = 23.751, p < 0.001) but no effects of pretreatment (main effect pretreatment: PL: F(1,28) = 0.035, n.s.; IL: F(1,28) = 0.645, n.s.; orbital: F(1,28) = 0.209, n.s.; lateral: F(1,28) = 0.092, n.s.) or interaction between pretreatment and challenge (pretreatment × challenge: PL: F(1,28) = 0.207, n.s.; IL: F(1,28) = 0.014, n.s.; orbital: F(1,28) = 0.008, n.s.; lateral: F(1,28) = 0.015, n.s.). In the midrange, increased numbers of c-fos positive nuclei in amphetamine-challenged animals were seen in the prelimbic (F(1,28) = 8.081, p < 0.01), orbital (F(1,28) = 7.941, p < 0.01), and lateral (7.029, p < 0.05) areas, but not the infralimbic area (F(1,28) = 1.545, n.s.). No effects were seen in the midrange segment of pretreatment (main effect pretreatment: PL: F(1,28) = 0.265, n.s.; IL: F(1,28) = 1.511, n.s.; orbital: F(1,28) = 1.453, n.s.; lateral: F(1,28) = 1.461, n.s.), and no pretreatment × challenge interactions were seen (pretreatment × challenge: PL: F(1,28) = 0.001, n.s.; IL: F(1,28) = 0.113, n.s.; orbital: F(1,28) = 0.928, n.s.; lateral: F(1,28) = 1.076, n.s.). In the light range, only the lateral prefrontal cortex showed a significant increase in c-fos positive nuclei in the amphetamine-challenged groups (main effect challenge: F(1,28) = 4.228, p < 0.05). No other challenge effects were observed (main effect challenge: PL: F(1,28) = 3.751, n.s.; IL: F(1,28) = 0.243, n.s.; orbital: F(1,28) = 0.302, n.s.). No effects of pretreatment (main effect pretreatment: PL: F(1,28) = 1.225, n.s.; IL: F(1,28) = 1.060, n.s.; orbital: F(1,28) = 0.863, n.s.; lateral: F(1,28) = 0.793, n.s.) or of sensitization (pretreatment × challenge interaction: PL: F(1,28) = 0.356, n.s.; IL: F(1,28) = 0.156, n.s.; orbital: F(1,28) = 0.015, n.s.; lateral: F(1,28) = 0.573, n.s.) were seen in the light range of the PFC c-fos immunopositive nuclei. Discussion In the present study, we set out to characterize the behavioral response to amphetamine in drug-naive and amphetamine-pretreated animals, as we had previously observed that treatment with a high dose (5 mg/kg) of amphetamine in drug-naive animals resulted in a similar pattern of cellular reactivity in the dorsal striatum as a challenge with an intermediate dose (1 mg/kg) in amphetamine-pretreated, behaviorally sensitized rats (Vanderschuren et al. 2002). Furthermore, we aimed to characterize the reactivity of subregions within the nucleus accumbens and prefrontal cortex during the expression of psychostimulant sensitization. Our results showed that previously demonstrated similarity of cellular activation patterns in the dorsal striatum after amphetamine in sensitized or drug-naive rats was not accompanied by similar behavioral responses. Thus, animals given acute amphetamine at a dose that produces heightened patch activation (5 mg/kg) displayed intense stereotypy, which was not seen during the expression of sensitization at a challenge dose (1 mg/kg) that also produces higher activation of patches than surrounding matrix. Cellular activation was sensitized in response to a drug challenge in the nucleus accumbens core, but not the shell or the prefrontal cortex. Examination of the distribution patterns of the optical densities of the individual immunopositive nuclei revealed an upward shift in the histograms of the core in the AA group, resulting in a significant increase in the midrange stained nuclei in the accumbens core in sensitized animals. Sensitization of locomotor activity and rearing without stereotypy After amphetamine pretreatment, an augmented locomotor and rearing response to amphetamine was seen compared to saline-pretreated, amphetamine-challenged animals, thus supporting the notion that stereotypy and patch hyperreactivity could take place independently. The behavioral pattern of the amphetamine-pretreated, amphetamine-challenged group most resembled the animals treated acutely with 1.0 or 2.5 mg/kg of amphetamine. No stereotyped behavior was observed in the sensitized animals after an amphetamine challenge, while intense stereotypy was seen in animals treated acutely with 5 mg/kg of amphetamine. In our previous paper (Vanderschuren et al. 2002), we noted that the decrease in locomotion normally observed in animals treated acutely with a high dose of amphetamine was likely a result of stereotypy, an idea which is strengthened by the present observations. In previous studies by our laboratory and others, animals treated acutely with the relatively high dose of 5 mg/kg of amphetamine (Graybiel et al. 1990; Vanderschuren et al. 2002) or with the same amphetamine regimen for sensitization followed by a challenge with 1 mg/kg amphetamine used in the present study (Vanderschuren et al. 2002) show enhanced ratios of response in patches compared to the matrix in the dorsal striatum, findings visually confirmed in the present set of experiments. We concluded on the basis of our previous results that the changes seen in striatal reactivity represent a shift in sensitivity to amphetamine rather than long-term adaptations in circuitry. If the increased sensitivity to amphetamine underlying behavioral sensitization paralleled the hyperreactivity of patches, the same pattern of behavior would be expected in animals showing similar immediate-early gene expression patterns, namely, the 5-mg/kg acutely challenged group and the amphetamine-sensitized, amphetamine-challenged group. However, the expression of stereotypy in the 5-mg/kg group, not seen in the sensitized animals, and the erratic locomotor activity seen in the 5-mg/kg group, clearly indicate behavioral differences after the two treatments. It is interesting to note that the dissociation of hyperreactivity of patches and sensitized behavioral responses manifests in a lack of stereotypical behavior in sensitized animals, as increased reactivity of patches compared to the matrix has been suggested to underlie stereotypical behavior (Canales and Graybiel 2000). Given the fact that the sensitization regimen and challenge dose used in the present study causes heightened patch to matrix ratios of c-fos expression, but does not cause stereotyped behavior, changes in striatal patterns of activity are apparently not causally related to stereotypy. Sensitization of immediate-early gene activation in a specific neural population of the nucleus accumbens core The sensitized immediate-early gene expression we observed in the nucleus accumbens core during the expression of psychostimulant sensitization after a withdrawal period of 14 days contradicts a number of previous immediate-early gene studies where no sensitization effect was observed in the core (Ostrander et al. 2003; Todtenkopf et al. 2002a), including one study from our own laboratory (Vanderschuren et al. 2002). There are two main differences between the present study and the previous study from our group. First, in our previous study, a 3-week abstinence period was observed, while in the present study, animals were tested after 14 days of abstinence. It is interesting to note that studies examining immediate-early gene expression after shorter abstinence periods (2 days) do show sensitization of reactivity within the core (Hedou et al. 2002; Todtenkopf et al. 2002a), indicating that abstinence time is of potential importance for sensitization of c-fos in the nucleus accumbens core. However, this explanation seems unsatisfactory as the study by Totenkopf et al. (2002a) also tested c-fos expression after 2 weeks of abstinence and found no sensitization of c-fos expression in the nucleus accumbens core. This discrepancy might be explained by a second difference between the studies where no effect was observed and the present study: in studies finding no effect of sensitization on the nucleus accumbens core, animals were at least partially pretreated and challenged in the same environment. Exposure to amphetamine in a relatively novel environment has been shown to potentiate the c-fos response to the drug in the nucleus accumbens core (Ostrander et al. 2003).The importance of the testing environment was recently underscored by a study showing that sensitized c-fos responses to cocaine in the nucleus accumbens only occurred when cocaine was always administered in a discrete environment outside of the home cage (Hope et al. 2006). Expression of c-fos in the striatum can be elicited by stimulation of dopamine receptors (Berretta et al. 1992), which have been demonstrated to be crucially involved in the expression of amphetamine sensitization (Vanderschuren and Kalivas 2000). As mentioned in the “Introduction”, a number of studies point to the sensitization of dopamine transmission in the accumbens core during the expression of sensitization (Cadoni et al. 2000; Phillips et al. 2003), and increased dopamine is also seen in the core compared to the shell in yoked controls in cocaine and heroin self-administration studies (Lecca et al. 2007a; Lecca et al. 2007b). It is thus possible that dopamine may play a role in eliciting the increased c-fos response seen in the accumbens core in the present study. In the accumbens core, more cells in the midrange of optical densities were present in the amphetamine-pretreated, amphetamine-challenged group than in all other groups. This “upward” shift in sensitized animals, visible in the histogram of optical density frequencies, indicates that the increase in c-fos positive nuclei is not due to nuclei that were under our detection threshold for c-fos increasing their reactivity enough to be measured. In the case of increased c-fos expression within the same population of nuclei that normally respond to amphetamine, a rightward shift in the histogram of optical densities would be expected, which would manifest in a sensitization effect in the dark range and increase of the average optical density. No increase was seen in the overall average optical density, no significant interaction observed between pretreatment and challenge in the dark group, and there is no significant difference between the AA and SA groups in the dark group. However, an obvious trend to sensitization in the darkly stained group is visible in Fig. 6. Although the lack of significance precludes firm conclusions, it might be that some neurons which normally respond to amphetamine by producing c-fos react by producing more c-fos after sensitization. The increase of frequency of neurons stained in the midrange without a significant shift in average optical density suggests the recruitment of a new population of nuclei becoming responsive to amphetamine in the sensitized animals. The population most likely to be involved in that sensitized response is the dopamine D2 receptor-containing, enkephalin-positive population that project via the subthalamic nucleus to the substantia nigra pars reticulata and entopenduncular nucleus. Indeed, sensitization of cocaine-induced c-fos expression in the nucleus accumbens in animals treated with the drug in a discrete environment was only found in enkephalin-positive cells, but not in dynorphin-positive, dopamine D1 receptor-expressing neurons that directly project to the substantia nigra (Hope et al. 2006). Differences in responsivity to amphetamine in these two cellular populations have also been demonstrated in a study where dopamine D2 receptor-containing neurons became activated when amphetamine is administered in a novel environment (Badiani et al. 1999). Remarkably, c-fos expression in dopamine D1 receptor-containing accumbens neurons was shown to be important for the induction of cocaine sensitization (Zhang et al. 2006), suggesting that these two different cell populations play distinct roles in the induction and expression of behavioral sensitization. Immediate-early gene responses to acute amphetamine in the nucleus accumbens shell and prefrontal cortex The cellular response of the nucleus accumbens shell to acute amphetamine, which did not sensitize after repeated amphetamine, corresponds well with previous results from our laboratory demonstrating the same effect (Vanderschuren et al. 2002). The role of the shell in the expression of psychostimulant sensitization is not clear-cut. On the one hand, psychostimulant sensitization-induced long-term changes in cellular reactivity or cellular morphology are not generally found in the nucleus accumbens shell after psychostimulant administration (Cadoni et al. 2000; Li et al. 2004; Todtenkopf et al. 2002a). Moreover, post induction lesions of the shell leave the expression of cocaine sensitization intact (Todtenkopf et al. 2002b), although preinduction lesions of a subarea of the shell result in reduced sensitized, but not acute responses to cocaine (Brenhouse and Stellar 2006). On the other hand, microinjection of amphetamine or cocaine into the shell (but not core) produces sensitized psychomotor responses and augmented dopamine levels in animals pretreated with cocaine (Filip and Siwanowicz 2001; Pierce and Kalivas 1995), and sensitized cellular reactivity has been observed in specific subareas within the accumbens shell after repeated cocaine administration (Brenhouse et al. 2006; Todtenkopf et al. 2002a). Discrepancies between studies finding the effects of sensitization in the shell and the present study could be due to the use of cocaine pretreatment and/or challenges in other studies and amphetamine pretreatment and challenge in the present study. Cocaine and amphetamine have been observed to produce different activation patterns in the dorsal striatum with acute amphetamine producing activation of patches and acute cocaine producing a more homogenous staining pattern (Graybiel et al. 1990); it is possible that cocaine and amphetamine pretreatment produce different patterns of reactivity in the ventral striatum as well. Within the prefrontal cortex, all areas showed c-fos expression in response to amphetamine, but there was no effect of amphetamine preexposure. To our knowledge, this is the first examination of c-fos expression in orbital and lateral prefrontal areas. However, the lack of sensitization of c-fos expression in the medial prefrontal cortex corresponds well with results from other groups showing similar results after 2 weeks of abstinence (Todtenkopf et al. 2002a). Sensitization of c-fos immunoreactivity does occur when animals are challenged after 2 days of abstinence (Hedou et al. 2002; Todtenkopf et al. 2002a, b), suggesting a role in the induction or early phases of sensitization. In support of a role for the prefrontal cortex in the induction and early phases of expression, lesion studies have demonstrated region-specific effects in the involvement of prefrontal areas in the induction of cocaine sensitization (Tzschentke and Schmidt 2000), and lesions of the entire medial prefrontal cortex have been shown to prevent the induction of amphetamine sensitization (Wolf et al. 1995; Cador et al. 1999, but see Tzschentke and Schmidt 2000). Studies examining expression of psychostimulant sensitization suggest that the dorsal areas of the medial prefrontal cortex that project to the nucleus accumbens core are involved in the expression of psychostimulant sensitization (Steketee 2003). Although no sensitization of immediate-early gene expression was observed in the present study, a dorsal–ventral gradient was seen in acute response to amphetamine with the prelimbic area showing a 74% increase in the SA group over the SS group and the infralimbic area showing a smaller increase at 45%. The high responsivity of lateral prefrontal neurons to amphetamine (114% increase in SA over SS) is also interesting in this respect, as the projections from this area to the nucleus accumbens are nearly exclusive to the nucleus accumbens core (Berendse et al. 1992). The highly active prefrontal cortical inputs projecting to the core may play a role in enhancing the activation of the core during the expression of sensitization. Remarkably, the highest responsivity to acute amphetamine was found in the orbital frontal area, a prefrontal cortex area that projects primarily to the dorsal striatum and only very sparsely to the nucleus accumbens (Berendse et al. 1992). Metabolic and structural changes within the orbital prefrontal cortex have been demonstrated in human drug addicts and nonhuman primates and rodents with a history of drug self-administration (Crombag et al. 2005; Porrino and Lyons 2000; Volkow and Li 2004). A recent study showed that rats sensitized to cocaine are impaired on a task that is sensitive to orbital prefrontal lesions (Schoenbaum et al. 2004). The high responsivity of this area to acute amphetamine indicates that corticostriatal circuits not involving the nucleus accumbens also play an important role in the acute response to amphetamine. In conclusion, challenging animals sensitized to amphetamine with a dose of the drug that produces preferential activation of dorsal striatal patches did not produce stereotyped behavior, unlike the behavioral pattern observed in animals treated acutely with a dose of amphetamine that causes the same pattern of neural activation. Sensitized immediate-early gene activity was found in the nucleus accumbens core, but not the accumbens shell or the prefrontal cortex. The gradients observed in the present study in cellular responses to amphetamine in subareas of the prefrontal cortex with specific afferents to the nucleus accumbens suggest a specific role for various corticostriatal loops in the behavioral responses to psychostimulant drugs, further supporting the importance of accumbens–prefrontal cortex interactions in drug addiction (Volkow and Li 2004; Robbins and Everitt 2002).

Eur_J_Epidemiol-4-1-2226058

Apolipoprotein E gene is related to mortality only in normal weight individuals: The Rotterdam study

Objective To investigate the relationship between the apolipoprotein E (APOE) gene and the risk of mortality in normal weight, overweight and obese individuals. Methods and Results In a population-based study of 7,983 individuals aged 55 years and older, we compared the risks of all-cause and coronary heart disease (CHD) mortality by APOE genotype, both overall and in subgroups defined by body mass index (BMI). We found significant evidence for interaction between APOE and BMI in relation to total cholesterol (p = 0.04) and HDL cholesterol (p < 0.001). Overall, APOE*2 carriers showed a decreased risk of all-cause mortality. Analyses within BMI strata showed a beneficial effect of APOE*2 only in normal weight persons (adjusted hazard ratio (HR) 0.7[95% CI 0.5–0.9]). APOE*2 was not associated with a lower risk of all-cause mortality in overweight or obese persons. The effect of APOE*2 in normal weight individuals tended to be due to the risk of CHD mortality (adjusted HR 0.5 [95% CI 0.2–1.2]). Conclusion The APOE*2 allele confers a lower risk of all-cause mortality only to normal weight individuals. Apolipoprotein E is a plasma protein involved in the metabolism of cholesterol. The apolipoprotein E isoforms ε2, ε3 and ε4, determined by the three alleles of the apolipoprotein E gene (APOE*2, APOE*3 and APOE*4), differ in their binding affinity to serum cholesterol and hence in their ability to clear dietary fat from the blood [1]. Accordingly, total serum cholesterol levels differ between APOE genotypes, with APOE*2 carriers having lower total serum cholesterol levels and APOE*4 carriers having higher levels than APOE*3 homozygotes [1–4]. In line with the clear differences in total serum cholesterol levels associated with the APOE genotype, one would expect that APOE*2 carriers have lower risks of cardiovascular morbidity and mortality and APOE*4 carriers have higher risks. Yet, the evidence for this association is inconsistent. A recent meta-analysis showed that APOE*4 was associated with a higher risk of coronary heart disease (CHD), but found no association with APOE*2 [5]. APOE*2 has been associated with a lower risk of mortality [6] and APOE*4 with a higher risk [7–11], but also these relationships were not found by others [12–19]. A possible explanation for these inconsistencies may be found in the role of body weight. Several studies have demonstrated that the relationship between the APOE gene and lipids differs between normal weight, overweight and obese individuals [20–22]. Total serum cholesterol levels were higher among obese than among non-obese APOE*4 carriers [20], and obese APOE*2 carriers had higher serum triglycerides and higher LDL-cholesterol levels than non-obese APOE*2 carriers [20, 21]. While APOE*2 carriers generally have the lowest cholesterol levels, those with high BMI were found to have similar total serum cholesterol levels compared to APOE*3 homozygotes and APOE*4 carriers [20]. These findings suggest that the adverse effects of high body weight may outweigh the beneficial effect of APOE*2 status. If so, a protective effect of APOE*2 on mortality may be largest among normal weight individuals. Despite this evidence for a moderating effect of body weight on the relation between the APOE gene and total serum cholesterol, no studies so far have performed a stratified analysis on the relationship between the APOE gene and the risk of mortality. The aim of the present study was to investigate the extent to which the relationship between APOE genotypes and the risk of mortality differs between normal weight, overweight and obese individuals. We examined this relationship for both all-cause and CHD mortality. Methods Study population and procedures The present analyses were performed in the Rotterdam Study, an ongoing population-based study on the determinants of disease and disability among persons 55 years and older. Details of this study have been described elsewhere [23]. Baseline data were collected between 1990 and 1993. During home visits, a trained investigator obtained information on health status, medical history, medication use and lifestyle. Subsequently, participants were invited to the study center where they underwent an extensive clinical examination. The Medical Ethics Committee of the Erasmus Medical Center approved the study protocol and all participants provided written informed consent. From 10,275 eligible subjects, 7,983 (78%) individuals agreed to participate in the study and were examined at baseline. Data collection Clinical and laboratory assessments The clinical examination included weight, height, systolic and diastolic blood pressure, serum glucose, total serum cholesterol and high-density lipoprotein (HDL) cholesterol levels. BMI was computed as weight (kg) divided by height squared (m2). Participants were classified as underweight (BMI < 18.5 kg/m2), normal weight (18.5 kg/m2 ≤ BMI < 25 kg/m2), overweight (25 kg/m2 ≤ BMI < 30 kg/m2) or obese (BMI ≥ 30 kg/m2) [24]. The waist and hip circumferences were measured, and the waist-hip-ratio (WHR) was calculated as indirect assessment of abdominal fat. Systolic and diastolic blood pressures were measured twice in a sitting position after 5 min rest using a random-zero sphygmomanometer. The mean of the two measurements was used for the analyses. Hypertension was defined as systolic blood pressure higher than 160 mmHg, diastolic blood pressure higher than 100 mmHg, or the use of antihypertensive medication indicated to treat high blood pressure (hypertension grades 2 and 3) [25]. Diabetes was diagnosed based on a random or post-load glucose level higher than 11.0 mmol/l and/or the use of anti-diabetic medication [26]. Total serum cholesterol and HDL-cholesterol were measured using an automated enzymatic procedure [27]. Hypercholesterolemia was defined as a total serum cholesterol level above 6.2 mmol/l [28]. The APOE genotype was determined on DNA samples using a polymerase chain reaction followed by enzymatic digestion using methods previously described [29]. The frequencies of APOE genotypes were APOE 2/2 0.8%, APOE 2/3 12.8%, APOE 2/4 2.7%, APOE 3/3 58.4%, APOE 3/4 22.9% and APOE 4/4 2.4%. The proportions of the APOE alleles and genotypes were in Hardy-Weinberg equilibrium (p = 0.71). Mortality data Information on the vital status of the participants was obtained at regular intervals from the municipal population registry. Causes of death were obtained from the general practitioners by means of a standardized questionnaire relating to the circumstances of death, most likely cause of death and time and place of death. Two independent research physicians coded all events according to the International Classification of Diseases, 10th edition (ICD-10) [30]. CHD mortality was defined as death from diseases coded I20–I25, I46, I50 or R96. Mortality data were available up to December 2003. Statistical analyses Of the 7,983 individuals who participated at baseline, 5,817 (73%) had APOE genotyped successfully and had complete information on BMI, WHR and cholesterol levels. Because persons with an extremely low BMI may suffer from life-threatening diseases such as cancer, individuals with a BMI below 18.5 kg/m2 (n = 51) were excluded from the analyses. Also, APOE 2/4 individuals were excluded from the analyses (n = 158) to distinguish the effect of the APOE alleles unambiguously. Hence, data from 5,608 participants were available for the analyses. Differences in baseline characteristics by APOE genotypes were tested using the chi-squared statistic (categorical variables) or ANOVA (continuous variables). P for trend was obtained by testing the linearity of the sum of squares from the ANOVA analyses. Multiple linear regression analysis was used to assess the interaction between APOE genotype and BMI on total serum cholesterol and HDL-cholesterol levels. Differences in survival probabilities were examined by comparing Kaplan-Meier survival curves and tested using the log rank test. Kaplan-Meier plots were constructed using age as the time scale to take proper account of the effect of age [31, 32]. Risks of mortality were quantified as hazard ratios (HRs) using Cox proportional hazards analyses with age as the time scale. The most common genotype (APOE 3/3) was used as the reference category. The proportionality assumption of all models was verified by testing the Schoenfeld residuals [33]. Because the proportionality assumption was not met, HRs were calculated for early and late mortality using 80 years as the cut-off age. This cut-off age was selected because it led to proportionality of the models below and above the cut-off age. HRs were calculated adjusted for gender, smoking status, education level, total serum cholesterol, HDL-cholesterol, waist to hip ratio hypertension and diabetes mellitus. Results The mean age at entry was 68.7 years (SD = 8.7 years) and 41.8% of the participants were men. The mean follow up time was 11.1 years (SD = 3.8 years). Baseline characteristics of APOE*2 carriers, APOE*3 homozygotes and APOE*4 carriers are presented in Table 1. As expected, cholesterol levels differed significantly between the genotype groups. The prevalence of hypercholesterolaemia was lowest among APOE*2 carriers (49%) and highest among APOE*4 carriers (66%). Furthermore, there were statistically significant differences in BMI scores between the genotype groups, with APOE*2 carriers having the highest mean BMI (26.7 ± 3.6 kg/m2) and APOE*4 carriers the lowest (26.1 ± 3.4 kg/m2). There were no statistically significant differences in WHR between the genotype groups. Table 1Baseline characteristics by APOE genotype groupsAPOE*2 carriersAPOE 3/3APOE*4 carrierspn7823,3691,457Sex (men)3842420.05Age at entry (years)68.8 (8.8)68.9 (8.7)68.3 (8.5)0.14Body mass index (kg/m2)26.7 (3.6)26.4 (3.6)26.1 (3.4)<0.01Normal weight343940Overweight494747<0.05Obese171413Waist-to-hip ratio0.9 (0.1)0.9 (0.1)0.9 (0.1)0.18Systolic blood pressure (mmHg)140 (22)140 (23)138 (2)0.04Diastolic blood pressure (mmHg)74 (11)74 (12)73 (11)0.06Hypertension3634330.25Total serum cholesterol (mmol/l)6.4 (1.3)6.6 (1.2)6.8 (1.2)<0.001HDL-cholesterol (mmol/l)1.4 (0.4)1.3 (0.3)1.3 (0.4)<0.001Cholesterol lowering medication3230.71Hypercholesterolemia496066<0.001Smoking status (current)3935330.02Education level    Lower434345    Intermediate4342400.21    Higher141515Diabetes mellitus101090.75Values are means (standard deviations) for continuous variables and percentages for categorical variables. p values are obtained by ANOVA for continuous variables and by χ2 for categorical variables Figure 1 presents mean total serum cholesterol and HDL-cholesterol levels by APOE genotype in normal weight, overweight and obese persons. In all BMI strata, APOE*2 carriers had the lowest and APOE*4 carriers the highest total serum cholesterol (pfor trend < 0.01). The association between HDL-cholesterol and APOE genotype was in the opposite direction (pfor trend < 0.01). Note that even though APOE*2 carriers had the lowest total serum cholesterol levels in each BMI group, still 50% of the overweight and 57% of the obese APOE*2 carriers had hypercholesterolaemia compared to 41% of the normal weight APOE*2 carriers (p = 0.003). There was no significant evidence for interaction between APOE genotype and BMI in relation to total serum cholesterol (p = 0.26) and HDL-cholesterol (p = 0.25). However, when BMI was analyzed as a continuous trait, then the p value for interaction was 0.04 for total serum cholesterol and less than 0.001 for HDL-cholesterol. Fig. 1Mean total serum cholesterol and HDL-cholesterol levels by APOE genotypes for normal weight, overweight and obese individuals. HDL: high-density lipoproteins A total of 1,918 deaths occurred during follow-up. Of those, 257 occurred in 782 APOE*2 carriers (33%), 1,160 in 3,369 APOE*3 homozygotes (33%) and 501 in 1,457 APOE*4 carriers (34%). Differences between the Kaplan-Meier survival curves of the APOE*2 carriers, APOE*3 homozygotes and APOE*4 carriers were small but statistically significant (p = 0.03) with APOE*2 carriers having slightly lower mortality risks (Fig. 2). After adjustment for other cardiovascular risk factors, APOE genotype was not associated with all-cause mortality before or after 80 years of age in the overall population (Table 2). The differences between the curves were only statistically significant in normal weight individuals (p = 0.04; Fig. 3). Normal weight carriers of the APOE*2 allele had a significantly lower risk of mortality before age 80 years (HR [95% CI] = 0.7 [0.5–0.9]), but not after age 80 years (crude HR [95% CI] = 1.0 [0.8–1.4]). There were no significant differences in the risk of mortality between genotype groups among overweight or obese persons (Table 2). Fig. 2Kaplan Meier survival curves for all-cause mortality by APOE genotypesTable 2Incidence rates and hazard ratios for all-cause mortality by APOE genotypes and BMI categoriesFollow-upIncidence rateHazard ratioFollow-upIncidence rateHazard ratioDeaths(py)Deaths/1,000 py (95% CI)(95% CI)Deaths(py)Deaths/1,000 py (95% CI)(95% CI)Before 80 years of ageAfter 80 years of ageAllAPOE*2 carriers1016,78214.9 (12.3–18.1)0.9 (0.7–1.1)15621,7647.2 (6.1–8.4)0.9 (0.8–1.1)APOE 3/348128,70316.8 (15.3–18.3)67995,8087.1 (6.6–7.6)APOE*4 carriers19812,69315.6 (13.6–17.9)1.0 (0.8–1.2)30341,6377.3 (6.5–8.1)1.2 (1.0–1.4)Total78048,17816.2 (15.1–17.4)1,138159,2097.1 (6.7–7.6)Normal weightAPOE*2 carriers282,37011.8 (8.2–17.1)0.7 (0.5–0.9)577,2897.8 (6.0–10.1)1.0 (0.8–1.4)APOE 3/320611,31418.2 (15.9–20.9)25638,0456.7 (6.0–7.6)APOE*4 carriers785,08015.4 (12.3–19.2)1.0 (0.7–1.3)12717,3057.3 (6.2–8.7)1.3 (0.9–1.7)Total31218,76416.6 (14.9–18.6)44062,6397.0 (6.4–7.7)OverweightAPOE*2 carriers593,30517.9 (13.8–23.0)1.1 (0.9–1.5)6511,0645.9 (4.6–7.5)0.9 (0.7–1.1)APOE 3/321713,57316.0 (14.0–18.3)31045,3666.8 (6.1–7.6)APOE*4 carriers906,04714.9 (12.1–18.3)0.9 (0.7–1.2)13719,3987.1 (6.0–8.4)1.2 (1.0–1.5)Total36622,92416.0 (14.4–17.7)51275,8276.8 (6.2–7.4)ObeseAPOE*2 carriers141,10712.6 (7.5–21.4)1.0 (0.5–1.8)343,41110.0 (7.1–14.0)1.0 (0.6–1.5)APOE 3/3583,81615.2 (11.7–19.7)11312,3989.1 (7.6–11.0)APOE*4 carriers301,56619.2 (13.4–27.4)1.3 (0.9–2.0)394,9347.9 (5.8–10.8)1.1 (0.8–1.6)Total1026,48915.7 (12.9–19.1)18620,7429.0 (7.8–10.4)Hazard ratios adjusted for sex, waist-hip-ratio, smoking, education level, total serum cholesterol, HDL cholesterol hypertension and diabetes; py: person-years; CI: Confidence intervalFig. 3Kaplan Meier survival curves for all-cause mortality by APOE genotypes for normal weight, overweight and obese individuals To investigate whether the lower risk of mortality observed among normal weight APOE*2 carriers was due to a lower risk of mortality from CHD, we calculated the HRs separately for CHD mortality. APOE genotype was not significantly associated with CHD mortality. Analysis within BMI strata showed that normal weight APOE*2 carriers tended to have lower risks of CHD mortality (adjusted HR [95% CI] = 0.5 [0.2–1.2]). No differences in the risk of CHD mortality between genotype groups were found in overweight or obese participants (Table 3). Table 3Incidence rates and hazard ratios for CHD mortality by APOE genotypes and BMI categoriesFollow-upIncidence rateHazard ratioFollow-upIncidence rateHazard ratioDeaths(py)Deaths/1,000 py (95% CI)(95% CI)Deaths(py)Deaths/1,000 py (95% CI)(95% CI)Before 80 years of ageAfter 80 years of ageAllAPOE*2 carriers266,7823.8 (2.6–5.6)1.0 (0.6–1.5)412,10319.5 (14.4–26.5)1.1 (0.8–1.5)APOE 3/312628,7034.4 (3.7–5.2)1718,74419.6 (16.8–22.7)APOE*4 carriers5912,6934.6 (3.6–6.0)1.1 (0.8–1.4)663,47219.0 (14.9–24.2)1.1 (0.8–1.4)Total21148,1784.4 (3.8–5.0)27814,31919.4 (17.3–21.8)Normal weightAPOE*2 carriers52,3702.1 (0.9–5.1)0.5 (0.2–1.2)1566322.6 (13.6–37.5)1.6 (0.9–2.9)APOE 3/35611,3144.9 (3.8–6.4)483,07515.6 (11.8–20.7)APOE*4 carriers265,0805.1 (3.5–7.5)1.2 (0.7–1.9)211,28616.3 (10.6–25.1)1.1 (0.6–1.8)Total8718,7644.6 (3.8–5.7)845,02416.7 (13.5–20.7)OverweightAPOE*2 carriers163,3054.8 (3.0–7.9)1.4 (0.8–2.4)1598915.2 (9.1–25.2)0.7 (0.4–1.3)APOE 3/35113,5733.8 (2.9–4.9)924,03522.8 (18.6–28.0)APOE*4 carriers216,0473.5 (2.3–5.3)0.8 (0.5–1.4)351,61921.6 (15.5–30.1)1.0 (0.7–1.5)Total8822,9243.8 (3.1–4.7)1426,64321.4 (18.1–25.2)ObeseAPOE*2 carriers51,1074.5 (1.9–10.8)1.1 (0.4–3.4)1145024.4 (13.5–44.1)1.4 (0.7–3.1)APOE 3/3193,8165.0 (3.2–7.8)311,63419.0 (13.3–27.0)APOE*4 carriers121,5667.7 (4.4–13.5)1.6 (0.8–3.5)1056817.6 (9.5–32.7)1.1 (0.5–2.2)Total366,4895.5 (4.0–7.7)522,65119.6 (14.9–25.7)Hazard ratios adjusted for sex, waist-hip-ratio, smoking, education level, total serum cholesterol, HDL cholesterol hypertension and diabetes; py: person-years; CI: Confidence interval Discussion Our study shows that the APOE*2 allele has a protective effect on overall mortality. This protective effect seems to be limited to mortality from CHD. Analyses within BMI strata demonstrated that this genetic advantage of APOE*2 was observed only in normal weight individuals, but not in overweight and obese APOE*2 carriers. The results of the present analyses may seem to contradict our earlier work in this population. We previously reported that the APOE gene was not significantly related to mortality [12]. While our findings come from the same population, the mean follow-up time of the participants has increased from 5.4 years to 11.1 years and the number of deaths increased by 53% from 18% to 34%. This extended follow-up has improved the statistical power of our study to show a modest effect of APOE genotype on overall mortality. Research on the relationship between APOE and mortality has merely provided conflicting results [6–19, 34]. Three out of eleven studies showed a protective effect of APOE*2 [6, 14, 18]. and three a deleterious effect of APOE*4 [8–10]. The latter effect may partly mediated by the increased risk of Alzheimer disease in association with APOE*4 and mortality [10, 11]. Note that our analysis showed small significant difference in the Kaplan-Meier survival curves, demonstrating a protective effect of APOE*2 before age 80. This may be due to the effect of age that becomes apparent when performing survival analysis with age as time scale, as we did in the present paper. Such finding is compatible with previous reports suggesting that genes affecting human lifespan might be age-specific [35]. Genetic and environmental interactions at older ages vary from those at early ages and therefore explain the differences in the association between APOE and mortality before and after 80 years of age. A previous report showed that among nonagenarians, APOE genotype has no effect on cognitive function, cognitive decline or survival [36]. In line with previous studies [20–22], we found that the total serum cholesterol levels associated with APOE genotype differed between normal weight, overweight and obese individuals. As expected, mean levels of total serum cholesterol increased with BMI and APOE*2 carriers had the lowest mean total serum cholesterol in each BMI group. APOE*4 carriers had higher cholesterol levels and lower HDL-cholesterol concentrations than APOE*2 carriers despite the presence of a lower BMI and equal WHR. Previous reports showed that APOE genotype, BMI and WHR determine together the lipid levels [20–22, 37]. However, in our study abdominal fat, as measured by WHR, did not have any effect on the lipid levels by APOE genotype, while BMI was an important determinant. The significant interaction between APOE genotype and BMI in relation to total serum cholesterol found previously [20], was also supported by this study. Although the mechanism of such interaction has not been yet elucidated, evidence suggests that obesity and abdominal fat increase low density lipoprotein cholesterol and therefore total cholesterol at a higher degree among APOE*4 carriers [20–22, 37]. This is in line with our observation that the deleterious effect of increased BMI prevails over the beneficial effects of the APOE*2 allele. Furthermore, previous research in the Rotterdam Study and the Dutch population indicated that among APOE*2 homozygous carriers, the expression of Hyperlipoproteinemia type III is determined to a great extent by hyperinsulinemia and the presence of insulin resistance syndrome [38], conferring and increased risk of CHD through accelerated atherosclerosis [39]. Although the association between high BMI and high cholesterol levels is well known and the APOE gene is associated with cholesterol levels as previously described, this is the first study that examined the effect of APOE genotype on mortality in BMI strata. We found that presence of the APOE*2 allele was associated with a decreased risk of mortality only in normal weight individuals and that this lower risk was partly explained by lower risk of CHD mortality. These results were in line with the lower total cholesterol and higher HDL levels among normal weight APOE*2 carriers. Our finding that the protective effect of APOE*2 was only found in normal weight individuals suggests that the negative influences of increasing BMI outweigh the positive effects of genetic predisposition. When these results are confirmed by other population-based studies, a challenging question remains whether weight loss in overweight and obese APOE*2 carriers restores the genetic advantage of their APOE status.

Dev_Genes_Evol-3-1-1784541

Free-living flatworms under the knife: past and present

Traditionally, regeneration research has been closely tied to flatworm research, as flatworms (Plathelminthes) were among the first animals where the phenomenon of regeneration was discovered. Since then, the main focus of flatworm regeneration research was on triclads, for which various phenomena were observed and a number of theories developed. However, free-living flatworms encompass a number of other taxa where regeneration was found to be possible. This review aims to display and to compare regeneration in all major free-living flatworm taxa, with special focus on a new player in the field of regeneration, Macrostomum lignano (Macrostomorpha). Findings on the regeneration capacity of this organism provide clues for links between regeneration and (post-)embryonic development, starvation, and asexual reproduction. The role of the nervous system and especially the brain for regeneration is discussed, and similarities as well as particularities in regeneration among free-living flatworms are pointed out. Introduction Regeneration research in flatworms has a long standing history, but was focussed on triclads since its beginning in 1774, when Pallas (Pallas 1774) described the regeneration of the triclads Dendrocoelum lacteum and Bdellocephala punctata (from Brøndsted 1969). Since then, a vast amount of studies about regeneration in flatworms has appeared (for recent reviews, see Agata and Watanabe 1999; Sánchez Alvarado 2000, 2003, 2004, 2006; Saló and Baguñà 2002; Newmark and Sánchez Alvarado 2002; Agata 2003; Agata et al. 2003; Reddien and Sánchez Alvarado 2004; Reuter and Kreshchenko 2004; Sánchez Alvarado and Kang 2005; Saló 2006), but many of the old questions and many of the most intriguing phenomena that have been discovered cannot be explained today. Why can some animals regenerate, while others cannot? How is the duplication of heads or tails (heteromorphoses) possible in adult organisms? What conditions are sufficient and what are necessary for successful regeneration? Is regeneration a side-product of asexual reproduction, or the other way round? Is regeneration recapitulating pathways used in embryonic and postembryonic development? How are stem cells controlled and directed in regeneration? Are stem cells in adult flatworms totipotent, and can they be likened to embryonic blastomeres? This review provides a brief synopsis on what is known about the regeneration capacity in free-living flatworms, and addresses open questions about regeneration, with special emphasis on an emerging model organism, Macrostomum lignano (Egger et al. 2006a,b). A comparison of the regeneration capacity of the macrostomorph flatworm M. lignano with other free-living flatworms necessarily involves a look at their phylogenetic relationship. The Macrostomorpha belong to the largest taxon Rhabditophora, which also encompasses the Polycladida, Lecithoepitheliata, Prolecithophora, Proseriata, Bothrioplanida, Tricladida, Rhabdocoela, and the parasitic Neodermata (Ehlers 1985; Rieger 1996; Tyler et al. 2006). Macrostomorpha are considered to be the most basal taxon within the Rhabditophora due to a simple pharynx, entolecithal eggs, and the lack of vitellaria (Ehlers 1985). Referring to their small size (millimeter range), macrostomorphans are members of the “microturbellarians”, a nonsystematic denomination. Polyclads and triclads are the only taxa belonging to the “macroturbellaria”, with forms generally in the centimeter range. Triclads, as more derived rhabditophorans, have vitellaria, a rather complicated embryonic development (e.g., Cardona et al. 2005) and an often very pronounced regeneration capacity (see Brøndsted 1969). The status of the acoels While many molecular (e.g., Baguñà and Riutort 2004a,b) and also some morphological studies (e.g., Reuter et al. 2001, summed up in Egger and Ishida 2005) have placed the acoels outside the Plathelminthes, a number of characteristics (apomorphies) speak in favor of acoels being flatworms (Ehlers 1985), most notably the stem cell system and the special mode of epidermal replacement (Smith et al. 1986; Tyler and Hooge 2004). In this review, acoels and nemertodermatids are considered members of the Plathelminthes. What is a planarian? In flatworm regeneration research, other groups than triclads are often neglected and can easily be overlooked. This may in part be due to an ambiguous use of the term “planarian”, which is most often referring merely to triclads (more specifically, to paludicolans, e.g., Agata 2003; Reddien and Sánchez Alvarado 2004), but is also commonly used as a synonym for all free-living flatworms (e.g., Baguñà 1998; Saló and Baguñà 2002) or even used interchangeably for free-living flatworms and triclads alike (e.g., Newmark and Sánchez Alvarado 2001). The term “freshwater planarian”, while somewhat more specific, is still ambiguous, as a number of nontriclad flatworms, e.g., Stenostomum and Microstomum, are freshwater dwellers as well. In publications dealing with several flatworm taxa, it is often unclear what the term planarian signifies. It seems therefore desirable to agree on using less ambiguous denominations in scientific contexts, such as unequivocal scientific names, and to get rid of the term planarian. Overview of regeneration capacity in free-living flatworm groups All major free-living flatworm taxa with their tentative general regeneration capacities are given in Fig. 1. Asexually reproducing forms are only known for acoels, catenulids, macrostomorphans, and triclads. Within free-living flatworms, either catenulids (Ehlers 1985) or acoelomorphs (acoels and nemertodermatids, Rieger 1996) are considered as the basal-most taxon. All other flatworms are encompassed by the rhabditophorans, with macrostomorphans and polyclads at the base (Ehlers 1985; Rieger 1996). Fig. 1The regeneration capacity of the major free-living flatworm taxa. In all groups with asexual reproduction, species with a pronounced regeneration capacity can be found (green shades). Not all species in the listed taxa are necessarily showing the same regeneration capacity; only a broad classification of regeneration capacities is given. The major taxa were compiled after Ehlers (1985), Rieger (1996), and Tyler et al. (2006) Catenulida The regeneration capacity of two catenulid genera, namely, Stenostomum and Catenula, has been extensively investigated. Probably linked with asexual reproduction (paratomy), excellent regeneration capacity has been reported after artificial amputation (Ritter and Congdon 1900; Child 1903a,b; Ruhl 1927a; van Cleave 1929; and Palmberg 1990 for Stenostomum; Moraczewski 1977 for Catenula). Regeneration from the so-called organ primordia was found to be possible for all organ systems. Acoela Based on experiments mainly with Polychoerus caudatus (Stevens and Boring 1905; Child 1907; Keil 1929) and Paramecynostomum diversicolor (Peebles 1913), Hyman notes that acoels “have limited powers of regeneration” (Hyman 1951, p 182). The studied species were not able to regenerate ganglia (brain) or the statocyst. Later studies with Amphiscolops langerhansi (Steinböck 1954, 1955, 1963a,b) and Hofstenia giselae (Steinböck 1966, 1967) illustrated a substantial regeneration capacity in these acoels, where H. giselae was found to even regenerate the statocyst after amputation (Steinböck 1966). Several species can regenerate their eyes (Hanson 1967 for A. langerhansi and Pseudohaplogonaria sutcliffei, Yamasu 1991 for Praesagittifera naikaiensis, and Åkesson et al. 2001 for Convolutriloba longifissura). Acoels are now known to show all main modes of asexual reproduction, that is architomy (e.g., C. longifissura, Åkesson et al. 2001), paratomy (e.g., Paratomella unichaeta, Ax and Schulz 1959), and budding (e.g., C. retrogemma, Hendelberg and Åkesson 1988). Architomy is characterized by fission before formation of new organs, paratomy by formation of new organs before fission. Budding refers to a special case of paratomy, where the axis of the new piece is not corresponding to the axis of the old piece—in C. retrogemma, the axis of the buds is tilted by 180° in relation to the main animal (Hendelberg and Åkesson 1988, 1991). Nemertodermatida No account of regeneration potential or asexual reproduction has been given for animals from this taxon. Macrostomorpha As already observed in catenulids and acoels, the regeneration capacity of macrostomorphans is closely associated with the mode of reproduction. Thus, in the asexually reproducing taxa Microstomum lineare and Alaurina (paratomy), the regeneration capacity is most pronounced and all organs can be regenerated (von Graff 1908; Ruhl 1927a; Palmberg 1986, 1990, 1991). Regeneration research with the exclusive sexually reproducing genus Macrostomum started with Ruhl (1927a), who found that Macrostomum tuba cannot regenerate the head (brain, eyes, and pharynx) and requires at least a quarter of the gut in the anterior piece to be able to regenerate posterior parts, including copulatory organs. More recently, Ladurner et al. (1997) noted that in Macrostomum hystricinum marinum and Macrostomum pusillum the anterior part can fully regenerate the posterior part, while the posterior part cannot regenerate the anterior part. Similar findings were reported for M. lignano: cut in midbody, only the anterior part was found to regenerate the missing part (Rieger et al. 1999). Salvenmoser et al. (2001) described the regeneration of the musculature in M. lignano after the removal of the posterior end. Using the same species, Egger et al. (2006a,b) characterized the regeneration capacity at various cutting levels. Anterior regeneration is possible if the animals are cut in front of the brain; posterior regeneration can be successful from cutting levels starting with the posterior end of the pharynx. A distinct regeneration blastema occurs in M. lignano, but not in M. lineare, where organ primordia are formed without conspicuous accumulation of neoblasts at the wound site (Palmberg 1986). Polycladida The best-studied polyclad genus regarding regeneration is Leptoplana. Leptoplana alcynoe (now Letoplana alcinoi) and Leptoplana velutinus, as well as Thysanozoon brocchi and Cryptocelis alba, were shown to regenerate posterior parts by Monti (1900a). Using Leptoplana atomata, Schultz (1901, 1902) observed the regeneration of the posterior part, including gonads and copulatory organs, whereas head and pharynx were not regenerated. Similar results were obtained with Leptoplana tremellaris and L. alcinoi, although the pharynx was found to regenerate here (Child 1904a–c, 1905a; von Levetzow 1939). L. tremellaris can regenerate both anterior and posterior parts if only about half the brain is removed by transversal amputation. At amputation levels posterior of the brain, no complete anterior regeneration is possible (Child 1904a–c). Notoplana humilis shows a comparable regeneration capacity (Ishida 1998). Besides more posterior parts, T. brocchi is also able to regenerate the main tentacles, gut, and pharynx (Monti 1900a; von Levetzow 1939). Studying regeneration in Leptoplana saxicola, Pucelis litoricola, and Planocera californica, Olmsted (1922a) reports that these species follow the “polyclad rule for regeneration”, that is, regeneration of all amputated parts (including eyes) is possible if the brain remains intact. Several studies were dedicated to find out whether polyclads can regenerate the brain (Morgan 1905 for Leptoplana littoralis and von Levetzow 1939 for T. brocchi), and how the behavior changes after excision of the brain (Morgan 1905; Olmsted 1922b; von Levetzow 1939; Koopowitz et al. 1975, 1976) or after severance of the main lateral nerve cords (Olmsted 1922b; Koopowitz et al. 1975, 1976). Severed nerve cords and brain halves can fuse and resume their functions (Olmsted 1922b; Faisst et al. 1980), but the brain is not regenerated. Full anterior regeneration, including the brain, after complete removal of the brain was only observed in Cestoplana, provided the amputation level is a short distance behind the brain (Child 1905b). All mentioned species, except T. brocchi, are members of the taxon Acotylea (Tyler et al. 2006). From what is known, both Cotylea and Acotylea seem to share a comparable regeneration capacity. Lecithoepitheliata Only scarce reports about regeneration are available for this neoophoran taxon. For Prorhynchus stagnalis, Ruhl (1927a) remarks that no regeneration capacity whatsoever could be detected at anterior, posterior, and midbody cutting levels. The posterior pieces survived for 30 days. When cutting Geocentrophora baltica a short distance behind the pharynx, no regeneration could be observed (Steinböck 1927). All pieces died at the latest after 17 days. A closer examination of the regeneration capacity of Geocentrophora sphyrocephala revealed that these animals can approximately rebuild the mouth opening anteriorly in 3–5 weeks but cannot regenerate the brain. Both parts die after 40 days if the animals are cut just behind the pharynx. Cut in midbody or more posteriorly, the anterior part can regenerate the tail with adhesive organs in 2–3 weeks, although the animals were never observed to reach the original body length. The posterior part dies without regeneration after 1 day (Hagleithner 1946). Prolecithophora According to Pechlaner (1957), Plagiostomum lemani is reported to have a low regeneration capacity (Keller 1894), while Plagiostomum girardi can regenerate all amputated parts (Monti 1900b). Proseriata von Graff (1882) describes amputation experiments with Monocelis fusca, where posterior pieces resumed movement after 3–5 days. In M. fusca, Monocelis lineata and other monocelidids (Coelogynopora, Archilopsis), and also in otoplanids (Itaspiella, Bothriomolus), Giesa (1966) observed posterior regeneration of parts containing the brain in 3–4 days. Even pieces amputated just behind the brain can regenerate. The brain does not regenerate, though. In Otomesostoma auditivum, Pechlaner (1957) found promising regeneration potential, severely limited by a short life span of 9–10 months and low living temperatures (about 4°C). Gut, part of the gonads, and sensory organs can regenerate. First signs of pharynx, eye, and brain regeneration were observed after 9–15 weeks, but the animals died before regeneration was complete. The statocyst was not regenerated. Bothrioplanida While no extensive regeneration studies have been undertaken, Sekera (1911) found that Bothrioplana semperi is able to regenerate the pharynx posteriorly. Tricladida Dalyell (1814) noted that Planaria nigra (today: Polycelis nigra) could “almost be called immortal under the edge of the knife” (in Brøndsted 1969). However, not all triclad species are equally powerful regarding regeneration. Ŝivickis (1930) distinguished five groups of triclads according to their ability to regenerate a head. This classification was subsequently refined into eight types by Teshirogi et al. (1977). For the Phagocata velata type, the Dugesia dorotocephala type, the Polycelis auriculata type, and the Dendrocoelopsis lactea type 2 (Sapporo population), regeneration of a head is possible from any part of the body, but with varying probability of success. Animals of the Dendrocoelopsis lactea type 1 (Kuroishi population) fail to regenerate a head from tail pieces up to the male gonopore. The Dendrocoelopsis ezensis type and the Dendrocoelum lacteum type 1 are very similar in that they are generally only able to regenerate a head if amputated anterior to the pharynx, whereas the Bdelloura type (e.g., Bdelloura candida) completely lacks the ability to regenerate a head. While a vast amount of literature on regeneration in triclads is available, recent publications are not so much concerned with studying the regeneration capacities of different triclads, but more with unraveling cellular and molecular mechanisms of regeneration in only a few model triclads, such as Dugesia japonica and Schmidtea mediterranea. Rhabdocoela The typhloplanoid flatworm Mesostoma ehrenbergi cannot regenerate any organs, only approximately restore the original shape posteriorly (Steinmann and Bresslau 1913). Regeneration experiments with Mesostoma productum, Mesostoma lingua, and Mesostoma punctatum revealed that anterior regeneration of parts in front of the brain is possible, although often done imperfectly. The tail can regenerate; the more posterior the cut is made, the better is the regeneration. Brain and pharynx cannot be regenerated (Fulinski 1922). In Phaenocora unipunctata and Phaenocora megalops (also typhloplanoids), no regeneration capacity was found (Fulinski 1922). For the the dalyelliid Dalyellia millportiana and the kalyptorhynch Gyratrix hermaphroditus, Ruhl (1927a) remarks curtly that all animals died within 1 day after amputation and no regeneration being observable. Dalyellia viridis and Typhloplana viridata survive the amputation of the caudal-most part of the tail, but show no signs of regeneration, not even of the adhesive organs. In Rhynchomesostoma rostratum, another typhloplanoid, the amputated tail tip is regenerated, or at least a tail-like shape is restored (Hein 1928). Comparison of the regeneration capacity between free-living flatworm taxa While the experimental evidence for some groups is still patchy, we find three main types of regeneration capacity in free-living flatworms. Some taxa cannot regenerate at all, such as most rhabdocoels and some lecithoepitheliates. On the other end of the scope are species that can regenerate all organs, such as the acoel Convolutriloba, the catenulid Stenostomum, the macrostomorphan Microstomum, the polyclad Cestoplana, and many triclads. Most taxa however lie in between these extremes, lacking the ability to regenerate brain, eyes, pharynx, or statocyst (many acoels), but are able to regenerate gonads, copulatory organs, sucker, duo-gland adhesive systems, gut, and the anterior-most part in front of the brain. The regeneration capacity of M. lignano can be likened most closely to triclads of the Bdelloura group, which cannot regenerate a head (brain or eyes). Most polyclads, although often able to regenerate a pharynx, are also roughly comparable to Macrostomum regarding their regeneration capacity. Members of other diverse taxa like lecithoepitheliates and rhabdocoels are bearing resemblance with regeneration in Macrostomum as well. Such similarities can be found crossing the boundaries of the major flatworm taxa, making the regeneration capacity a feature of no immediate phylogenetic value. Close relatives like Microstomum and Alaurina (both macrostomids) surpass Macrostomum’s regeneration capacity by far. Asexual reproduction and regeneration Not surprisingly, the ability to regenerate is often linked to asexual reproduction (Fig. 1, Brøndsted 1969; Sánchez Alvarado 2000). In taxa lacking asexual reproduction, the regeneration capacity is generally less pronounced than in taxa with asexual reproduction, where regeneration is necessarily involved (Brøndsted 1969; Reuter and Kreshchenko 2004). As described in the section “Acoela”, in architomy, regeneration events follow fission, while in paratomy and budding, the term “pregeneration” seems to be more accurate, as organs are duplicated in the presence of the old organs in the new animal (zooid) before fission occurs. The notable difference between regeneration and pregeneration is the presence of a brain in the latter, which is regarded as a decisive organ for regeneration in many species (Henley 1974; Kreshchenko et al. 2001). It is interesting to note that the catenulid Stenostomum leucops, reproducing asexually by paratomy, often fails to regenerate a head if amputated artificially (van Cleave 1929). This finding hints at an actual difference between paratomy and architomy regarding the influence on regeneration capacity. Asexual reproduction in flatworms occurs in more basal taxa (Acoela, Catenulida, and Macrostomorpha) as well as in more derived taxa (Tricladida and Neodermata). Triclads, like acoels, show paratomy (Dugesia fissipara and Dugesia paramensis, Hyman 1951) as well as architomy (e.g., Dugesia tahitiensis, Peter et al. 2001). Just like regeneration capacity, asexual reproduction is too dispersed among diverse taxa to provide insight into phylogenetic relationships. Why can some flatworm species regenerate while others cannot? If a profound regeneration capacity is a plesiomorphic feature, there might have been a selection pressure against regeneration during evolution (Goss 1991; in Sánchez Alvarado 2000). Why would such a useful feature be reduced or abolished? Because regeneration is dangerous, as a plethora of evidence of double-headed or double-tailed flatworms testify. Such aberrant forms are easily produced in species that can regenerate (see section “Heteromorphoses”), but are not known from flatworms lacking regeneration capabilities. Other than as a side effect of asexual reproduction, the ability to regenerate is seemingly not useful enough to outweigh the inherent dangers. What are the chances that a predator spares a tail fragment that can regenerate to a complete animal in a couple of days if left alone? What are the chances that a disease only affects a part of the animal, but spares the rest to readily regenerate the damaged organs? A totipotent stem cell system out of control might be too high a risk for comparably little reward. This could be the reason why regeneration capabilities decreased in species that no longer propagate through asexual reproduction. In this light, it seems more likely that in evolutionary terms, asexual reproduction is a primitive and not a derived feature (Rieger 1986; Gremigni 1992; Reuter and Kreshchenko 2004), and that regeneration capability results from asexual reproduction as an added bonus. Regeneration can be seen as a form of asexual reproduction, only triggered by different stimuli. What is the decisive factor that determines the regeneration capacity of a species? Citing regeneration experiments with Dendrocoelum and Bdellocephala, Brøndsted (1969) comes to the conclusion “that the posterior part of the body in these two species simply lacks the potential to respond to head inducing substances”. This may also hold true for M. lignano, considering that in transversal amputations a head was never regenerated, but in longitudinal incisions a second head—in the presence of the original head—could be formed (Fig. 2c,d, Egger et al. 2006a,b). This experiment suggests that the ability to induce head formation, although present in adult M. lignano, is restricted to the head region only. Serotonin and possibly other substances related to the brain or nervous system might be key factors deciding on the regeneration capacity of flatworms. Fig. 2Heteromorphoses resulting from embryogenesis and regeneration. a Twin embryo of M. lignano, 8-cell stage. The two embryos are not separated by a dividing wall. In some cases, such embryos may grow to siamese twins. Diameter of individual embryos is about 150 μm. b Different animal than a found in culture dishes. The juvenile animals are connected at the level of the pharynx, possibly the result of two embryos growing together. c Adult M. lignano, 1 week after longitudinal incision (arrowhead). Only a small indentation is visible. d Same specimen as in c 29 days after longitudinal incision. A second head with brain, a single eye, and pharynx has emerged. Scale bars are 100 μm. c, d Same scale bar It is interesting to note that double-headed specimens of M. lignano can sometimes be observed in untreated culture dishes. These aberrant forms might rather be the result of twin embryos (Fig. 2a) that have grown together during embryonic development, than adults that were accidentally incised longitudinally (Sekera 1911; Steinmann and Bresslau 1913; Seilern-Aspang 1958). Ovaries restricting regeneration? The possible evolutionary origin and function of regeneration aside, what genetical, physiological, developmental, or environmental factors are responsible to determine to what extent an organism can regenerate? Even close relatives may differ vastly in regeneration capacity (e.g., Microstomum–Macrostomum), and also the same species in different developmental stages. Stenostomum alternates seasonally between asexual and sexual reproduction (protandric hermaphroditism). Asexual and male specimens of Stenostomum grande can regenerate all organs, but lose the ability to regenerate a head after the appearance of ovaries (van Cleave 1929). While the presence of ovaries and the loss of regeneration capacity might be coincidental, it appears possible that in S. grande ovaries are producing substances that inhibit regeneration. Could sexualization be directly affecting a general decrease of regeneration potential? At present, the molecular nature of such an inhibitory substance remains unclear. Vasa-related genes are expressed strongly both in gonads and in the early blastema in D. japonica and in M. lignano (Shibata et al. 1999; Pfister and Ladurner 2005), making an inhibitory function of vasa during regeneration improbable, at least in these two species. Also, contrary to the findings in Stenostomum, several triclads are known to regenerate all organs in the presence of gonads (e.g., S. mediterranea, Sánchez Alvarado 2003). Two possibilities may explain differences in the regeneration capacity in the presence of ovaries: either the ovaries in Stenostomum are producing a unique inhibitory substance that is not synthesized in Schmidtea, or Schmidtea is able to counteract the effects of such an inhibitory agent. Regeneration in juveniles If gonads, specifically ovaries, are indeed hindering regeneration in some species, then sexually immature juveniles could possibly be more potent regenerates than adults. In the rhabdocoel Mesostoma, young animals (the author does not specify their age) are reported to regenerate better than adults (Fulinski 1922). Contrastingly, in the acoel species H. giselae, no regeneration even of large juveniles (2.5 mm long) takes place (Steinböck 1967). In most cases, however, juveniles regenerate as well (or bad) as adults. An explanation is that gene expression for the building of gonads is already occurring in immature animals, thus diminishing possible differences in the regeneration capacity between juveniles and adults. Regeneration in juvenile triclads has been studied in B. punctata by Brøndsted, who found that animals recently hatched from cocoons were able to regenerate the head at the same rate as mature adults (Brøndsted 1942, 1969). Pechlaner (1957) studied the regeneration of freshly hatched animals, juveniles and mature adults in the proseriate O. auditivum. He did not observe significant differences in the regeneration capacity between these age groups. Neither juvenile nor adult dalyellids were found to regenerate (Hein 1928). In the lecithoepitheliate G. sphyrocephala, the regeneration capacity of juveniles corresponds to that of adults (Hagleithner 1946). The results for M. lignano support these findings and show that even very young animals (1 day after hatching) are capable of regeneration, indicating that the neoblast stem cell system is already present and functional as found in adults. Accordingly, BrdU labeling of 1-day-old hatchlings revealed a similar pattern of S-phase cell distribution as in adults, with the majority of labeled cells along the sides of the animal, and no labeled cells in the rostrum. Just like adults, juveniles were not able to regenerate a head (Egger et al. 2006a,b). Similarities between postembryonic development and regeneration However similar, regeneration has some characteristics not shared with postembryonic development. Unlike postembryonic development, regeneration in flatworms is triggered by a disruptive event like fission or amputation (but not by chemical destruction of tissue, burning, or high voltage; see Brøndsted 1969). As in other flatworms, in M. lignano, the wound is soon closed after amputation by contraction of the ring musculature minimizing the wound surface, and by flattening of the surrounding epidermal cells (Salvenmoser et al. 2001). In the next days, undifferentiated cells (neoblasts) accumulate at the wound site, forming a distinct blastema, covered by new and old epidermal cells. Subsequently, differentiation of microorgans and organs becomes prevalent within the blastema, superseding the blastema proper (Egger et al. 2005, 2006a,b). Epidermal cells are not undergoing mitoses, as new epidermal cells and blastema cells are derived from neoblasts located in the mesodermal space. Different to triclads, neoblasts are proliferating also within the blastema, not only at the blastema border (Egger et al. 2005). While wound closure and blastema formation are regeneration-specific events, similarities between regeneration and postembryonic development become prevalent starting with differentiation in the (former) blastema. In extreme cases, where only a small piece remains for regeneration (e.g., in pharynx-level regenerates, Egger et al. 2006a,b), resemblance with postembryonic development is particularly striking (Fig. 3). The remaining organs in these regenerates undergo extensive morphallactic changes to accommodate to the new body proportions (Fig. 3n–o in Egger et al. 2006a). Fig. 3Comparison of M. lignano hatchlings with posterior regenerates amputated at the pharynx level. b and d are depicting the same individual. a Hatchling up to 24 h old. b Pharynx level regenerate 7 days after amputation. c 7-day-old hatchling. d Pharynx level regenerate 14 days after amputation. Scale bar for all subpanels is 100 μm One week after amputation, pharynx-level regenerates of M. lignano are confronted with similar challenges as freshly hatched juveniles: both have to build gonads and the male and female genital organs, grow a full set of duo-gland adhesive glands, and increase their size by five or six times (Figs. 3a,b, and 4). It seems plausible, that after wound healing and restoration of the basic functionality of the gut, pharynx-level regenerates are making use of similar or same developmental pathways that are used in hatchlings (Fig. 4) (cooption, Sánchez Alvarado 2000). This view is supported by the ability of hatchlings and juveniles to regenerate just like adults, and by the similarity of the stem cell system in hatchlings, juveniles, and adults (Ladurner et al. 2000; Egger et al. 2006a,b). Vasa gene expression is similar in hatchlings and in gonad-recovering regenerates as well (Pfister and Ladurner 2005). Also, the time needed for developing to a full-grown adult in M. lignano is about the same for hatchlings (Morris et al. 2004) and 1-week-old pharynx-level regenerates, that is about 2 weeks (Fig. 3 and Egger et al. 2006a,b). In the proseriate M. fusca, the regeneration of small anterior pieces also goes through stages similar to postembryonic development (Giesa 1966), which is observed in triclads as well (Baguñà 1998). Starved adults, reducing their size and breaking down the gonads (Nimeth et al. 2004), are also likely recapitulating their postembryonic development while growing back to full size (Fig. 4). Not only is regeneration akin to postembryonic development, amputated animals can even be rejuvenated by regeneration (Fig. 4 and Haranghy and Balázs 1964 for triclads, Egger et al. 2006a,b for M. lignano). While untreated animals were found to live for about 10 months in culture conditions, they were outlived by animals amputated 45 times over a period of 20 months, effectively doubling their life span so far (Egger et al. 2006a,b; Egger, unpublished observations). At the time of writing, many of these repeatedly amputated animals are still alive. Fig. 4Similarities between (post-)embryonic development, regeneration, and starvation considering as example M. lignano. Successive developmental stages from left to right, from embryonic to postembryonic development (formation of gonads and genital organs) until adulthood. Regeneration If an adult animal is amputated just posterior of the pharynx, the anterior piece will pass stages similar to early juveniles and subadults before regeneration to a normal adult is completed. The same animal can be amputated repeatedly, each time seemingly repeating parts of its postembryonic development. Repeated regeneration was also shown to have a rejuvenation effect. A connection between the formation of organ primordia during embryonic development and the organ differentiation in regenerating animals is possible, but has not been shown yet for M. lignano. Some developmental pathways are likely shared between late embryos and regenerates as well, e.g., the initial building of microorgans like the duo-gland adhesive systems. Both in embryos and posterior regenerates, these organs are built from scratch. Starvation Animals starved for 30 days or longer break down their gonads and genital organs shrink, but can recover and regain sexual prowess after feeding. Starvation leading to an extended life span was shown for triclads and is also likely to occur in M. lignano A fountain-of-youth effect through starvation was shown to occur in triclads (Haranghy and Balázs 1964), but was not yet proven for M. lignano (Fig. 4). Embryonic development and regeneration Early embryogenesis is vastly different between diverse free-living flatworm taxa (for recent publications, see Jondelius 2004 for nemertodermatids; Ramachandra et al. 2002 for acoels; Morris et al. 2004 for macrostomorphans, Younossi-Hartenstein and Hartenstein 2000 for polyclads, Cardona et al. 2005 for triclads, and Younossi-Hartenstein and Hartenstein 2001 for rhabdocoels). Even comparably close relatives show significant differences in their early embryonic development. Lacking vitellaria and therefore producing entolecithal eggs, Macrostomorpha and Polycladida are the only archoophoran Rhabditophora, placing them at the basal-most positions in this largest flatworm taxon (Ehlers 1985; Rieger 1996). Nonetheless, their early embryonic development starts to deviate after about the third cleavage (Fig. 5). If developmental pathways from the embryonic development are reused in regeneration, it is most likely from the formation of organ primordia and later stages (Fig. 4). Cardona et al. (2005) point out the similarities between embryogenesis and early regeneration in triclads concerning the building of a provisional epidermis and the development of muscles and nerves, suggesting that not only postembryonic, but also embryonic developmental pathways are utilized during regeneration, a viewpoint shared with Vannini (1966). A comparison of gene expression patterns between regenerates (see Reddien et al. 2005 for triclads), late embryos and hatchlings seem to be a worthwhile task. Fig. 5Early embryonic development of two basal rhabditophorans. a The 12-cell stage of the polyclad Pseudostylochus intermedius. Egg without eggshell. b 12-cell stage of the macrostomorphan M. lignano. Different to the clear spiral cleavage of polyclads, M. lignano forms the so-called hull cells after the third cleavage to enclose the rest of the embryo. The diameter of the embryos is about 150 μm Early blastomere ablation in polyclads has revealed a determinative development as in other spiralians (Boyer 1986), the fate of blastomeres depending on micromere–macromere interaction (Boyer 1989). Polyclads do not have an invariant cell lineage, however (Boyer 1992). Polyclad embryos can be obtained devoid of an eggshell (Fig. 5a) and still develop normally in a culture dish treated with antibiotics. Regeneration experiments with polyclad embryos may thus be relatively easy to carry out and could provide additional clues about the requirements for regeneration in flatworm embryos. Also, the regeneration capacity of polyclad larvae, as compared to hatchlings of directly developing polyclads, would be interesting to know. A connection between regulation in early embryonic development and regeneration capacity does not seem to exist. This notion is supported by regeneration experiments with embryos of the triclad P. nigra. Regeneration does not take place before organ systems, such as the nervous system, reach a certain degree of differentiation in the embryo, so that stage five embryos are unable to regenerate, while stage six embryos with a more differentiated nervous system readily regenerate lost parts (Le Moigne 1966). This author held that the nervous system plays a critical role in the formation of a blastema during regeneration. Nervous system and role of the brain for regeneration The nervous system, especially the brain, has long been considered to play a key role in regeneration (Kreshchenko et al. 2001). In many free-living flatworms, posterior parts can be regenerated, while the brain cannot. In anterior regenerates of M. lignano missing the brain, a blastema-like structure was observed, which was unable to give rise to all missing organs. The more posterior the animals were amputated, the less pronounced was this abortive anterior blastema (Egger et al. 2006a,b). In Macrostomum, are the brain and the nerve cell clusters associated with the pharynx (Fig. 6) required to release factors needed for the regeneration of brain, eyes, and pharynx? The study of gene expression in head-amputated regenerates of the triclads D. japonica and S. mediterranea revealed necessary genes for regeneration of the brain (Cebrià et al. 2002a; Koinuma et al. 2003; Cebrià and Newmark 2005). Also, with nou-darake a factor was identified that restricts brain tissue to the head region of D. japonica. If nou-darake is interrupted by RNA interference, brains appear in all body regions (Cebrià et al. 2002b). In species that are not able to regenerate a brain, nou-darake or a similar factor might be expressed not only in the brain region but also throughout the body, thus preventing the formation of a new brain even if the old brain was lost. Fig. 6Immunocytochemical stainings of the nervous system of M. lignano. Confocal images. a GYIRFamidergic immunoreactivity. Clusters of immunoreactive cells are found lateral to the pharynx and in front of the eyes. b, c Serotonergic immunoreactivity in the head region. b Full stack of confocal images. Note the stained cell clusters lateral to the pharynx and in front of the eyes. c Subset of image stack in b. In this focal plane, stained cells anterior to the eyes become apparent. Arrowheads denote the level of the eyes In M. hystricinum marinum and M. pusillum, large clusters of serotonergic cells have been found besides the posterior part of the pharynx (Ladurner et al. 1997). These results were corroborated for M. lignano by staining of the serotonergic and the GYIRFamidergic nervous system (Fig. 6). Amputees cut in the middle of the pharynx lose this cluster of nerve cells, as well as a large part of the pharynx, which is known to be a critical structure for regeneration in some triclads (Brøndsted 1969) and polyclads (Ishida 1998). Ladurner et al. (1997) held that the distribution of serotonergic nerve cells is correlated with the regeneration capacity. In species where serotonergic neurons are restricted to the head area, the posterior part of the animals cannot regenerate the head. In the asexually reproducing species M. lineare and S. leucops, serotonergic nerve cells are found over the whole length of the main nerve cords. The inability of M. lignano to regenerate eyes after transversal amputation or oblique amputation between the eyes, is possibly linked with a severe brain damage caused by these amputations, and the loss of strongly immunoreactive cells in front of the eyes (Fig. 6c). In two cases of longitudinal incisions, lost eyes were regenerated or supernumerary eyes were built—but in these cases, the brain was only slightly injured. In a series of experiments, Lender (1950, 1951a–c) and Wolff and Lender (1950a,b) stress the importance of the head ganglia for the regeneration of eyes in P. nigra. If the brain was repeatedly excised or X-irradiated, no regeneration of eyes took place: an intact brain was obviously required to induce eye regeneration. These results were corroborated for Dugesia (now Schmidtea) lugubris by Török (1958). However, L. littoralis, L. saxicola, and P. litoricola can regenerate eyes even if the brain was excised (Morgan 1905; Olmsted 1922a). Vannini (1965) observed in S. lugubris and P. nigra that the testes do not regenerate in the absence of a brain. Ghirardelli (1965) found that if a two-headed Dugesia was decapitated on one side, the testes on the same side degenerated. Contrastingly, the gonads in M. lignano are viable for an extended period of time after decapitation or oblique amputation between the eyes, and are only reduced if the pharynx is lost as well, causing malnourishment and starvation (Egger et al. 2006a,b). This may be an indication that the building of the gonads is not under the primary control of the brain in M. lignano, just like the tail plate can be built without the presence of a head. The regeneration of a tail in headless regenerates is possible in M. lignano as well as in triclads of the Dendrocoelum group (Brøndsted 1939, 1969), in the polyclads L. littoralis, L. saxicola, and P. litoricola (Morgan 1905; Olmsted 1922a), and also in the rhabdocoel Mesostoma (Fulinski 1922). Therefore, the head in all of these organisms is not required to control the regeneration of the tail. In Cestoplana and T. brocchi, the pharynx can regenerate in the absence of the brain (Child 1905c; von Levetzow 1939); in Cestoplana often a secondary pharynx is built in the presence of the old pharynx (Child 1905c). Also, the regulation of tissue breakdown (e.g., gonads) and body size reduction during starvation is possible in decapitated pieces, so the brain is not responsible for this regulation as well. Headless M. lignano with intact pharynx are able to feed, in accordance with the polyclads Notoplana acticola and Planocera gilchristi (Koopowitz et al. 1976). These animals are able to identify and devour food in the absence of a brain, while Enchiridium punctatum, a polyclad possessing a tubular instead of a plicate pharynx, cannot feed without the brain (Koopowitz et al. 1976). Similarly, some triclads, also equipped with a tubular pharynx, are not able to feed if the brain was amputated (Hyman 1951). Brainless M. ehrenbergi cannot feed but can move normally (Steinmann and Bresslau 1913), whereas brainless Phaenocera is reported to behave similarly to normal animals (Fulinski 1922). G. sphyrocephala, different to M. lignano, cannot adhere to the substrate after decerebration (Hagleithner 1946). The main function of the brain in M. lignano is probably to be found in directed motion (e.g., light avoidance, predator avoidance; see also Koopowitz et al. 1976 and literature therein), satiation reflex (Koopowitz et al. 1976), mating behavior (Scharer et al. 2004), and chemical analyzation of the medium. The totipotency of neoblasts Pharynx-level regenerates of M. lignano have shown that the regeneration of gonads is possible after removal of all gonadal tissue. The viability of the regenerated gonads has been demonstrated by breeding experiments with fully regenerated animals. Besides the gonads, also the gut and the copulatory organs have been rebuilt de novo in pharynx-level regenerates (Egger et al. 2006a,b). In similar experiments, Morgan (1902) removed all gonads in Schmidtea lugubris, and the animals regenerated to fully functional sexual organisms. It is still possible that the remaining neoblasts in the amputee are only pluripotent and not totipotent, different types of neoblasts being competent for different (groups of) tissues. A convincing experiment to prove the totipotency of neoblasts would be the injection of a single neoblast into a lethally X-irradiated animal that is to be amputated some weeks after the injection. In case the animal survives, and is able to restore all removed organs, then either redifferentiation of cells would have taken place, or (at least some) adult neoblasts would be totipotent. A similar experiment, with the injection of 20,000–24,000 cells enriched in neoblasts (30–88% neoblasts) into X-irradiated S. mediterranea, was performed by Baguñà et al. (1989). This experiment was not replicated in the literature, but the results made highly probable that neoblasts as a whole are totipotent in routine cell turnover, leaving open the questions if this holds also true for regeneration and if this totipotency applies to one single cell type or to several different types of stem cells. Are neoblasts in species with limited regeneration potential not totipotent, or are they just not receiving the right stimulus or guidance for regeneration? Supporting the latter notion is the ability of Macrostomum to duplicate heads after longitudinal incision (Fig. 2d and Egger et al. 2006a,b), while decapitated animals were never observed to regenerate a head (Egger et al. 2006a,b). In this species, it seems that stimuli from the brain are required to duplicate or regenerate the brain. Is there a specific subset of predetermined “brain” neoblasts in species that cannot regenerate a brain? These neoblasts could be located in the vicinity of the brain and therefore lost by amputation of the brain. Brain excision experiments in polyclads, where only the encapsuled brain was removed (von Levetzow 1939; Koopowitz et al. 1976), do not support the existence of such brain-specific neoblasts, as these animals were not able to regenerate the brain. Also, amputation experiments in M. lignano, where only brain and the neoblast-free rostrum were amputated, showed no brain regeneration (Egger et al. 2006a,b). While neoblasts are required for regeneration in general, they do not determine whether a species is able to regenerate a specific organ such as the head, as was found with grafting experiments in triclads (Brøndsted 1969). Can a single neoblast from an adult develop to a whole organism in an adequate environment, just like a zygote? For the parasitic flatworm Taenia crassiceps, it was shown that single cells of trypsinated cysticerci are able to produce complete cysticerci in the mouse host (Toledo et al. 1997). So in these cestodes, a single totipotent cell, other than a zygote, is able to recover a whole animal. No such experiment is known for free-living flatworms. Besides the difficulty in providing a suitable culture medium, distinct differences between a single neoblast and a zygote are yolk content or surrounding yolk cells and maternal determinants that are present in eggs, but not in neoblasts. Do (all or some) neoblasts have embryonic characteristics, as some authors held (e.g., Keller 1894)? When and where do neoblasts origin in the embryo? Polyclads with their clear and traceable embryonic development appear to be suitable flatworms for investigating these questions. Heteromorphoses Hetermorphoses, i.e., the occurrence of supernumerary heads or tails, in species that cannot regenerate an amputated head, have shown that the regeneration potential of neoblasts is not to be equated with the regeneration capacity of the species. Heteromorphoses are also a placative warning sign for the dangers coming with the ability to regenerate (all) organs. Last but not the least, hetermorphoses in free-living flatworms show that the duplication of axes is not only possible during embryonic development (Fig. 2b, siamese twins), but also in adults, even after comparably minor surgery. The M. lignano specimen depicted in Fig. 2c,d shows that a longitudinal incision in the anterior part of the animal can establish a second anteroposterior body axis parallel to the original axis. The new anteroposterior axis reaches from the rostrum tip to the caudal tip and duplicates all organs including head, pharynx, female genital opening, and male copulatory organ. Even the breadth of the animal approximates that of two single specimens. Over time, however, the animal degenerated in both halves losing all eyes, and was not able to recover to a viable state (Egger et al. 2006a,b). A decreased viability in animals with heteromorphoses was also observed in Macrostomumappendiculatum and Bothrioplana bohemica (Sekera 1911), as well as in S. leucops (Ruhl 1927b). In triclads, duplicated or even multiple heads (e.g., 10-headed D. lacteum, Lus 1924) and tails have been generated and studied by a great number of researchers (see Brøndsted 1969). Randolph (1897) made longitudinal and lateral incisions in Planaria maculata [today: Dugesia (Girardia) tigrina] leading to additional eyes and pharynges. It was found that when a duplicated head or tail was amputated close enough to the second main axis, regeneration was inhibited by the presence of the remaining head or tail (Rand and Mildred 1926). In the polyclad Leptoplana, double-heads and double-tails were experimentally produced by longitudinal incision (Child 1905a). With the same method, tails, but not heads, could be duplicated in the rhabdocoel Mesostoma (Fulinski 1922) and in the acoel P. caudatus (Keil 1929). Constrastingly, heads, but not tails could be duplicated by incision in the catenulid S. leucops (Ruhl 1927b). Another type of heteromorphoses is created by provoking a body axis reversed to the original body axis. This was found in M. lignano after longitudinal incision in the head, that induced the creation of a tail plate appendage in the anterior body region (Egger et al. 2006a,b). Agata et al. (2003) suggested that one of the main functions of a blastema in triclads is producing positional signals and inducing intercalary regeneration between stump and blastema. An incision may cause the building of a very small blastema at the wound site, which provides anterior or posterior positional signals. Depending on the signal the blastema produces, either a head or a tail plate is subsequently formed. Incisions may also have a long-range effect, as was observed by Randolph (1897) in triclads, where outgrowths appeared at body regions not affected by incision. A similar phenomenon as in M. lignano was observed by Steinböck (1967) in H. giselae, where an animal was divided into 10 pieces, one of which developed a tail-like structure at the anterior end besides the head. Just as was observed in M. lignano (Egger et al. 2006a,b), the anterior tail in H. giselae was gradually moving along the side of the animal towards the posterior end, when after 36 days the additional tail was located at the posterior tip of the animal. Conclusion A pronounced regeneration capacity and the existence of possibly totipotent stem cells in adults are two key features of many flatworms not found in this combination in any other bilaterian taxa. It could be shown that regeneration follows similar principles in many different taxa of free-living flatworms, even if the regeneration capacity can vastly differ. Such comparisons make it possible to recognize unique patterns, and also to form general ideas how, why, and when regeneration works. For this reason, it appears to be valuable to establish several flatworm model organisms from various taxa, to address old and upcoming questions regarding regeneration and the neoblast stem cell system. One of these comparably new model flatworms is M. lignano, a small but versatile worm that facilitates regeneration studies on the cellular and molecular level.

Eur_J_Nucl_Med_Mol_Imaging-3-1-1998890

Treatment of patients with gastro-entero-pancreatic (GEP) tumours with the novel radiolabelled somatostatin analogue [177Lu-DOTA0,Tyr3]octreotate

Medical treatment and chemotherapy are seldom successful in achieving objective tumour reduction in patients with metastatic neuroendocrine tumours. Treatment with the radiolabelled somatostatin analogue [90Y-DOTA0,Tyr3]octreotide may result in partial remissions in 10–25% of patients. The newer analogue [DOTA0,Tyr3]octreotate (octreotate) has a ninefold higher affinity for the somatostatin receptor subtype 2 as compared with [DOTA0,Tyr3]octreotide. Also, labelled with the beta- and gamma-emitting radionuclide 177Lu, it has proved very successful in achieving tumour regression in animal models. The effects of 177Lu-octreotate therapy were studied in 35 patients with neuroendocrine gastro-entero-pancreatic (GEP) tumours who underwent follow-up for 3–6 months after receiving their final dose. Patients were treated with doses of 100, 150 or 200 mCi 177Lu-octreotate, to a final cumulative dose of 600–800 mCi, with treatment intervals of 6–9 weeks. Nausea and vomiting within the first 24 h after administration were present in 30% and 14% of the administrations, respectively. WHO toxicity grade 3 anaemia, leucocytopenia and thrombocytopenia occurred after 0%, 1% and 1% of the administrations, respectively. Serum creatinine and creatinine clearance did not change significantly. The effects of the therapy on tumour size were evaluable in 34 patients. Three months after the final administration, complete remission was found in one patient (3%), partial remission in 12 (35%), stable disease in 14 (41%) and progressive disease in seven (21%), including three patients who died during the treatment period. Tumour response was positively correlated with a high uptake on the octreoscan, limited hepatic tumour mass and a high Karnofsky Performance Score. Because of the limited efficacy of alternative therapies, many physicians currently adopt an expectant attitude when dealing with patients with metastatic GEP tumours. However, in view of the high success rate of therapy with 177Lu-octreotate and the absence of serious side-effects, we advocate its use in patients with GEP tumours without waiting for tumour progression. Introduction Neuroendocrine gastro-entero-pancreatic (GEP) tumours, which comprise pancreatic islet cell tumours, non-functioning neuroendocrine pancreatic tumours and carcinoids, are usually slow growing. If the tumour is localized, the therapy of choice is surgery. When a metastatic tumour causes a syndrome by hormonal overproduction (i.e. carcinoid syndrome, hypergastrinaemia), treatment with somatostatin analogues results in symptomatic relief in most cases. In terms of objective tumour reduction (complete and partial remission), however, treatment with somatostatin analogues is seldom successful, whether or not it is given in combination with interferon-α [1, 2, 3]. A new development in cytoreductive therapy for GEP tumours is the use of radiolabelled somatostatin analogues. Initial studies with high doses of [111In-DTPA0]octreotide (111In-octreotide; Octreoscan) in patients with metastatic neuroendocrine tumours were encouraging, although partial remissions were exceptional [4, 5]. However, 111In-coupled peptides are not ideal for peptide receptor radionuclide radiotherapy (PRRT) because of the small particle range and therefore short tissue penetration. Therefore, another radiolabelled somatostatin analogue, [90Y-DOTA0,Tyr3]octreotide, was developed. Using this compound, partial remissions have been reported in 10–25% of patients with neuroendocrine tumours [6, 7, 8]. Recently, it was reported that the somatostatin analogue [DOTA0,Tyr3]octreotate has a ninefold higher affinity for the somatostatin receptor subtype 2 as compared with [DOTA0,Tyr3]octreotide [9]. Also, labelled with the beta- and gamma-emitting radionuclide 177Lu, this compound was shown to be very successful in achieving tumour regression and animal survival in a rat model [10]. In a comparison in patients, we found that the uptake of radioactivity, expressed as a percentage of the injected dose of [177Lu-DOTA0,Tyr3]octreotate (177Lu-octreotate), was comparable to that after 111In-octreotide for kidneys, spleen and liver, but was three- to fourfold higher for four of five tumours [11].We concluded that 177Lu-octreotate potentially represents an important improvement because of (a) the higher absorbed doses that can be delivered to most tumours with about equal doses to potentially dose-limiting organs and (b) the lower tissue penetration range of 177Lu as compared with 90Y, which may be especially important for small tumours. In this study we present the first data on the side-effects as well as the antitumoral effects of 177Lu-octreotate therapy in 35 patients with GEP tumours, who had a follow-up of 3–6 months after receiving their final dose. Materials and methods Patients Thirty-five patients with GEP tumours were studied. All patients had tumour uptake on the octreoscan preceding the therapy that was at least as high as the uptake in the normal liver tissue. None of the patients had received prior treatment with other radiolabelled somatostatin analogues. Eight patients used Sandostatin s.c.; this medication was discontinued from 1 day before to 1 day after the treatment. Prerequisites for treatment were: Hb ≥6 mmol/l, WBC ≥2×109/l, platelets ≥80×109/l, creatinine ≤150 μmol/l and Karnofsky Performance Score ≥50. All patients gave written informed consent to participate in the study, which was approved by the medical ethical committee of the hospital. Methods [DOTA0,Tyr3]octreotate was obtained from Mallinckrodt, St Louis, Mo., USA. 177LuCl3 was obtained from NRG, Petten, the Netherlands and Missouri University Research Reactor (MURR), Missouri, Mo., USA, and was distributed by IDB, Baarle-Nassau, the Netherlands. 177Lu-octreotate was prepared as described previously [11]. Granisetron 3 mg was injected i.v. and an infusion of amino acids (lysine 2.5%, arginine 2.5% in 1 l 0.9% NaCl; 250 ml/h) was started 30 min before the administration of the radiopharmaceutical and lasted up to 3.5 h afterwards. The radiopharmaceutical was co-administered via a second pump system. The treatment doses of 100 mCi were injected in 20 min and those of 150 and 200 mCi were injected in 30 min. The interval between treatments was 6–9 weeks. Patients were treated up to a cumulative dose of 750–800 mCi (27.8–29.6 GBq) (corresponding to a radiation dose to the bone marrow of 2 Gy) [11], unless dosimetric calculations indicated that the radiation dose to the kidneys would then exceed 23 Gy; in these cases the cumulative dose was reduced to 600–700 mCi. Routine haematology, liver and kidney function tests, hormone measurements and serum tumour markers were measured 1 week prior to each therapy, as well as at follow-up visits. EORTC quality of life forms (QLQ-C30) [12] were filled out by the patients at each visit. CT or MRI scanning was done within 3 months before the first therapy, and 6–8 weeks, 3 months and 6 months after the last treatment. In vivo measurements. The tumours on the CT or MRI scans were measured and scored according to the WHO solid tumour response criteria. The uptake on the pretreatment octreoscans was scored visually on planar images on a 4-point scale: grade 1, lower than normal liver tissue uptake; grade 2, equal to normal liver tissue uptake; grade 3, greater than normal liver tissue uptake; grade 4, higher than normal spleen/kidney uptake. Statistics Analysis of variance (ANOVA), paired t tests and chi-square tests were used. P values <0.05 were considered significant. Results The study population comprised 14 men and 21 women with a mean age of 54 years(range 19–78 years). Twelve had carcinoid tumour, 12 neuroendocrine pancreatic tumour, 8 neuroendocrine tumour of unknown origin and 3 gastrinoma. Twelve patients had been operated in the past, 1 had had radiotherapy, 3 had had chemotherapy and 14 had been treated with octreotide (Sandostatin). Sixteen of the 35 (46%) patients had documented progressive disease within 1 year before the start of the therapy. Cycle doses were 100 mCi in seven patients, 150 mCi in 14 and 200 mCi in the remaining 14. Higher dosages were not administered, since 200 mCi 177LuCl3 is typically bound to 180–300 μg [DOTA0,Tyr3]octreotate, and higher doses would result in a lower percentage tumour uptake owing to receptor saturation. In 30 patients, the final intended cumulative dose of 600–800 mCi was administered. Three of the five remaining patients had progressive disease and died before completing their treatment; the other two patients, who were both elderly, stopped their treatment after reaching a cumulative dose of 600 mCi because of the burden they felt the treatment to be. Nausea and vomiting within the first 24 h after the administration were present in 30% and 14% of the administrations, respectively, and were independent of the administered dose. Some abdominal pain was noticed during 11% of the administrations, especially in patients with liver metastases. Increased, mild hair loss was present in 17 of the 35 patients; hair regrowth had occurred at follow-up 3 and 6 months after the final administration. WHO toxicity grade 2 or 3 anaemia (Hb 4.95–6.2 or 4.0–4.9 mmol/l, respectively), leucocytopenia (WBC 2.0–2.9 or 1.0–1.9×109/l, respectively) and thrombocytopenia (platelets 50–74.9 or 25.0–49.9×109/l, respectively) occurred after 8% and 0%, 5% and 1%, and 3% and 1% of the administrations, respectively. Toxicity grade 2 or 3 leucocytopenia or thrombocytopenia occurred in two out of three (67%) patients who had had previous chemotherapy, as against seven out of 32 (22%) patients who had not. Mean haematological parameters rose again during the follow-up after the final administration. Serum creatinine, creatinine clearance and serum HbA1c did not change significantly (data not shown). In patients without thyroid hormone medication, serum TSH and fT4 levels did not change. In women, serum LH and FSH concentrations did not change significantly; in men, serum testosterone decreased and serum LH concentrations increased significantly. Also inhibin-B concentrations decreased and serum FSH levels increased significantly. The effects of the therapy on tumour size were evaluable in 34 patients. Three months after the final administration (on average 9 months from the start of the treatment), a complete remission (CR), evaluated with CT scanning, MRI and somatostatin receptor imaging, was found in one patient (3%), partial remission (PR) in 12 (35%), stable disease (SD) in 14 (41%) and progressive disease (PD) in 7 (21%), including the three patients who died during the treatment period (Table 1) (Figs. 1, 2). Follow-up evaluation 6 months after the final therapy was available for 19 of the 34 patients. All seven patients who had PR after 3 months still had PR after 6 months; in 10 of the 12 patients with SD, the evaluation was unchanged, whereas one had a minimal response (MR) and one had PD.Table 1. Tumour responses in 34 patients, 3 months after the final administration of 177Lu-octreotate. Three patients with PD died before reaching their final doseTumour typeResponseTotalCRPRSDPDCarcinoid4 (33%)6 (50%)2 (17%)12NE pancreas1 (8%)1 (8%)7 (58%)3 (25%)12NE unknown origin4 (57%)1 (14%)2 (29%)7Gastrinoma3 (100%)3Total1 (3%)12 (35%)14 (41%)7 (21%)34Fig. 1.A–C Planar scans of the abdomen, 3 days after the injection of 200 mCi 177Lu-octreotate in a patient with liver metastases of an operated neuroendocrine pancreatic tumour. A After the first treatment; B after the second treatment; C after the fourth treatment. Note the loss of intensity of uptake in the liver lesions (arrows in A). This sign virtually always indicates a tumour volume response. D, E CT scans of the same patient: before treatment (D) and 3 months after the last treatment (E). Tumour (arrows in D) is not demonstrated on E. Neither MRI nor octreoscan could demonstrate definite tumour deposits at that timeFig. 2. A, B MRI scans in a woman with hepatic metastases of a gastrinoma. Before treatment (A) she was inoperable because of the size and localization of the metastases. Three months after completion of her treatment with 800 mCi 177Lu-octreotate she had a PR (B). Note also that the hypertrophy of the gastric wall (arrow in A) had diminished. Concomitantly, serum concentrations of gastrin (closed dots) and chromogranin-A (open dots) had decreased markedly (C). At 6 months the MRI showed no further regression. The patient underwent left partial hepatectomy and the two right-sided lesions were drained and injected with alcohol. MRI 3 months after surgery showed further regression of the right-sided lesions Tumour response (CR or PR) was significantly more frequent in patients whose tumours had a high uptake on the octreoscan [6/7 (86%) with grade 4 uptake vs. 7/27 (26%) with grade 2 or 3 uptake; chi-square test: P<0.05]. Five out of the seven patients (71%) with PD had hepatomegaly and diffuse liver metastases vs 6 out of 27 (22%) with CR, PR or SD (chi-square test: P<0.05). Tumour response was more frequent in patients with documented PD within 1 year before the start of the treatment (8/15; 53%) than in those without (5/19; 26%), although this difference was not statistically significant (P=0.11; chi-square test). Of the 15 patients with documented PD before the start of the treatment, 3 (20%) had SD at follow-up. Patients who had a Karnofsky Performance Score of less than 80 before the treatment more frequently had PD (5/8; 63%) than those who had higher scores (2/26; 8%) (P<0.01; chi-square test). The serum tumour marker chromogranin-A was elevated in 29 patients. During the treatment and follow-up, there was a clear decrease in serum chromogranin-A concentrations in patients with PR or CR, whereas these concentrations were virtually unchanged in patients with SD, and initially showed an increase in patients with PD (Fig. 3).Fig. 3. Serum chromogranin-A concentrations during and after therapy in patients with PR, SD or PD. The reduction in the number of patients during the course of the follow-up was due to death or missing values. Note the logarithmic y-axis The patient-assessed quality of life, according to the EORTC-QLQ30 questionnaire, was evaluated in 25 patients. Ten patients were excluded because of progressive disease/death (n=5) or missing forms during the follow-up period (n=5, foreign patients). Scores before the start of the treatment, after receiving 400–600 mCi 177Lu-octreotate and at follow-up 3 months after the final treatment were evaluated. There were no significant differences for functional scales or single symptom scales. The global health scale, on which patients were asked to assign marks regarding both their general health and their quality of life, and which ranged from 0 to 100, was judged as higher than 70 by nine patients (36%) before the start of the treatment, by 17 (68%) patients during the treatment and by 16 (64%) patients after the treatment (P<0.05; chi-square test). Discussion There are few treatment options for metastatic GEP tumours. The use of radiolabelled somatostatin analogues for tumour regression is a promising new development. With [90Y-DOTA0,Tyr3]octreotide, PR (or CR) has been reported in 10–25% of patients with neuroendocrine tumours [6, 7, 8]. In the present study we found objective tumour shrinkage in 38% of the patients, but PD before the start of the treatment was documented in only 46% of the patients. This last fact is important, because documented PD was present in more than 80% of patients in two of the reported series treated with [90Y-DOTA0,Tyr3]octreotide [6, 8]. As with chemotherapy, an objective tumour response after PRRT is more likely in patients with (fast) growing tumours. Indeed, in our series we also found a trend towards a more favourable treatment outcome in patients with documented PD before the start of the treatment. It may therefore be concluded that the percentage of patients with significant tumour shrinkage in our study might have been even higher had the percentage of patients with PD prior to treatment been comparable to that in studies with [90Y-DOTA0,Tyr3]octreotide. The side-effects of treatment with 177Lu-octreotate are few and mostly transient, with mild bone marrow depression as the most common finding. Neither renal nor pituitary function deteriorated in any of our patients. Other side-effects can be ascribed to the radiation dose to the testes in men. This dose leads to significantly lower serum testosterone and inhibin-B levels which in turn give rise to higher serum LH and FSH concentrations, thereby substantiating that the pituitary function is unimpaired. The patient-assessed global health score improved in 30% of patients during the treatment and the follow-up period. This is an important finding which reflects the improvement in patient well-being and stresses the scarcity of side-effects as perceived by patients. The fact that other scores mainly addressing symptoms did not change significantly is likely due to the diversity of symptoms between patients, the use of Sandostatin by symptomatic patients and the small size of the patient group. Many physicians adopt an expectant attitude when dealing with patients with metastatic GEP tumours owing to the low success rates of chemotherapy protocols. However, given that treatment with 177Lu-octreotate resulted in PR (or CR) in 38% of our patients, this attitude may be questioned. Although there was a (statistically non-significant) tendency towards more frequent tumour regression in patients who had documented PD before the start of the treatment, those of our patients who had an objective tumour response more frequently had a limited tumour load. This implies that waiting for tumour progression might place patients in a worse position, as PD during or after treatment was more frequent in patients with an extensive tumour load, especially in the liver. We therefore advocate treatment with177Lu-octreotate at an earlier stage of metastatic disease, when even CR may be possible. Another argument in favour of early treatment is that neuroendocrine tumours can dedifferentiate in the course of the disease, and lose their somatostatin receptors, making treatment with radiolabelled somatostatin analogues impossible. If serious side-effects of treatment with 177Lu-octreotate remain absent during longer patient follow-up, such treatment could also be considered in an adjuvant setting in patients with neuroendocrine tumours who are operated on with curative intent. Lastly, as it has been shown in animal experiments that 90Y-labelled somatostatin analogues are more effective for larger tumours and 177Lu-labelled somatostatin analogues are more effective for smaller tumours [13, 14], combination therapy with 90Y-labelled and 177Lu-labelled octreotate may be tried in the near future. Such therapy might yield even better results than treatment with either of the radionuclides alone [15].

Glycoconj_J-4-1-2234449

Characterization of anticoagulant heparinoids by immunoprofiling

Heparinoids are used in the clinic as anticoagulants. A specific pentasaccharide in heparinoids activates antithrombin III, resulting in inactivation of factor Xa and–when additional saccharides are present–inactivation of factor IIa. Structural and functional analysis of the heterogeneous heparinoids generally requires advanced equipment, is time consuming, and needs (extensive) sample preparation. In this study, a novel and fast method for the characterization of heparinoids is introduced based on reactivity with nine unique anti-heparin antibodies. Eight heparinoids were biochemically analyzed by electrophoresis and their reactivity with domain-specific anti-heparin antibodies was established by ELISA. Each heparinoid displayed a distinct immunoprofile matching its structural characteristics. The immunoprofile could also be linked to biological characteristics, such as the anti-Xa/anti-IIa ratio, which was reflected by reactivity of the heparinoids with antibodies HS4C3 (indicative for 3-O-sulfates) and HS4E4 (indicative for domains allowing anti-factor IIa activity). In addition, the immunoprofile could be indicative for heparinoid-induced side-effects, such as heparin-induced thrombocytopenia, as illustrated by reactivity with antibody NS4F5, which defines a very high sulfated domain. In conclusion, immunoprofiling provides a novel, fast, and simple methodology for the characterization of heparinoids, and allows high-throughput screening of (new) heparinoids for defined structural and biological characteristics. Introduction Glycosaminoglycans (GAGs) are long, unbranched, negatively charged polysaccharide chains. The precursor of heparin and heparan sulfate (HS) is composed of alternating N-acetylglucosamine (GlcNAc) and d-glucuronic acid (GlcA) residues. In chondroitin sulfate (CS) and dermatan sulfate (DS), the aminosugar is a N-acetylgalactosamine (GalNAc) instead of GlcNAc. The polysaccharide chain precursors are modified by a sequential series of enzymatic reactions, including N-deacetylation/N-sulfation, C5-epimerization of GlcA to l-iduronic acid (IdoA), and O-sulfation at various positions [27, 32]. Heparin is widely used as an anticoagulant. A specific pentasaccharide sequence (GlcNAc/NS(6OS)–GlcA–GlcNS(3,6OS)–IdoA(2OS)–GlcNS(6OS)) [5, 23, 37] in heparin binds with high affinity to the enzyme inhibitor antithrombin III, causing a conformational change, resulting in the exposition of its active sites and in inactivation of factor Xa. For the inactivation of factor IIa, an additional site in heparin proximal to the pentasaccharide is necessary. The formation of a ternary complex between antithrombin III, factor IIa and heparin, which requires at least 18 saccharide residues for efficient formation, subsequently results in the inactivation of factor IIa [23]. Next to heparin, other heparinoids are used, including preparations containing other GAGs such as HS and DS, low molecular weight heparins (LMWHs), and fondaparinux, a synthetic pentasaccharide identical in structure to the antithrombin III-binding pentasaccharide. Heparinoids are used to treat patients with venous thrombosis and pulmonary embolism [2, 5, 10, 22]. In addition, they are used to prevent thrombosis in patients that undergo surgery and to prevent clotting in the extracorporeal circulation during hemodialysis [2, 5, 10, 22, 24, 28]. Due to their lower molecular weight, LMWHs have a high anti-factor Xa activity compared to their anti-factor IIa activity resulting in a more refined regulation of coagulation [4, 24]. In addition, the biological half-life of LMWHs/fondaparinux is longer in comparison to heparin, and the incidence of heparin-induced thrombocytopenia (HIT) is substantially reduced [6, 7, 23]. LMWHs are synthesized from heparin by controlled chemical or enzymatic depolymerization processes, resulting in the production of oligosaccharides of various lengths (Table 1). Dalteparin and nadroparin are produced by deamination using nitrous acid, resulting in the formation of an unnatural 2,5-anhydromannitol residue at the reducing end of the resulting oligosaccharides. Enoxaparin is produced by chemical β-elimination at alkaline pH, whereas tinzaparin is produced by enzymatic β-elimination by heparinase. Both processes result in the formation of an unsaturated uronic acid residue at the non-reducing end of the GAG fragments [12, 17]. Table 1Characteristics of therapeutic heparinoidsHeparinoid: generic name (brand name)Composition (and method of heparin modification)Molecular weight, kDaHeparinHeparin5.0–30.0Dalteparin (Fragmin)LMWH (deaminative cleavage with nitrous acid)4.0–6.0Nadroparin (Fraxiparine)LMWH (deaminative cleavage with nitrous acid)2.4–7.2Enoxaparin (Clexane, Lovenox)LMWH (β-eliminative cleavage by alkaline treatment)3.5–5.5Tinzaparin (Innohep, Logiparin)LMWH (β-eliminative cleavage by heparinase)3.4–5.6Sulodexide (Vessel Due F)80% ‘fast-moving’ heparina, 20% DS<8.0Danaparoid (Orgaran)84% HS, 12% DS, 4% CS4.0–10.0Fondaparinux (Arixtra)Pentasaccharide: C31H43N3Na10S81,728All drugs are derived from porcine intestinal mucosa, with the exception of the chemically synthesized pentasaccharide fondaparinux.LMWH Low-molecular weight heparin, HS heparan sulfate, DS dermatan sulfate, CS chondroitin sulfateaHeparin that is less sulfated and has a lower molecular weight [42]. All heparinoids, with the exception of fondaparinux, are heterogeneous mixtures of different GAG chains. Structural analysis of heparinoids is generally done by nuclear magnetic resonance spectroscopy, high pressure liquid chromatography, mass spectrometry, capillary electrophoresis, or polyacrylamide gel electrophoresis [17]. In general, these techniques require sophisticated instrumentation and sometimes extensive sample preparation. Alternative methodologies that are fast and simple are needed. Immunoprofiling of heparinoids could be a fast and simple method to characterize the complex heparinoids. Recently, a number of phage–display-derived antibodies were generated, reactive with various domain structures within heparin. In this study nine antibodies–all defining different heparin epitopes–were applied to study their reactivity with heparinoids. Materials and methods Heparinoids The following heparinoids were included in the study: heparin (Sigma, St. Louis, MO), dalteparin (Pharmacia, Woerden, The Netherlands), nadroparin (Sanofi-Synthélabo, Maassluis, The Netherlands), enoxaparin (Aventis Pharma, Hoevelaken, The Netherlands), tinzaparin (LEO Pharma, Breda, The Netherlands), sulodexide (Alfa Wassermann S.p.A., Bologna, Italy), danaparoid (Organon, Oss, The Netherlands), and fondaparinux (Sanofi-Synthélabo, Maassluis, The Netherlands). Characteristics are given in Table 1. Quantification of glycosaminoglycans by Farndale assay A Farndale assay was done to determine the concentration of GAGs in the preparations [13]. To 50 μl of the standard (heparin, 0–100 μg/ml) and heparinoids 1 ml Farndale reagent (40.5 mmol/l glycine, 40.6 mmol/l NaCl, and 0.05 mmol/l 1,9-dimethylmethylene blue, pH 3.0) was added and absorbance was measured directly at 525 nm. Determination of glycosaminoglycan classes by agarose gel electrophoresis The different classes of GAGs in the heparinoids were separated by agarose gel electrophoresis, fixed in 0.1% (w/v) azure A in 10 mmol/l MgCl2·6H2O, 50 mmol/l Na–formiate and 0.001% (w/v) Na–azide (pH 3.5), and stained by combined azure A–silver treatment [38]. Enzymatic digestion of heparinoids Heparinoids were diluted in heparinase I digestion buffer (50 mmol/l NaAc and 50 mmol/l Ca(Ac)2, pH 7.0) and/or chondroitinase ABC digestion buffer (25 mmol/l Tris and 2 mmol/l Mg(Ac)2, pH 8.0) to a final content of 4 mg/ml. The enzymes were added to a final concentration of 0.02 IU/ml for heparinase I (E.C. 4.2.2.7; from Flavobacterium heparinum; Seikagaku, Tokyo, Japan) and 0.04 IU/ml for chondroitinase ABC (E.C. 4.2.2.4; from Proteus vulgaris; Sigma). Heparinase I cleaves the linkages GlcNS±6S–IdoA2S predominantly present in heparin and to a minor extent in HS [11]. Chondroitinase ABC cleaves CS and DS [11]. The solutions were incubated for 16 h at 37°C. Determination of glycosaminoglycan size by polyacrylamide gel electrophoresis Heparinoids and their enzymatic digestion products were separated by polyacrylamide gel electrophoresis using a 33% gel [36], fixed in 0.8% (w/v) alcian blue in 2% (v/v) HAc, and stained by combined azure A–silver treatment [38]. Production of anti-heparin antibodies Antibodies (Table 2) were obtained by biopanning against HS or heparin, using bovine kidney HS (antibodies HS4C3 [41], HS3A8, HS4E4 [8], and LKIV69 [43]), porcine intestinal mucosa heparin (antibodies EW3D10 and EW4G2 [39]), human lung HS (antibody NS4F5; Smits et al., unpublished results), and mouse and human skeletal muscle GAG (antibodies AO4B08 and RB4EA12 [15], respectively). Large scale production of soluble antibodies was performed as described before [8, 41]. Table 2Characteristics of epitope-specific anti-heparin antibodiesAntibodyVH CDR3 sequenceVH familyDP geneGAG used for selectionEssential modificationsInhibitory modificationsHS4C3GRRLKD338Bovine kidney HS [41]N-sulfation 2-O-sulfation (minor importance)6-O-sulfation3-O-sulfationHS3A8GMRPRL338Bovine kidney HS [8]N-sulfation C5-epimerization2-O-sulfation6-O-sulfation (likely)HS4E4HAPLRNTRTNT338Bovine kidney HS [8]N-acetylation2-O-sulfationa 6-O-sulfationN-sulfationC5-epimerizationLKIV69GSRSSR338Bovine kidney HS [43]N-sulfation6-O-sulfationC5-epimerization (likely)2-O-sulfationEW3D10GRTVGRN338Porcine intestinal mucosa heparin [39]Sulfation required, position(s) unknownbEW4G2GKVKLPN338Porcine intestinal mucosa heparin [39]Sulfation required, position(s) unknownbNS4F5SGRKGRMR353Human lung HScN-sulfation C5-epimerization2-O-sulfation (high)6-O-sulfation (high)AO4B08SLRMNGWRAHQ347Mouse skeletal muscle GAG [15]N-sulfation C5-epimerization2-O-sulfationd6-O-sulfation (high)RB4EA12RRYALDY332Human skeletal muscle GAG [15]N-acetylation N-sulfation6-O-sulfationMPB49 (control)WRNDRQ338––Given are the antibody name, amino acid sequence of the VH complementary determining region 3 (CDR3), VH germ line gene family, DP gene number, selection moiety, and relevant modification reactions required for antibody binding. Table derived from Wijnhoven et al. unpublished results.GAG Glycosaminoglycan, HS heparan sulfateaAntibody HS4E4 possibly requires as yet unspecified O-sulfated residues [16].bAlthough the chemical nature of the specific heparin/HS structure recognized is not exactly known, the antibody defines a unique sulfated heparin/HS structure as demonstrated by its specific staining pattern on renal cryosections [8, 39].cSmits et al., unpublished resultsdAntibody AO4B08 requires an internal 2-O-sulfated iduronic acid residue [16]. Enzyme-linked immunosorbent assay Reactivity of the anti-heparin antibodies with the heparinoids was evaluated by enzyme-linked immunosorbent assay (ELISA). Wells of microtiter plates were coated overnight by incubation with the heparinoids that bound by passive adsorption [41]. After blocking with PBS containing 3% (w/v) BSA and 1% (v/v) Tween-20 for 1 h, anti-heparin antibodies [in PBS containing 1% (w/v) BSA and 0.1% (v/v) Tween-20] were added for 1.5 h. As a negative control, antibody MPB49 was used, which is >95% identical to most antibodies used, but does not bind any GAG. Bound antibodies were detected using mouse IgG anti-VSV tag antibody P5D4 (1:10; Boehringer Mannheim, Mannheim, Germany), followed by incubation with alkaline phosphatase-conjugated rabbit anti-mouse IgG (1:2,000; Dako, Glostrup, Denmark), both for 1 h. Enzyme activity was detected using 100 μl 1 mg p-nitrophenyl phosphate (ICN, Aurora, OH)/ml 1 M diethanolamine/0.5 mM MgCl2 (pH 9.8) as a substrate. Absorbance was measured at 405 nm [43]. Background reactivity, measured with the control antibody MPB49, was substracted. The absorbance with control antibody MPB49 was comparable to the absorbance measured when primary antibody was omitted. Immunofluorescence staining Human kidney cryosections (2 μm) were incubated with anti-heparin antibodies in the absence or presence of 100 μg/ml of the heparinoids. Bound antibodies were detected as described [43]. Results General biochemical analysis of heparinoids The different classes of GAGs in the heparinoids were separated by agarose gel electrophoresis (Fig. 1). Heparin showed a band at about the same position as the HS standard, whereas the LMWHs dalteparin, nadroparin, enoxaparin, and tinzaparin migrated somewhat further into the gel. Danaparoid displayed bands at the level of HS and DS, whereas the CS band, which only represents 4% of the prepration (Table 1), could not be detected. Sulodexide clearly showed two bands, representing the ‘fast-moving’ heparin (heparin that is less sulfated and has a lower molecular weight [42]) and DS fractions. The chemically synthesized pentasaccharide fondaparinux, which has a much lower molecular weight compared to the other heparinoids, migrated further into the gel. Fig. 1Agarose gel electrophoresis of heparinoids. The different classes of GAGs in the drugs were visualized by a combined azure A–silver staining procedure. The standard contains 40 ng HS, 20 ng DS, and 20 ng CS To analyze the molecular range of the heparinoids, gel electrophoresis using 33% polyacrylamide was applied. The LMWHs clearly displayed distinct fragments with a lower molecular weight compared to heparin (Fig. 2a). The smaller fragments of enoxaparin compared to tinzaparin (both prepared by β-elimination) can be explained by a preference of heparinase for 2-O-sulfated iduronic acid in contrast to chemical alkaline treatment, which results in cleavage at both sulfated and non-sulfated iduronic acid residues. Sulodexide and danaparoid showed a broad smear with less distinct fragments, in line with their methods of preparation, which does not include a specific depolymerization step. The pentasaccharide fondaparinux demonstrated primarily one individual band. To further establish the nature of the heparinoids, they were subjected to enzymatic digestion using heparinase I and/or chondroitinase ABC (Fig. 2b for dalteparin, enoxaparin, and sulodexide). Heparin, dalteparin, nadroparin, enoxaparin, tinzaparin, and fondaparinux were only cleaved by heparinase I, in accordance with their derivation from heparin. Sulodexide was cleaved by heparinase I as well as chondroitinase ABC, which is in line with the heparin/DS nature of the preparation. Danaparoid was primarily cleaved by chondroitinase ABC, reflecting the presence of DS/CS, whereas heparinase I was less effective. Fig. 2Thirty-three percent polyacrylamide gel electrophoresis of undigested (a) and enzymatically digested (b) heparinoids. Heparinoids were visualized by a combined alcian blue-silver staining procedure. Treatment with heparinase I resulted in cleavage of dalteparin and enoxaparin, whereas treatment with chondroitinase ABC did not. Sulodexide was cleaved by heparinase I as well as chondroitinase ABC Characterization of heparinoids using anti-heparin antibodies Reactivity of anti-heparin antibodies with immobilized heparinoids was analyzed by ELISA (Table 3). For reference, reactivity of antibodies with heparin was set at 100%. A distinct antibody-binding profile could be observed for all antibodies, although also similar reactivity with a number of antibodies was noticed. For instance, enoxaparin and tinzaparin, both prepared by β-elimination, reacted differently with antibodies LKIV69 and EW4G2, whereas they reacted similarly with the other antibodies. Nadroparin reacted very poorly with all the antibodies tested, which may be in line with its overall low amount of sulfated disaccharides in comparison with other LMWHs [17]. In contrast, sulodexide and danaparoid reacted strongly with antibody HS4E4, which defines a low-sulfated and N-acetylated epitope, in comparison to the LMWHs. This is in line with the presence of low-sulfated heparin (HS) and HS in sulodexide and danaparoid, respectively, and which contain a substantial amount of N-acetylated glucosamine. Compared to heparin, the other heparinoids reacted poorly with antibody NS4F5, which defines a very highly sulfated oligosaccharide. Table 3Reactivity of anti-heparin antibodies with heparinoids (ELISA) HS4C3HS3A8HS4E4LKIV69EW3D10EW4G2NS4F5AO4B08RB4EA12Heparin100100100100100100100100100Dalteparin75 ± 644 ± 48 ± 53 ± 133 ± 649 ± 80 ± 016 ± 413 ± 2Nadroparin29 ± 35 ± 24 ± 30 ± 03 ± 17 ± 40 ± 01 ± 11 ± 1Enoxaparin80 ± 956 ± 87 ± 310 ± 236 ± 826 ± 71 ± 128 ± 721 ± 6Tinzaparin84 ± 155 ± 28 ± 344 ± 1045 ± 1274 ± 19 ± 220 ± 325 ± 8Sulodexide102 ± 571 ± 11283 ± 2151 ± 743 ± 10100 ± 30 ± 0102 ± 1526 ± 8Danaparoid69 ± 1122 ± 358 ± 172 ± 22 ± 11 ± 10 ± 063 ± 63 ± 1Anti-heparin antibodies were applied to heparinoids immobilized on microtiter plates. Given is the reactivity of the antibodies with the heparinoids relative to unmodified heparin. Values are given as mean ± SEM (n = 4). Each hepainoid shows a distinct immunoprofile. Heparinoids could be ranked according to their reactivity with antibodies. For instance, reactivity of antibody EW4G2 was highest with heparin and sulodexide, followed by tinzaparin, dalteparin, and enoxaparin. Reactivity with the other heparinoids was very low or absent. Antibody LKIV69 reacted strongly with heparin, followed by sulodexide and tinzaparin, whereas other heparinoids reacted poorly. Reactivity of antibody AO4B08 was highest with heparin and sulodexide, followed by danaparoid and enoxaparin/tinzaparin/dalteparin, whereas reactivity with nadroparin was absent. These data indicate that distinct heparin domain structures are present in different commercially available heparinoids. The only chemically defined heparinoid studied was the pentasaccharide fondaparinux (GlcNS(6OS)–GlcA–GlcNS(3,6OS)–IdoA(2OS)–GlcNS(6OS)). Due to its small size (1,728 kDa), it was assayed using a competition ELISA (Table 4). Heparin was included as a reference and values obtained for fondaparinux were related to those obtained for heparin. Antibodies HS4E4 and RB4EA12, both requiring N-acetylation in their epitopes, did not react with fondaparinux. Antibody NS4F5, which reacts with an extremely high sulfated oligosaccharide, was also not reactive. These results are in line with the chemical composition of fondaparinux, which contains no N-acetylated glucosamine and a non-sulfated glucuronic acid residue. Reactivity with antibody LKIV69, which reacts poorly with 6-O-sulfated epitopes, was very weak and reflects the high amount of 6-O-sulfation in fondaparinux. Reactivity with antibody AO4B08, which needs C5-epimerization, N-sulfation, high 6-O-sulfation and an internal 2-O-sulfated iduronic acid residue, was only moderate/poor, reflecting the lack of an internal 2-O-sulfated iduronic acid (i.e. flanked by other iduronic acid residues). Reactivity with antibody HS4C3, which requires N-, 3-O-, and 6-O-sulfation, and antibody HS3A8, which requires at least C5-epimerization and N- and 2-O–sulfation, was high, which corresponds with the chemical structure of fondaparinux. Reactivity was also high with antibodies EW3D10 and EW4G2, both defining an as yet undefined, but sulfated oligosaccharide. These data indicate that reactivity with antibodies may provide some chemical information of heparinoids. Table 4Reactivity of anti-heparin antibodies with heparin and fondaparinux (competition ELISA) HS4C3HS3A8HS4E4LKIV69EW3D10EW4G2NS4F5AO4B08RB4EA12Heparin0.3 ± 0.10.4 ± 0.12.7 ± 0.1 0.2 ± 0.00.1 ± 0.00.1 ± 0.00.1 ± 0.01.2 ± 0.2<0.01Fondaparinux1.7 ± 0.21.3 ± 0.1 >20010.3 ± 0.80.2 ± 0.01.5 ± 0.5 11.7 ± 0.8 12.3 ± 3.821.5 ± 0.5Ratio fondaparinux/heparin5.73.3>7451.521511710.3>2150Given are the amounts of heparinoids (micrograms per milliliter), which result in a 50% inhibition of the binding of the anti-heparin antibodies to immobilized HS/heparin (IC50 values), and the ratio between antibody reactivity with fondaparinux and heparin. Values are given as mean ± SEM (n = 3). Antibody EW3D10 showed a strong reactivity with fondaparinux, followed by antibodies HS4C3, HS3A8, and EW4G2. The other antibodies showed a relatively weak or no reactivity. Reactivity of the anti-heparin antibodies with immobilized heparinoids was supported by immunofluorescence analysis (Fig. 3). When human renal cryosections were simultaneously incubated with anti-heparin antibodies and heparinoids, staining was completely abolished with those heparinoids, which reacted strongly with the antibodies in ELISA (e.g. heparin and sulodexide for antibodies HS4C3 and HS4E4). An obvious decreased staining was noticed for those heparinoids, which reacted moderately with the antibodies (e.g. nadroparin for antibody HS4C3), whereas no major differences in staining were observed for those heparinoids, which did not bind to antibodies (e.g. nadroparin for antibody HS4E4). Fig. 3Immunofluorescence staining of human renal cryosections with anti-heparin antibodies HS4C3 and HS4E4 in the absence and presence of heparinoids. Staining was abolished when renal cryosections were incubated with HS4C3 or HS4E4 in the presence of heparin or sulodexide. HS4C3 staining was decreased in the presence of nadroparin, whereas HS4E4 staining was unaffected. Bar represents 50 μm; magnification is identical for each photograph. G Glomerulus, BC Bowman’s capsule, T renal tubule Discussion In this study eight heparinoids were characterized by their reactivity with nine different anti-heparin antibodies. All heparinoids showed a distinct antibody-binding profile in line with differences in GAG composition (heparin, HS, CS, DS), molecular weight, and method of preparation. A number of antibodies reacted strongly with certain heparinoids, but not with others. It may therefore be possible to deduce chemical and biological information of a specific heparinoid by virtue of its antibody profile. For instance, antibody NS4F5 showed reactivity with only heparin. This antibody defines a stretch of trisulfated (N-, 2-O-, and 6-O-sulfated) and epimerized disaccharides and therefore defines a highly sulfated heparin domain (Smits et al., unpublished results). The lack of reactivity of the other heparinoids for NS4F5 indicates the absence or the destruction of the domain during their preparation. Absence may hold for sulodexide and danaparoid, which contain HS rather than heparin. Destruction by the depolymerization processes may hold for the LMWHs. In this respect, reactivity with antibody NS4F5 may be indicative for the degree of depolymerization. The antibody profile of the chemically synthesized pentasaccharide fondaparinux is compatible with its chemical structure GlcNS(6OS)–GlcA–GlcNS(3,6OS)–IdoA(2OS)–GlcNS(6OS). Reactivity with antibodies HS4C3 and HS3A8 indicates it to contain iduronic acid and N-, 2-O-, 3-O-, and 6-O-sulfate residues, whereas poor/no reactivity with antibodies HS4E4, LKIV69, AO4B08, and RB4EA12 indicates the lack of internal 2-O-sulfated iduronic acid residues (i.e. flanked by other iduronic acids) and N-acetylated glucosamine residues. Lack of reactivity with antibody NS4F5 indicates that fondaparinux does not contain a stretch of tri-sulfated disaccharides. Next to information on the chemical properties of the heparinoids, the antibody profile may be indicative for their biological activity, e.g. the anti-Xa/anti-IIa ratio. This ratio is of essential importance since it is associated with a more predictable anticoagulant effect. The anti-Xa/anti-IIa ratio is 1 for unfractionated heparin, 2–4 for LMWHs, 0.5 for sulodexide, and infinite for fondaparinux [2, 23]. Antibodies such as HS4E4 and AO4B08 (Tables 3 and 4), which are highly reactive with heparin and sulodexide, but poorly/not with LMWHs and fondaparinux, likely bind to a heparin domain structure at a site proximal to the pentasaccharide sequence and may be indicative for a high anti-factor IIa activity. These results indicate that when a heparinoid is recognized by antibodies HS4C3, HS3A8, EW3D10, and EW4G2 (which showed a high reactivity with fondaparinux; Table 4), as well as by antibodies HS4E4 and AO4B08, the heparinoid is able to form a ternary complex with antithrombin III and factor IIa, resulting in inactivation of both factors Xa and IIa [23]. On the other hand, when a heparinoid is recognized by antibodies HS4C3, HS3A8, EW3D10, and EW4G2, but not by antibodies HS4E4 and AO4B08, it is does not contain the heparin domain structure at the site proximal to the pentasaccharide required for factor IIa binding. In this case only factor Xa will be inactivated. When a ratio is calculated from reactivity of heparinoids with antibodies HS4C3 and HS4E4 (indicative for 3-O-sulfates and for the presence of domains allowing IIa activity, respectively), and heparin is assigned a ratio of 1, the following numbers are found: dalteparin, 9; nadroparin, 7; enoxaparin, 11; tinzaparin, 11; sulodexide, 0.4; and fondaparinux, >>13 (no antibody HS4E4 reactivity observed). These numbers are in line with the above mentioned anti-Xa/anti-IIa ratios, which were determined by enzymatic assays [2, 23], and indicate that biological characteristics are associated with antibody profiles. Heparin-induced thrombocytopenia (HIT) is a major complication in patients administered heparin. Heparin binding to platelet factor 4 (PF4) causes a conformational change in the protein, rendering it antigenic. Susceptible patients develop antibodies to the heparin-PF4 complex, and binding of the antibody-PF4-heparin complex to its receptor on platelets results in platelet activation and aggregation. In addition, platelets are removed by activated immune cells, resulting in thrombocytopenia [6]. Highly sulfated domains within heparin/HS are involved in the binding to PF4 [31], and an increase in sulfation and size is associated with an increased risk of HIT [6, 19]. This may explain why the incidence of HIT is lower in patients treated with LMWHs, danaparoid, or fondaparinux compared to unfractionated heparin [7, 23]. Antibody NS4F5, which defines a highly sulfated domain, reacted strongly with heparin, but not with the other heparinoids. Reactivity with this antibody may thus be used to obtain information about the risk of a heparinoid to induce HIT. Although we have focussed here on the immunoprofiling of heparinoids in the context of anticoagulation, this technique may also be useful to evaluate heparinoids in other settings. Heparin and LMWHs have been used to treat patients with kidney disease since an ameliorating effect on a.o. proteinuria has been found [1, 14, 21, 25, 30, 33, 34, 40]. Likewise, heparinoids are reported to have anti-tumor and anti-metastatic effects [3, 20, 29], the degree of sulfation being of crucial importance. Heparinase III-digested heparin (highly sulfated fragments), inhibited tumor growth, whereas heparinase I-digested heparin (low sulfated fragments) promoted tumor growth [18], underscoring the necessity to characterize the preparation used. In addition to the implementation of these antibodies to characterize the therapeutic potential of heparinoids, they may be used to stage a disease (e.g. kidney disease, cancer) characterized by a specific HS profile in tissue, blood, and/or urine. Since the antibody profile of the heparinoids used in these studies may be correlated to a specific biological effect, antibodies may be used for the identification and subsequent characterization of the active structures within the heparinoids, which may result in the development of more specific drugs. The antibodies described here may also be used to counteract bleeding in patients administered an overdose of heparinoids. This is generally accomplished by treatment with protamine sulfate [9, 26], but this rather unspecific polycation may cause hypotension, bradycardia, and thrombocytopenia. In addition it may cause other severe side effects including anaphylactic and anaphylactoid reactions, resulting in respiratory distress, circulatory collapse, capillary leak, and pulmonary hypertension [26]. There is a need for safer agents that inhibit the anticoagulant activity of heparinoids, and some antibodies described here may be useful, especially those that also react with fondaparinux, which is difficult to inhibit in vivo. It has already been described that antibodies HS4C3 and EW3D10 strongly inhibited heparin-induced anticoagulation (APTT clotting assay) [35, 39]. Antibody HS4C3 also blocked the anticoagulant activities of heparin and fondaparinux in an anti-factor Xa assay [35]. These studies indicate that anti-heparin antibodies may react with heparinoids thereby neutralizing their anticoagulant effect. Finally, our antibodies may be used for fast industrial screening of heparinoids. A quality check of heparinoid batches may be performed using the antibodies since each heparinoid shows a specific immunoprofile. In conclusion, this study presents a novel way of characterizing heparinoids using immunoprofiling. Reactivity with antibodies was found indicative for chemical and biological aspects of the heparinoids, and may therefore be used for a fast and simple screening of industrial batches for defined characteristics.

Ann_Hematol-4-1-2413090

Cytogenetic features in myelodysplastic syndromes

Myelodysplastic syndromes (MDS) comprise a group of bone marrow diseases characterized by profound heterogeneity in morphologic presentation, clinical course, and cytogenetic features. Roughly 50% of patients display clonal chromosome abnormalities. In several multicentric studies, the karyotype turned out to be one of the most important prognostic parameters and was incorporated into statistical models aiming for a better prediction of the individual prognosis like the International Prognostic Scoring System. However, due to the profound cytogenetic heterogeneity, the impact of many rare abnormalities as well as combinations of anomalies occurring in a substantial portion of patients with MDS is still unknown and can only be delineated on the basis of large international multicentric cooperations. Recently, the German–Austrian MDS Study Group presented cytogenetic findings in 2,072 patients with MDS, which serve as a basis for the characterization of the cytogenetic subgroups discussed in this article. The availability of new therapeutic options for low- and high-risk MDS targeted against distinct entities characterized by specific chromosome abnormalities, like 5q-deletions, monosomy 7, and complex abnormalities underlines the important role of cytogenetics for the clinical management of MDS. This article thus focuses on the clinical and prognostic relevance, the molecular background, and therapeutic perspectives in these three cytogenetic subgroups. Introduction The profound heterogenetity of myelodysplastic syndromes (MDS) is well-known from morphological and clinical studies, which finally lead to the establishment of classification, and prognostic scoring systems. It is becoming increasingly obvious that this heterogeneity also manifests itself against the background of genetic heterogeneity of MDS. In sharp contrast to the situation in chronic myeloid leukemia (CML) where one single cytogenetic abnormality, the translocation t(9;22)(q34;q11), is the genetic hallmark of the disease, there is an enormous variability of cytogenetic abnormalities in MDS hampering not only the prognostic classification but also the delineation of the molecular background of cytogenetic aberrations in MDS. As an example in our recent multicentric cytogenetic analysis of patients with MDS, we observed 684 different types of chromosome abnormalities in a cohort of 1,080 patients with MDS and an abnormal karyotype [1]. The cytogenetic profile of MDS In general, MDS show a characteristic genetic profile with an overweighing of unbalanced abnormalities. Most frequently, a loss of genetic material in the form of deletions and monosomies can be observed. A gain of genetic material with the appearance of total or partial trisomies is less frequent. Loss or gain of genetic material can also be the result of unbalanced translocations, which are frequently observed in MDS with multiple abnormalities. Taken together, it is thus obvious to assume that a prime molecular mechanism in MDS is the loss or inactivation of tumor suppressor genes, while the activation of oncogenes seems to be less relevant in myelodysplasia. In contrast to AML, balanced structural abnormalities like translocations and inversions are rare in MDS. Due to the profound genetic heterogeneity, the knowledge about distinct cytogenetic alterations was mainly restricted to the most frequent abnormalities (−5/5q−, −7/7q−, +8, 20q−, and −Y), although, in MDS, rare chromosome abnormalities are present in a substantial portion of patients. In a recent analysis of the German–Austrian MDS Study Group, 59% of all 2,370 abnormalities observed in 1,080 patients with MDS where rare; that is, they occurred with a frequency of less than 2% [1]. In this situation, prognostic knowledge can only be deepened by large-scale multicentric studies with a high enough number of abnormal cases with follow-up data. This development was initiated by the International MDS Risk Assessment Working Group (IMRAW) ending up with the establishment of the International Prognostic Scoring System (IPSS), which was based upon the analysis of 816 patients with de novo MDS of whom 327 had abnormal karyotypes [2]. A next step forward was the study of the Spanish cooperative group with 500 abnormal cases [3] followed by our German–Austrian dataset with 1,080 patients with chromosome abnormalities [1]. The situation is furthermore complicated by the fact that chromosome abnormalities in principle can occur in three different conditions: as isolated abnormality, together with one additional change, and as part of complex abnormalities with at least two additional cytogenetic alterations. Table 1 shows the incidence of the 21 most frequent abnormalities according to the number of accompanying alterations occurring in our German–Austrian patient cohort. Table 1Incidence of chromosome abnormalities in MDS related to 2,072 patients examined successfullyAnomalytotal, n (% of all cases)Isolated, n (%a)With one additional abnormality, n (%a)As part of complex abnormalities, n (%a)5q−312 (15.1)146 (47)52 (17)114 (36)−7/7q−230 (11.1)86 (37.5)31 (13.5)113 (49)+8173 (8.4)81 (46.8)37 (21.4)55 (31.8)−18/18q−78 (3.8)3 (3.8)2 (2.6)73 (93.6)20q−74 (3.6)36 (48.6)10 (13.5)28 (37.8)−569 (3.3)1 (1.4)4 (5.8)64 (92.8)−Y58 (2.8)41 (70.7)5 (8.6)12 (20.7)+2145 (2.2)5 (11.1)18 (40)22 (48.9)−17/17p−42 (2.0)1 (2.4)1 (2.4)40 (95.2)inv/t(3q)41 (2.0)16 (39)8 (19.5)17 (41.5)−13/13q−40 (1.9)5 (12.5)6 (15)29 (72.5)+1/+1q37 (1.8)3 (8.1)6 (16.2)28 (75.7)−2133 (1.6)3 (9.1)4 (12.1)26 (78.8)+1128 (1.4)6 (21.4)4 (14.3)18 (64.3)−1226 (1.3)02 (7.7)24 (92.3)12p−25 (1.2)7 (28)6 (24)12 (48)t(5q)24 (1.2)6 (25)3 (12.5)15 (62.5)11q−23 (1.1)8 (34.8)4 (17.4)11 (47.8)9q−23 (1.1)8 (34.8)3 (13)12 (52.2)t(7q)22 (1.1)6 (27.3)6 (27.3)10 (45.5)−2022 (1.1)0022 (100)aOf cases with the respective abnormality Cytogenetic prognosis The first large-scale cytogenetic studies in patients with MDS encompassing more than 100 patients where published more than 20 years ago [4–7]. In the following years, the patients cohorts increased step by step from nearly 250 patients [8] to some 400 patients [9, 10]. In 1997, Greenberg published a collaborative multicentric international data set of more than 800 patients with de novo MDS, which was the basis for the establishment of the IPSS [2]. This database was further surmounted by the analyses of the Spanish cooperative group with 968 patients examined (Table 2), which represented the largest number of patients with MDS karyotyped so far [3]. Recently, the German–Austrian Study Group published their multicentric analyses of more than 2,100 patients with MDS [1]. Table 2Cytogenetic prognostic findings in publications with greater than 100 patients examinedAuthor, yearNumber of patientsAbnormal (%)FavorableIntermediateUnfavorableKnapp 198517466 (38)Normal (NN)ComplexNowell 1986–198914463 (44)5q−, 20q−+8−7/7q−Billström 198816974 (44)NN, 5q−+8−7/7q−Pierre 1989247106 (43)NNComplexMorel 1993408151 (37)NN, 5q−, −Y, −7/7q−, 20q−+8ComplexToyama 1993401200 (50)+8−7/7q−ComplexWhite 199419875 (38)NN, 5q−12p−, +21Complex, +8, 20q−Greenberg 1997816327 (40)NN, 5q−, 20q−, −YAll othersComplex, abnormal #7Solé 2005968500 (51)NN, 5q−, 20q−, −Y, 11q−, 12p−rea 3q, + 8, +9, t11q, 17p−Complex, −7/7q−, i17qHaase 20072,0721,080 (51)NN, +1/+1q, t(1q), 5q−, t(7q), 9q−, 12p−, abnormal #15, t(17q), 20q−, −21, + 21, −X, −Yrea 3q, −7, 7q−, +8, 11q−, t(11q23), +19, complex (=3)Complex (>3), t(5q)rea Rearranged What has changed and what has been achieved during these last 25 years of cytogenetic analyses in MDS? If the aberration rate is considered, there is a gradual increase in the portion of abnormal cases from under 40% in the most earlier studies to 50% in the more recent analyses of Toyama, Solé, and Haase [1, 3, 10]. Although the composition of the patient cohorts may play a role for the portion of clonal abnormalities in a given collective, other factors like the improvement of culture conditions possibly due to the use of recombinant myeloid growth factors [11] might have contributed to the increase in the aberration rate in cytogenetic studies. To date, an aberration rate of 50% can be regarded as an international standard. Prognostic relevance of chromosome abnormalities The prognostication of patients with MDS has become an important means for the development of therapeutic strategies based on more individualized risk assessment. Even in the most recent prognostic scoring system, the World Health Organization classification-based Prognostic Scoring System, cytogenetics play a decisive role [12]. In general, three to four prognostic cytogenetic categories can be distinguished. In several multicentric investigations, cytogenetics have been proven to be highly relevant independent prognostic parameters ([2, 3, 13], see also Table 2). Good prognosis Even in the earliest study considered here, Knapp et al. [4] described the prognostic relevance of cytogenetic findings with a normal karyotype as favorable and complex abnormalities as being associated with bad prognosis. The good prognosis of a normal karyotype was confirmed by nearly all other groups. Thus, in contrary to the situation in AML where patients with a normal karyotype have an intermediate outcome and prognosis is significantly influenced by additional molecular alterations, in MDS, a normal karyotype is undisputedly associated with good prognosis. Further, well-established abnormalities defining good-risk subgroups are deletions of 5q and 20q as well as loss of the Y-chromosome, although the question whether this latter abnormality is age related, or a real clonal marker is unsolved as yet. The Spanish group identified further new cytogenetic abnormalities with a good prognosis (12p− and 11q−) [3]. In the German–Austrian multicentric study, the following abnormalities were associated with a favorable clinical course with a median survival between more than 9 years and 32 months: normal karyotype, t(1q), 5q−, t(7q), 9p−, 12q−, t(15q), t(17q), 20q−, +21, −21, −X, −Y. However, prognosis was only favorable when not more than one additional abnormality was present [1]. Intermediate prognosis In most studies, patients with trisomy 8 displayed an intermediate clinical course. Until the Spanish group published their results on more than 980 patients, the knowledge of cytogenetic findings with an intermediate prognosis was very limited. In the IPSS, all abnormalities neither belonging to the good-risk group (isolated 5q−, 20q−, and loss of Y-chromosome) nor to the bad risk cohort (complex [greater than or equal to three abnormalities] or any chromosome 7 abnormality) were designated to be of intermediate prognosis not by availability of survival data but by definition [2]. In their large Spanish dataset, Solé et al. delineated new abnormalities associated with an intermediate clinical course (3q abnormalities, trisomy 9, 11q translocations, and 17p deletions) [3]. The results of White et al. [14] and Toyama et al. [10] who identified −7/7q− and 12p− and trisomy 21, respectively, as abnormalities with an intermediate prognosis were not confirmed by other groups. Recently, the multicentric database of the German–Austrian MDS Study Group revealed several infrequent abnormalities with an intermediate prognosis. A median survival time of 23–26 months was observed in patients with trisomy 8 and 11q− (called Intermediate-I). A more worse prognosis with a median survival time between 20 and 14 months was found for 11q23 translocations, chromosome 3q aberrations, trisomy 19, 7q deletions, and complex abnormalities with three different chromosomal alterations and monosomy 7 (called Intermediate-II) [15]. Poor prognosis There is a great consensus in all publications on cytogenetic prognosis in MDS that complex abnormalities characterize a MDS subgroup with bad prognosis and a median survival time significantly below 1 year, although the threshold at which the number of abnormalities confers bad prognosis is a matter of debate. While in most publications, the term “complex” is used for three or more abnormalities, in the Medical Research Council AML trials, bad prognosis was assigned to cases with five or more abnormalities [16]. Recently, we clearly could show that, in patients with MDS, median survival is significantly reduced only when more than three abnormalities are present, reducing the median survival from 17 months in cases with three abnormalities to less than 9 months in cases with four or more anomalies [1]. Besides complex changes, partial and total monosomy 7 was ranked to the cytogenetic findings with an unfavorable clinical course [3, 5–7]. Within the IPSS, any chromosome 7 abnormality was counted to the bad prognostic findings [2]. As mentioned above, in the German–Austrian dataset, −7/7q− showed a significant better prognosis (14 and 19 months median survival time, respectively) as compared to the complex abnormality group defined by four or more chromosomal changes (8.7 months) and was thus attributed to the intermediate-II group [15]. The findings of Toyama et al. [10] who observed an unfavorable course in patients with trisomy 8 or 20q deletions was not confirmed by any other group. Solé et al. described a median survival of under 12 months in ten patients with an isochromosome of 17q [3]. Since, in the German–Austrian dataset, this abnormality was too infrequent for a statistically meaningful analysis, the Spanish findings are convincing but still need further confirmation. Prognostic scoring In the IPSS, three different cytogenetic subgroups were established and weighted against bone marrow blast counts and cytopenias by multivariate analysis [2]. Applying the cytogenetic part of the IPSS to our German–Austrian dataset, 59% of the patients had a good risk karyotype (normal, isolated 5q−, 20q−, or loss of Y-chromosome) with a median survival of 54 months, 19% of the collective had an intermediate prognosis (neither good risk nor bad risk cytogenetics) with a median survival of 31 months, and 22% displayed a dismal prognosis (any chromosome 7 anomaly, complex [≥3] changes) with a median survival of 11 months [1]. The application of the new cytogenetic findings derived from the German–Austrian dataset allows the definition of four different cytogenetic prognostic subgroups (Table 3 and Fig. 1). The low-risk group covers 73% of patients with 14 cytogenetic categories and a median survival of 55 months, the intermediate groups I and II (15.5% of patients) with eight cytogenetic categories and a median survival of 29 (intermediate-I) and 15 months (intermediate-II), and the high-risk group (11.5% of patients) with a median survival of 8 months [15]. In comparison to the IPSS, our new cytogenetic prognostication splits the intermediate group into two distinctly separate subgroups with significantly different median survival. The bad-risk group is more strictly defined with a lower median survival time (8 vs. 11 months). While the IPSS assigned an intermediate risk by exclusion and not by knowledge, our intermediate cytogenetic subgroup was based exclusively on available survival data of patients treated with supportive care only thus reflecting the natural course of the disease. Taking into account only those patients with a known prognosis (in IPSS only good- and bad-risk patients), 81% of our patients could be categorized by the IPSS in comparison to 93% by the German–Austrian prognostic system. Fig. 1Kaplan–Meier survival curves according to the cytogenetic prognostic classification of the German–Austrian MDS Study Group. Log-rank test: p < 0.0001 (3 degrees of freedom) [15]Table 3New cytogenetic prognostic subgroups of the German–Austrian MDS Study Group in 1,202 patients treated with supportive care onlyCytogenetic riskCytogenetic findingNumber (%)Median survival (months)Good12p−7 (0.6)n.r.9q−6 (0.5)n.r.t(15q)6 (0.5)n.r.15q−5 (0.4)n.r.+2113 (1.1)100.85q−132 (11)77.220q−24 (2)71.0−X6 (0.5)56.4normal karyotype622 (51.7)53.4−Y33 (2.8)39.4t(1q)7 (0.6)34.7t(7q)7 (0.6)34.7t(17q)6 (0.5)32.1−216 (0.5)32.0Intermediate-I11q−11 (0.9)26.1+864 (5.3)23.0Intermediate-IIt(11q23)6 (0.5)20.0Any 3q abnormality16 (1.3)19.9+195 (0.4)19.87q−11 (0.9)19.0Complex (=3 anomalies)32 (2.7)17.0−742 (3.5)14.0PoorComplex (>3 anomalies)134 (11.1)8.7t(5q)7 (0.6)4.4n.r. Median survival not reached A further element of the IPSS, the weighting of cytogenetics in comparison to other relevant parameters like the bone marrow blast counts, has become questionable. According to the IPSS, 0 scoring points are attributed to good-risk cytogenetics and blasts less than 5%, intermediate cytogenetics as well as 5–10% blasts get 0.5 scoring points, and bad-risk cytogenetic findings are scored with 1.0. Eleven to 20% blasts are scored with 1.5 and 21–30% blasts with 2.0 points. Thus, bad-risk cytogenetics with a median survival of 11 months get less scoring points than patients with 11–20% blasts and a median survival in our cohort of 16.5 months. According to our observations, only patients with 21–30% blasts (11.7 months median survival) had a prognosis comparable to patients with bad-risk cytogenetics. Thus in a revision of the IPSS the weight of bad-risk cytogenetics has to be readjusted [17]. Delineation of the most relevant cytogenetic subgroups 5q−, monosomy 7, and complex abnormalities 5q deletions Deletions within the long arm of chromosome 5 are the most frequent cytogenetic changes in MDS accounting for roughly 30% of abnormal cases [1, 2, 3]. The deletions can have variable size; however, the common deleted region always spans the chromosome band 5q31. Due to intensive research during the last decade, our knowledge of the molecular background of these abnormalities is increasing. It is generally accepted that not the illegitimate fusion of deoxyribonucleic acid (DNA) sequences at the variable breakpoints of the deletions but loss of genetic information represents the relevant pathomechanism. There could be two different deleted regions in 5q31. One that is located more centromeric is possibly associated with bad prognosis, complex abnormalities, and high-risk, as well as therapy-related, MDS. A second area is located more telomeric in the vicinity of band 5q32 and is supposed to be related to the good-risk 5q− syndrome [18–21. Recently, it has been convincingly shown by the means of sophisticated systematic knockout experiments on 41 candidate genes in the critical 5q− region that the gene for a ribosomal subunit protein, RPS41, seems to be a relevant gene in patients with 5q− syndrome on the basis of haploinsufficiency [22]. The 5q− syndrome has been first described by Van den Berghe in 1974 [23]. It is cytogenetically characterized by an isolated deletion of chromosome 5q. A female preponderance is well documented. The clinical appearance is characterized by a refractory macrocytic anemia, normal or elevated platelets, and mild leukocytopenia. The clinical course is mild and long lasting with a very low risk for leukemic transformation. The bone marrow smears reveal less than 5% blasts, while dysplasia of the erythroid and granulocytic lineages usually are discrete or even absent. Megakaryopoesis shows profound and characteristic dysplasias with either separated multiple nuclei, hypolobulation, microkaryocytes, and most characteristically monolobulated megakaryocytes with a round nucleus. The prognosis of 5q deletions in MDS is generally favorable if they are not part of complex abnormalities; however, it is significantly modified by single additional cytogenetic changes [1, 24]. Figure 2 shows the Kaplan–Meier survival curves for patients treated with supportive care only with isolated 5q−, 5q− plus one additional abnormality, and 5q− as part of complex abnormalities derived from the German–Austrian database [1]. Fig. 2Median survival according to accompanying abnormalities in patients with 5q deletions. Log-rank test: p ≤ 0.0001 (2 degrees of freedom), p = 0.30 (isolated vs. +1), p ≤ 0.0001 (isolated vs. complex), p = 0.0001 (+1 vs. complex) It is evident that additional abnormalities negatively influence survival in patients with 5q deletions. However, due to cytogenetic heterogeneity, the additive prognostic impact of distinct single additional abnormalities remains obscure as yet (Table 4). Table 4Frequencies (in percent of all cases with the respective primary abnormality) of accompanying abnormalities5q− (n = 59)−7/7q− (n = 38)trisomy 8 (n = 44)Additional anomalyPercentAdditional anomalyPercentAdditional anomalyPercent+817+21105q−23+21135q−10+2111−20/20q−8+810der(3q21/q26)7−77inv(3q)5+117der/del(12p)7del(12p)5−75der/del(3p)5t(11q23)5del(12p)5der/t(21q)5iso(17)(q10)5+135t(11q23)3Others50+145Others35+8*5del(1p)5Others22 Over a long period, the therapeutic standard in patients with noncomplex 5q deletions was supportive care only. In clinical trials, retinoic acid turned out to be inefficient, and low-dose cytarabine was more effective but led to a pronounced increase in severe neutropenic infections [25]. A preferential response to other modern therapeutic strategies like immunosuppression and suppression of DNA methyltransferase or histone deacetylase was not reported as yet. However, recently, List et al. reported a remarkably high response rate to the immunomodulating agent lenalidomide, which was especially pronounced in MDS patients with 5q deletions. Nine out of 12 patients with a 5q deletion displayed complete cytogenetic remissions, and 10 out of 12 experienced an erythroid response [26]. Thus, a new promising agent may target cell clones in MDS bearing 5q deletions. These results could recently be confirmed on the basis of a large muliticentric trial [27]. Monosomy 7 Monosomy 7 is the second most frequent distinct chromosome abnormality in MDS occurring in some 25% of abnormal cases. It can present as total or partial monosomy. In the latter case, variable deletions of parts of the long arm lead to loss of genetic material of different size. As yet, no significant differences concerning the prognostic relevance have been observed between total and partial monosomy 7. In the German–Austrian dataset, 36% of monosomy 7 were isolated ones, 14% displayed one additional abnormality, and 50% occurred as part of complex abnormalities [1]. Comparable to the situation in 5q, also in 7q, at least two and maybe more distinct regions of common deletions have been identified: the band 7q22 and the more telomeric regions 7q31–32 and 7q36. It is interesting to note that interstitial as well as terminal deletions might be due to cryptic unbalanced translocations. In single individuals, more than one different deletion has been found in one and the same patient, either within the same copy of chromosome 7 with retention of sequences between the deleted regions or within different cell clones [28, 29]. These observations clearly underline the regional genetic instability of the long arm of chromosome 7, which makes this region especially prone to deletions of different size and localization as well as to structural abnormalities. On the other hand, it can be assumed that there must be a cluster of genes with tumor-suppressive features distributed over several chromosome bands in 7q with the same or very similar consequences when being inactivated by chromosomal abnormalities. However, in contrast to some presumptions in the 5q deletions, as yet, no prognostic differences have been elaborated between different regions of deletions (reviewed in [20]). Investigations of the molecular background of monosomy 7 MDS are on the way. An association of this cytogentic subgroup with RAS mutations, mutations of AML1, and hypermethylation of p15INK4B have been reported [30, 31]. In an analysis of gene expression profiles in CD34+ cells from MDS patients with monosomy 7, a malignant phenotype with highly proliferative potential was found with an overexpression of HOX9A, PRAME, BMI-1, PLAB, and the DNA repair gene BRCA2. Parallelly, downregulation of the tumor suppressor gene p21, GATA2, and MAP was observed [32]. Clinically, monosomy 7 is characterized by a lower median age of the affected patients as compared to 5q deletions, severe refractory cytopenias, and a proneness to life-threatening infections. Predisposing conditions recurrently observed in cases with monosomy 7 are antecedent aplastic anemia, Fanconi’s anemia, neurofibromatosis type I, cyclic neutropenia, and long-term treatment with granulocyte colony-stimulating factor (G-CSF), as well as Schwachman’s syndrome. It is of special interest that myeloid cells harboring monosomy 7 seem to be preferentially responsive to a stimulation by myeloid growth factors like G-CSF and granulocyte–macrophage CSF in vitro [33, 34] as well as in vivo [35]. Although being obvious, this phenomenon never was exploited therapeutically for a targeted priming strategy for patients with monosomy 7. In extremely rare cases, for the so-called monosomy 7 syndromes, the abnormality has a familiar background. In a substantial portion of patients, a mutagen exposition (benzene, solvents, irradiation, or radio- and/or chemotherapy) was documented. In contrast to the situation in 5q deletions, in monosomy 7, additional abnormalities do not have such a profound impact on outcome as seen in 5q deletions, since monosomy 7 even as an isolated abnormality confers a significantly bad prognosis. In our Austrian–German dataset, median survival times were 14 months for the isolated abnormality, 11 months for cases with one additional change, and 8 months for monosomy 7 as part of complex abnormalities (Fig. 3) [1]. Fig. 3Median survival according to accompanying abnormalities in patients with −7/7q−. Log-rank test: p = 0.03 (2 degrees of freedom), p = 0.94 (isolated vs. +1), p = 0.07 (isolated vs. complex), p = 0.43 (+1 vs. complex) Therapeutic options in the monosomy 7 subgroup are unsatisfying as yet. If age and clinical condition are adequate, patients should be treated with allogeneic stem cell transplantation whenever possible. Conventional intensive chemotherapy bears a high risk of early death and nonresponse. Even if a complete remission can be achieved, this frequently is of only short duration with a high risk of early relapse. In a recent report on 34 patients with MDS or AML treated with 5-azacytidine, the group of Mufti observed a preferentially good response of MDS patients with monosomy 7 to the demethylating agent 5-azacytidine. In this trial, five of the seven (71%) complete responders had isolated chromosome 7 abnormalities and achieved a continuous complete remission (10 months follow-up) in contrast to patients with other karyotype abnormalities like trisomy 8 who relapsed within the first 6 months [36]. Supporting results have been recently published for the use of decitabine by Lübbert et al. [37]. Future multicentric therapy trials are needed to verify this observation in a higher number of patients. Complex abnormalities According to the International System for Human Cytogenic Nomenclature criteria, complex chromomosome abnormalities are defined by the simultaneous occurrence of at least three independent abnormalities within one cell clone [38]. This cytogenetic subtype is present in some 15% of all patients with MDS (roughly 30% of all abnormal cases) and thus occurs in a frequency comparable to 5q deletions [1–3]. Complex abnormalities may be the result of a multistep process with sequential accumulation of abnormalities, called karyotype evolution. In some cases, karyotype evolution can be observed if repeated cytogenetic analyses are performed, unraveling the stepwise accumulation of secondary abnormalities with the starting point of a primary abnormality. In a few patients, cytogenetic analysis detects cell clones with primary and others with primary and secondary and possibly further abnormalities at one and the same time point. However, frequently, this process cannot be proven in the individual patient who presents with multiple chromosomal changes at first diagnosis. In most of these cases, complexity is profound with aberration numbers exceeding the threshold of three by far. Frequently, in these cases, cell-to-cell variations occur, and the chromosomal complement presents with a chaotic, mutator-like shape resembling cytogenetics of solid tumors and malignancies in chromosomal breakage syndromes. Thus, it is conceivable that, in these cases, the process leading to complex abnormalities and genetic instability must proceed rapidly on the one hand and must involve mechanisms like DNA repair and cell cycle control comparable to the situation in DNA repair deficiency syndromes. Recent results of gene expression analyses in patients with MDS and complex chromosome abnormalities support these assumptions and are considered below in more detail [39]. In a substantial portion of patients with complex abnormalities, an exposition to therapeutically applied mutagens, like anthracyclines, topoisomerase II inhibitors, alkylating agents, and/or irradiation, can be traced in the medical history [40]. Within the subgroup of patients with complex changes, a broad range of the number of abnormalities with the majority of patients displaying greater than or equal to five chromosomal changes can be observed [39]. There is a strong association with unbalanced structural abnormalities affecting most frequently chromosomes 5q and 7q. Chromosomes less frequently involved in loss of genetic material were chromosomes 3 (p- and q-arms), 12p, 13q, 16q, 17p, 18q, and 20q. Genomic gains were observed for chromosome 8/8q, 11q, and 21q [41]. Mutations of p53 have been observed in up to 90% of patients with AML and complex abnormalities [39]. Matrix comparative genomic hybridization (CGH) analyses in patients with complex abnormalities revealed an association of complex karyotype changes with amplifications in the chromosomal regions 8q24, 9p24,,11q23, 12p13, 13q12, 20q11, and 2q1q22 [42]. Recently, characteristic profiles of complex abnormalities in AML were delineated by gene expression analyses. In this respect, a significant overexpression of RAD21, a double-strand-break DNA repair enzyme, as well as overexpression of other genes related to DNA-repair, apoptotic mechanisms, and cell-cycle control (RAD1, RAD 9A, RAD23B, RAD51AP1, NBS1, MSH6, SUMO1, and PARP2) was observed [39]. Taken together, these data might help to understand complex abnormalities as a maximum manifestation of genetic instability by allowing cells with significant DNA damages to circumvent physiologically protective apoptosis and to escape cytotoxicity of chemotherapy by intensified DNA repair. This model is also compatible with the high extent of chemo-resistance of hematopoietic cells with complex chromosome abnormalities, which is a well-known and severe problem in the clinical management of high-risk MDS. Further clinical associations are the high median age of patients within this cytogenetic subgroup [43] and, related to conventional intensive chemotherapy, short remission durations and a high risk for early relapse [44]. Allogeneic stem cell transplantation, the only curative option for these patients, will only be feasible for a minority of individuals with complex chromosomal changes due to age-related multimorbidity and organ dysfunctions. New therapeutic strategies targeting hypermethylation, deacetylation, and immunomodulation, which have proven their tolerability also in aged patient populations, are now available for the treatment of patients with high-risk MDS. In this respect, two studies have to be mentioned. Cytogenetic responses in a substantial portion of patients with high-risk MDS have been observed in a multicentric study of decitabine, a potent demethylating agent. Major cytogenetic responses (MajCR) were observed in 38% of 50 patients. Remarkably, 10 out of 26 patients (38%) with high-risk cytogenetics displayed a MajCR; five of them had complex chromosome abnormalities. In the patients’ group with MajCR, median survival time was 24 months as compared to the significantly shorter median survival time of 11 months in patients with persisting cytogenetic aberrations [45]. These observations are further supported by observations of our group. Within a cohort of 14 patients with high-risk MDS, we recently observed hematologic and cytogenetic responses to 5-azacytidine according to the modified International Working Group criteria [46] in three out of four patients with complex abnormalities [47]. Another new therapeutic mechanism, immunomodulation, might be effective in cases with complex karyotypic changes too. In this connection, a complete cytogenetic response to the immunomodulatory drug lenalidomide has been reported in patients with high-risk MDS with complex chromosome abnormalities harboring 5q deletions [48]. Further multicentric clinical trials are needed to confirm these observations. Conclusions In contrast to CML where the bcr–abl fusion forms the homogenous genetic basis of the disease [49], MDS show a profound heterogeneity, not only on the morphologic and clinical level but also on the genetic presentation. Three cytogenetic subgroups with prognostic relevance have been identified and were incorporated into prognostic scoring by the IPSS. However, in the great majority of cytogenetic changes, the prognostic relevance is still obscure. Genetic heterogeneity in MDS leads to the paradox situation that rare abnormalities can be observed frequently [1] with the consequence that, for a relevant portion of abnormalities, their prognostic impact is still unknown. The only chance to overcome these shortcomings is to establish multicentric cooperations, which are the goals of the German–Austrian MDS Study Group, which is now cooperating with the MD Anderson Cancer Centre as well as the IMRAW group [50–52]. New therapeutic strategies targeting immunomodulation and epigenetic changes proved to be of outstanding effectiveness and tolerability in comparison to established therapies not only for the low-risk 5q syndrome but also for high-risk MDS with as yet deleterious cytogenetic alterations like monosomy 7 and complex abnormalities. However, what about the subgroup of nearly 50% of patients presenting with a normal karyotype? New techniques will help to further delineate the genetic background in these cases. Thus, very recently, two groups published their results of single nucleotide polymorphism analyses in MDS [53, 54]. The group of Mufti performed high-resolution single-nucleotide polymorphism (SNP) array analyses in 119 low-risk MDS patients of whom 32% displayed clonal chromosome abnormalities to search for cryptic chromosomal aberrations not detectable by chromosomal banding analyses. They found uniparental disomy (UPD), a copy number neutral loss of heterozygosity, in 46% of patients, small deletions in 10%, and amplifications in 8% of the cohort. The changes of copy numbers were acquired while UPDs were constitutional. The former aberrations were associated with a worsened outcome of the patients. The authors speculated that the high frequency of UPD may be indicative for a constitutional genomic instability in these patients [53]. This could be a predisposing factor for MDS. Maciejewski et al. applied 250 K SNP arrays to 94 patients with MDS and observed SNP aberrations in 75% of patients as compared to 59% clonal chromosomal abnormalities detected by metaphase analyses. Previously unknown lesions were found in patients with a normal—as well as in others with an abnormal—karyotype. In this study, UPD occurred in 20% of the MDS patients [54]. New analytical tools like matrix CGH, microarray gene expression analyses, proteomics, and methylation profiling will add substantially to the understanding of pathogenesis, delineation of therapeutic targets, and individualization of therapy in MDS. Taken together cytogenetics is still the gold standard of genetic diagnosis in MDS providing “labels” like 5q−, monosomy 7, or complex abnormalities for disease entities with a common biological behavior and clinical as well as prognostic impact. In the future, scientific progress will depend on the ability to combine established and newly developed methods to gain a comprehensive understanding of MDS, which will translate into an individualized and highly effective and well-tolerable treatment for all patients with MDS.

Ann_Surg_Oncol-4-1-2234441

A Novel Function of the Receptor for Advanced Glycation End-Products (RAGE) in Association with Tumorigenesis and Tumor Differentiation of HCC

Background The expression of the receptor for advanced glycation end products (RAGE) has an impact on the mechanisms giving rise to characteristic features of various cancer cells. The purpose of this study was to elucidate the clinicopathological relevance of the level of RAGE expression in patients with hepatocellular carcinoma (HCC) and to explore the effect of RAGE expression on the characteristic features of HCC. Although many cancers arise from chronic inflammation, the relationships between carcinogenesis, cancer promotion, and its molecular characteristics remain poorly understood. Hepatocellular carcinoma (HCC), which typifies an inflammation-related tumor, is one of the most common malignancies in the world, especially in Asia and Africa. Japan has a high incidence of chronic viral hepatitis, cirrhosis, and HCC. The resolution of inflammatory activity at the molecular level may correlate with prevention of hepatocarcinogenesis and cancer promotion. The receptor for advanced glycation end-products (RAGE) is a multiligand receptor classified as an immunoglobulin superfamily cell surface molecule and acts as a counter-receptor for high-mobility group box 1 (HMGB1),1 advanced glycation end-products (AGEs), S100/calgranulins, and amyloid-β peptides. These interactions trigger the activation of key cell signaling pathways (e.g., p38 and p44/42 MAP kinase,2 NF-κB, cdc42/rac,3 and the generation of reactive oxygen species, and result in the production of proinflammatory cytokines.5 RAGE-mediated proinflammatory processes are now considered to contribute to the progression of many chronic diseases, such as neuropathy, nephropathy,6 macrovascular disease, amyloidoses, inflammatory conditions (e.g., rheumatoid arthritis and inflammatory bowel disease) and sepsis.5,7 In addition to RAGE-mediated proinflammatory events, recent studies have revealed that the interaction of RAGE and its ligands and the resultant signaling play a causative role in the characteristic modulation of cancer cell functions, i.e., increasing tumor invasion and metastasis.8 Furthermore, several clinical studies have demonstrated the (strong) association of RAGE expression with the malignant potential of various cancers such as gastric cancer,9 colon cancer,10,11 common bile duct cancer,12 pancreatic cancer,13 and prostate cancer,14 although one report showed a reverse correlation between RAGE expression and tumor progression.15 Thus, RAGE expression may be expected play have a significant role in the development of HCC, although no data have been reported for this tumor. In terms of HCC, it is known that the pathophysiological conditions or circumstances surrounding tumorigenesis are quite different from those reported for other cancers. For example, small HCC tumors at an early stage are continuously exposed to low oxygen and high glucose as well as the noncancerous hepatic tissues with cirrhosis in which tissue derangement occurs.15 In addition, RAGE is the receptor for AGEs produced from excessive glucose metabolism,16 extracellular HMGB1 released from necrotic cells which could be induced by hypoxia and inflammation, and serum amyloid A (SAA) produced in response to the proinflammatory cytokine IL-6,17 and these RAGE ligands are possibly generated and/or released from inflamed hepatic tissues. Our central hypothesis is thus that RAGE expression may play a particular role in tumorigenesis of HCC, in addition to the role in the invasive and/or metastatic potential of cancer cells. With this background, the objectives of this study were: (1) to clarify the relationship between RAGE expression and the clinico-pathological features of HCC and (2) to investigate the functional role of RAGE expression in HCC development. MATERIAL AND METHODS Human samples From March 2000 to September 2005, 65 patients with primary HCC were treated surgically in the Department of Surgical Oncology and Digestive Surgery, Kagoshima University School of Medicine. Of these 65 patients, 12 who had diabetes mellitus and 6 who underwent preoperative therapy were excluded from the study. RAGE expression for Diabetes patients upregulates at various tissues.18,19,20 A further 11 patients were excluded because their RNA samples were degraded. Samples from the remaining 36 patients (30 men and 6 women with a mean age of 67.1 years) were included in the study. Seven patients (19.4%) were positive for hepatitis B surface antigen and 21 (58.3%) were positive for the antibody to hepatitis C virus. Eight patients (22.2%) were negative for both of these viruses. Twenty-four patients had chronic hepatitis and four had liver cirrhosis. The mean tumor size was 49.7 mm (range, 16–150 mm). The histological grade of each tumor was determined according to the general rules for the clinical and pathological study of primary liver cancer (The Liver Cancer Study Group of Japan, 2000).21 Four tumors (11.1%) were well-differentiated HCC, 28 (77.8%) moderately differentiated HCC, and 4 (11.1%) poorly differentiated HCC. Postoperative tumor recurrence was observed for 11 patients (30.6%) (Table 1). For immunohistochemical study, a further 12 HCC samples (six well- and six poorly differentiated HCC), obtained surgically during the same period at JA Kagoshima Kouseiren Hospital, were added and used to verify the correlation of RAGE expression with tumor differentiation. Finally, a total of 48 HCC nodules (10 well-differentiated HCC, 28 moderately differentiated HCC, and 10 poorly differentiated HCC) were studied. As a control study, six normal liver samples were collected from patients with benign or metastatic liver tumors. Written informed consent, recognized by the ethical committees of Kagoshima University School of Medicine and JA Kagoshima Kouseiren Hospital, was obtained from each patient before tissue acquisition. TABLE 1.Background of patientsGender  Male30 (83%)  Female6 (17%)Mean age67.2 yearsVirus type  B7 (19.4%)  C21 (58.3%)  None8 (22.2%)Background of livers  Chronic hepatitis24 (60.7%)  Liver cirrhosis4 (11.1%)  Normal liver8 (22.2%)  Mean tumor size 49.7mmHistological grade  Well differentiated4 (11.1%)  Moderately differentiated28 (77.8%)  Poorly differentiated4 (11.1%)Postoperative tumor recurrence 11 (30.6%). Immunohistochemistry Consecutive 4-μm sections were cut from each paraffin-embedding block. Sections were immunostained by anti-RAGE antibody (Santa-Cruz, CA, USA) according to the conventional immunoperoxidase technique. Briefly, after peroxidase blocking with 3% H2O2/methanol for 10 min, specimens were blocked with phosphate buffered saline (PBS) containing 5% normal horse serum (Vector Laboratories, Inc., Burlingame, CA, USA). Anti-RAGE antibody was used at 1/200. After overnight incubation at 4°C with the primary antibody, specimens were briefly washed in PBS and incubated at room temperature with the secondary antibody conjugated with peroxidase. The specimens were then washed in PBS and color-developed by diaminobenzidine solution (DAKO). After washing with water, specimens were counterstained with Meyer’s hematoxylin (Sigma Chemical Co., St Louis, MO, USA). Immunostaining of all cases was performed at one time to ensure the same conditions of antibody reaction and DAB exposure. A total of 48 HCC nodules (10 well-differentiated HCC, 28 moderately differentiated HCC, and 10 poorly differentiated HCC) were studied. To evaluate the immunohistochemical staining, ten fields were selected and expression in 1000 tumor cells (100 cell/fields) was evaluated with high-power (×200) microscopy. The immunohistochemical expression of RAGE was defined as positive if distinct staining of the cell membrane was observed in at least 10% of tumor cells. Cell and cell culture Hepatoma cell lines HepG2, HuH7, HT17, and Li7 were kindly provided by the Cell Resource Center for Biomedical Research Institute of Development, Aging, and Cancer, Tohoku University and Hep3B was obtained from the European Collection of Cell Cultures. HepG2, Hep3B, and HT17 were cultured with Dulbecco’s Modified Eagle’s Medium (DMEM) (1000 mg/l) and HuH7 and Li7 were cultured with RPMI. RAGE-transfected Cos7 and its mock-transfectant were kindly provided by Drs. Yamamoto, Department of Biochemistry and Molecular Vascular Biology, Kanazawa University. These transfectants were maintained with DMEM supplemented with 10% Fetal Calf Serum (FCS) in the presence of 650 μg/ml G418. A hypoxic environment was created by placing a cell culture dish with one pouch of Anaero Pack into an airtight jar. (Mitsubishi Gas Chemical Co., Inc., Tokyo, Japan). This created nearly 0% O2 conditions. Gene silencing of RAGE with specific siRNA Cells were seeded in 12-well plates at a density of 1 × 105 cells per well and allowed to adhere overnight. Then siRNAs for the target gene or its control oligoribonucleotide mixed with Dharma FECT2TM transfection reagent (Dharmacon Inc. Chicago, USA) was added to the cells and incubated for 48h at 37°C The efficacy of gene silencing was evaluated using immunoblot analysis. MTT assay Cell viability was monitored after incubation for 24, 36, and 48 hours by MTT assay. Briefly, 0.5 mg/mL 3-[4,5]-2,5-diphenyltetrazolium bromide (MTT) in fresh medium was added to each well and the cells were incubated for an additional 3 hours. Afterwards, the blue formazan crystals were dissolved in 1 mL isopropanol and measured spectrophotometrically at 570 nm. Immunoblot analysis Whole cell lysates were prepared as per the Santa Cruz protocol. One milliliter of Radio-Immunoprecipitation Assay (RIPA) buffer was added to a 100 mm cell culture plate. The plates were gently rocked for 15 min at 4°C. Adherent cells were scrpaed with a cell scraper, followed by incubation for 30–60 min on ice. The cell lysate was microcentrifuged at 10,000g for 10 min at 4°C. The supernatant fluid was the total cell lysate. The supernatant was transferred to a new microfuge tube and the pellet was discarded. Twelve microgram lysates were subjected to immunoblot analysis using a 12.5% sodium dodecyl sulfate (SDS) -polyacrylamide gel followed by electrotransfer onto nitrocellulose filters. The filters were immunoreacted with anti-RAGE antibody (a gift from TORAY Research Institute, Sagamihara, Japan) or with anti-HMGB1 antibody (BD Biosciences, Tokyo, Japan) and then incubated with peroxidase-conjugated anti-goat IgG (Medical and Biological Laboratories, Nagoya, Japan). The immune complex was visualized using the Enhanced Chemiluminescence (ECL) Western blot detection system (PIERCE, Rockford, IL, USA). The amount of B-actin as an internal control was also examined using a specific antibody (Cytoskelton Inc., Denver, CO, USA and Santa Cruz, CA, USA). At least three independent experiments were performed. Quantitative RT-PCR For reverse-transcription PCR (RT-PCR) and real-time quantitative PCR, total RNA was extracted from 30 mg frozen tissue using Total RNA Mini (VIOGENE, CA, USA). For cDNA synthesis, the RNA samples (1 μg) were converted into cDNA by reverse transcription (RT) using random primers (TAKARA, Siga, Japan) according to the manufacturer’s instructions. To estimate the mRNA expression levels of several genes quantitatively, polymerase chain reaction (PCR) amplification was performed using a Light-Cycler system (Roche, Mannheim, Germany) and the Light-Cycler Fast Start DNA Master SYBER green I kit (Roche). Primers were as follows: RAGE: 5′-AAA CAT CAC AGC CCG GAT TG-3′ and 5′-TCC GGC CTG TGT TCA GTT TC-3′, HMGB1: 5′-GCT CAG AGA GGT GGA AGA CCA-3′ and 5′-GGT GCA TTG GGA TCC TTG AA-3′ (14) , GAPDH: 5′-TTG GTA TCG TGG AAG GAC TCA-3′ and 5′-TGT CAT CAT ATT TGG CAG GTT T-3′. Amplification was carried out in 20 μL reactions containing 4 mM MgCl2, 2 μL of primer, 2 μL of Light-Cycler-FastStart DNA Master SYBR green I reagent, and 2 μL of cDNA. Reaction conditions were an initial incubation at 95°C for 10 min, followed by 50 cycles at 95°C for 10 s for denaturation, 64°C for 10 s for annealing of the RAGE primers, 54°C for 10 s for annealing of the HMGB1 primers and 60°C for 10 s for annealing of the GAPDH primers and 72°C for 10 s for extension. Melting curves were obtained according to the protocol under the following conditions: 0 s denaturation period at 95°C, starting temperature of 65°C, end temperatures of 95°C, and rate of temperatures increase of 0.1°C/s. The quantitative value of the target gene (RAGE mRNA) in each sample was normalized using GAPDH expression as an internal control. The quantitative RT-PCR assay was carried out twice and the mean value was calculated. Finally, the mRNA expression ratio of cancerous (C) to noncancerous (N) tissues was calculated using the following formula: R=log{target gene (C)/GAPDH (C)}, R=log{target gene (N)/GAPDH (N)}. These experiments were carried out twice to confirm reproducibility. Statistical analysis Statistical analysis was performed using the JMP IN version 5.1.2 software system (SAS institute Inc., Cary, NC, USA). Each value of mRNA expression was log transformed before statistical analysis. Gene expression was compared among normal liver, hepatitis, and HCC using Student’s t-test. The relationships between RAGE-, HMGB1-mRNA expression levels and clinicopathological features were evaluated using Student’s t-test and the Mann–Whitney U test, as appropriate. Immunohistochemical study of RAGE in HCC tissues was evaluated using the χ2 test. A p value of less than 0.05 was considered to be statistically significant. RESULTS RAGE and HMGB1 expression in normal liver, hepatitis, and HCC RAGE antibody yielded a strong band compared to control mouse lung extracts. Using Western blotting and RT-PCR, protein and mRNA expression of RAGE and HMGB1 were examined in both cancer and noncancerous tissues from three cases (Fig. 1a, b). All 3 cases showed co-expression of RAGE and HMGB1 protein and mRNA in these tissues. FIG. 1.(a) RAGE expression by Western blotting (A) and RT-PCR (B). (b) HMGB1 expression by Western blotting (A) and RT-PCR (B). Abbreviations: normal, normal liver; CH, chronic hepatitis. Quantitative RAGE mRNA expression in HCC and noncancerous lesions Comparing the quantitative expression of RAGE mRNA in paired cancer and noncancerous tissues (i.e., chronic hepatitis or liver cirrhosis) of 36 cases, HCC tissues showed significantly higher expression than noncancerous tissues (p < 0.01, Fig. 2a). Moreover, the mean values of RAGE mRNA expression in cancer and noncancerous tissues were higher (p < 0.01 and p = 0.08) than that in normal liver tissues (i.e., non-inflamed liver from benign or metastatic liver tumor patients) (Fig. 2b). FIG. 2.(a) Quantitative RAGE mRNA expression in paired cancerous (HCC) and noncancerous tissues from 36 cases. (b) Quantitative RAGE mRNA expression in normal liver (n = 6) and noncancerous and cancerous tissues (n = 36). Relationship between RAGE mRNA expression and clinicopathological features To elucidate the biological significance of RAGE expression in HCC, we compared the levels of RAGE mRNA expression with the clinico-pathological features of 36 patients. As shown in Table 2, we noted significant differences in RAGE mRNA expression in association with gender, age, the levels of protein induced by vitamin K absence or antagonist (PIVKA-II), and postoperative recurrence. In terms of gender and age, the results are in accordance with a previous report.23 The expression levels of RAGE mRNA tended to be lower in poorly differentiated tumors compared to well- or moderately differentiated tumors (p = 0.06). Moreover, the levels of RAGE mRNA showed a negative correlation with PIVKA-II levels and the presence or absence of recurrence. There were no significant differences regarding tumor size, intrahepatic metastasis, and vascular invasion. On the other hand, it should be noted that a significant difference in the levels of RAGE expression was observed between hepatitis virus-positive and virus--egative tissues (p < 0.01) in the study of noncancerous inflammatory liver tissues (i.e., chronic hepatitis or liver cirrhosis) (data not shown). TABLE 2.Relationship between tumor RAGE expression and clinicopathological featuresFactorsTumor RAGE mRNA expressionp valuenMean ± SDGender  Male302.267 ± 0.3410.03  female62.777 ± 0.619Age  ≥65 years262.468 ± 0.5680.03  <65 years102.052 ± 0.638Tumor size (mm)  ≥30112.395 ± 0.6130.39  <30252.333 ± 0.620Portal invasion  Absent222.358 ± 0.6550.52  Present142.348 ± 0.594Venous invasion  Absent262.339 ± 0.5770.47  Present102.357 ± 0.633Vascular invasion  Absent192.362 ± 0.6330.46  Present172.340 ± 0.601Intrahepatic metastasis  Absent272.378 ± 0.6620.33  Present92.273 ± 0.435Gross classification  Localized type212.448 ± 0.6360.13  Invasive type152.217 ± 0.564Differentiation  Well42.365 ± 0.5660.43  Moderately282.420 ± 0.5800.08  Poorly41.860 ± 0.783Stage  I,II162.367 ± 0.6430.55  III,IV202.340 ± 0.597PIVKA‡U  Normal92.704 ± 0.385  High252.186 ± 0.6300.03AFP  Normal122.557 ± 0.5860.16  High242.250 ± 0.607Virus  B72.237 ± 0.606  C212.382 ± 0.569  None82.374 ± 0.779Recurrence  Absent252.485 ± 0.6280.04  Present112.050 ± 0.456 RAGE expression and tumor differentiation in HCC To verify the correlation of RAGE expression with HCC differentiation, we carried out an immunohistochemical examination. (Fig. 3a–c) Well- and moderately differentiated HCC showed a high percentage of RAGE positivity (70% and 64%, respectively); only 2 of 10 (20%) were positive in poorly differentiated HCC (Fig. 3d). Statistically, there were significant differences in the rates of positivity between well- and moderately differentiated HCC and poorly differentiated HCC. FIG. 3.RAGE expression by immunohistochemical staining: (a) well-differentiated HCC, (b) moderately (mod.) differentiated HCC, (c) poorly differentiated HCC, (e) the number of RAGE positive or negative cases according to tumor differentiation. RAGE expression in hepatoma cell lines RAGE antibody yielded a strong band on Western blot analysis of the control RAGE transfected Cos7 cells (Fig. 4a, lane 1). In this experimental setting, the well-differentiated HCC cell line HuH7 (lane 6) expressed RAGE protein at a modest level. The poorly differentiated HCC cell line HT17 (lane 7) expressed RAGE protein at a low level. Hepatoma cell lines Hep3B and Li7 (lane 4 and 5), derived from tumors of unknown state of differentiation, expressed RAGE protein at a modest level. HepG2 (lane 3) did not express RAGE protein at all. These results were very similar to RAGE mRNA expression and depend on the pathological differentiation of HCC. FIG. 4.(a) RAGE expression in hepatoma cell lines by Western blotting. (b) RAGE expression in hepatoma cell lines under hypoxic conditions by Western blotting. N, normoxic conditions; A, anaerobic conditions. Enhanced RAGE expression under hypoxic conditions Early carcinogenesis occurs in an avascular environment and under conditions of hypoxia.15,23,24 In order to clarify the role of RAGE expression under hypoxic conditions, all hepatoma cell lines were cultured in an anaerobic environment. Li7 could survive for 48 h, Hep3B for 12 h, and HuH7 for 24 h under such conditions, up to the appearance of signs of cell death. HepG2 and HT17 rapidly died under anaerobic conditions. The RAGE-positive cell lines (Li7, Hep3B and HuH7) were more tolerant of anaerobic conditions than RAGE-negative or weakly expressing cell lines (HepG2 and HT17). Moreover, RAGE expression in the cell lines that survived was clearly upregulated in response to the hypoxic conditions (Fig. 4b). Enhancement of cell survival in anaerobic conditions following RAGE transfection These findings led to the hypothesis that RAGE expression could confer tolerance to hypoxia in HCC. Therefore, we investigated the survival of RAGE-transfected and mock-transfected cells under hypoxic conditions. In continuous incubation under hypoxic conditions, as expected, RAGE-transfected Cos7 cells clearly survived longer than mock-transfected Cos7 cells (observed at 36 h and 48 h) (Fig. 5). FIG. 5.Comparison of cell survival of RAGE-transfected Cos7 and mock-transfected Cos7 cells under anaerobic conditions. Cell survival of both groups was estimated from six dishes at each time point using the MTT assay. *p < 0.01 by Student’s t-test. Decline of cell survival in hypoxic conditions following RAGE reduced by siRNA Consistent with the previous result, reduction of RAGE expression with anti-RAGE siRNA re-increased susceptibility to hypoxia-induced injury in RAGE-overexpressed Cos7 transfectant (Fig. 6a,b), suggesting that RAGE expression might play an important role in the acquirement of hypoxia-resistant cellular phenotype. FIG. 6.RAGE level in RAGE-transfected Cos7 cells were reduced by siRNA RAGE-transfected Cos7 cells reduced RAGE level by siRNA decreased more than the cells mixed with only transfection reagent. *p < 0.01 by Student’s t-test. DISCUSSION Hepatocellular carcinogenesis and associated tumorigenesis occur in a stringent and restrictive environment, such as the cirrhotic liver in which tissue oxygen supply is insufficient. Moreover, in the liver, with a dual blood supply from the hepatic artery and portal vein, liver-specific oxygen and glucose supply are available dependent upon the degree of liver disorder. The multiligand receptor RAGE is particularly relevant in this context. The ligands for RAGE may be produced abundantly during HCC development. For example, HMGB1 released from necrotic cells and AGE generated in the process of higher hepatic glucose metabolism are considered to be major candidates (i.e., the presence of HMGB1 and carboxymethyl lysine in HCC; data not shown). Hepatic SAA production in response to IL-6 may also be included.19 In addition, RAGE expression and its downstream signaling are now considered to play a significant role in the progression of various cancers as well as the development of inflammatory lesions. Another aspect of the biology of RAGE was suggested by the features of retinoic acid-induced neuroblastoma differentiation, in which RAGE expression played a more important role in cellular survival than in neurite outgrowth.25 In this model, inhibition of RAGE function partially blocked the increase in levels of the anti-apoptotic protein Bcl-2 in the process of neuronal differentiation,25 indicating that RAGE and its signaling also might contribute to the survival of certain cancer cell types undergoing “differentiation”. Furthermore, although indirectly, our previous study demonstrated that NF-κB, currently considered as a causative transcriptional factor for various inflammatory events and also known as a major RAGE-mediated signal, played an important role in the survival of neuronal tumor cells26,27 and in experimentally developed HCC.28 The present study demonstrated that increased RAGE expression was highly associated with the status of pathological “differentiation” in HCC, which played a significant role in acquisition of the hypoxia-resistant phenotype of tumor cells. This conclusion is supported by several lines of experimental evidence. First, the level of RAGE expression was higher in well- and moderately differentiated HCC, while it diminished as the tumors dedifferentiated to poorly differentiated HCC. This was consistent with the evidence that a negative correlation was observed between the level of RAGE mRNA expression and either the level of PIVKA II or the incidence of postoperative recurrence. Second, the analysis of five HCC lines revealed that three of these (Li7, Hep3B, and HuH7) that are resistant to hypoxic stress characteristically showed higher levels of RAGE expression compared to the two hypoxia-intolerant cell lines HepG2 and HT17. Third, sublethal hypoxia exposure induced significantly increased RAGE expression in hypoxia-resistant HCC lines. In the analysis of the association between the level of RAGE expression and the “differentiation status” of HCC lines, the level of RAGE expression was higher in “possibly differentiated” lines (i.e., HCC with low Alpha-Fet protein (AFP) production), consistent with the results from clinical samples. Finally, cells overexpressing RAGE exogenously showed prolonged survival under hypoxic conditions compared to control mock-transfected cells, and siRNA experiments demonstrated similar results. Our conclusion also provides a new hypothetical concept that hepatic RAGE expression may be relevant to the stage or severity of inflammation and the incidence of carcinogenesis and early tumorigenesis of HCC (Fig. 7). In the process of the development of an HCC lesion, increased RAGE expression by inflamed hepatocytes may confer adaptation to an advanced hypoxic environment during remodeling and carcinogenesis with accelerated cell proliferation. In this context, an HCC lesion which is highly associated with inflammation caused by either hepatitis viruses or drugs is developed through a process of multistage carcinogenesis, i.e., from inflammatory lesions (e.g., hepatitis, cirrhosis and precancerous lesion) to adenomatous hyperplasia, and eventually into HCC. Once cancer is established, HCC dedifferentiates step-by-step to a more malignant histology, from well- and moderately to poorly differentiated HCC. It has also been pointed out that the vascular supply changes significantly between each stage of tumorigenesis. According to the evidence from several clinical studies,15,29,30 both adenomatous hyperplasia and well-differentiated HCC are hypovascular tumors primarily fed by the portal vein system, while moderately and poorly differentiated HCC are hypervascular tumors primarily fed by arterial blood. In addition, early carcinogenesis and the development of the malignant phenotype generally occur in an avascular environment.31–33 Increased RAGE expression, at least in part, may thus play an important role in the mechanism of early HCC development from precancerous inflammatory lesions. In addition, about the expression of RAGE being lower in poorly differentiated HCC when compared to well and moderately differentiated HCC, gastric and colon cancers have been showed that RAGE and other ligands interacts act on tumor promotion in clinical materials.9,11 However, prostate cancer was demonstrated to show no correlation between RAGE expression and tumor differentiation and depth.14 Moreover, non-small-cell lung cancers have been showed a reverse correlation between RAGE expression and tumor stage.34 This result is partially similar to our results. RAGE expression of each clinical samples is various. FIG. 7Scheme of the change of RAGE expression according to the sequential change of liver tissue: normal → chronic hepatitis (CH) → cirrhosis → HCC. The value indicates the quantitative RAGE mRNA expression. Some reports have shown that RAGE-expressed cells have invasion and migration potential. Our data from clinical samples did no’ relate to the potential. In vivo, invasive and metastatic potential are reflected by many factors, which may have caused our results. Our results raise many questions concerning mechanistic and practical processes. It is important to know whether RAGE can bind HMGB1 secreted from activated macrophages in hepatic inflammation, or if occupancy of RAGE by inhibitors would obviate binding of the stimulatory ligands. The functional role of downstream signaling of the RAGE would also be important to determine the cytoprotective mechanism under hypoxia. Our findings have provided the first evidence of the clinical relevance and function of RAGE in HCC, namely differentiation-associated RAGE expression that confers a hypoxia-resistant phenotype. Although other mechanisms may also be important, our data also introduce the concept that RAGE and its functions may be possible candidates for therapeutic targets in the treatment of HCC.

Qual_Life_Res-3-1-2039846

Choosing between measures: comparison of EQ-5D, HUI2 and HUI3 in persons with hearing complaints

Objectives To generate insight into the differences between utility measures EuroQol 5D (EQ-5D), Health Utilities Index Mark II (HUI2) and Mark III (HUI3) and their impact on the incremental cost-effectiveness ratio (ICER) for hearing aid fitting Introduction Hearing loss affects the ability to exchange information and therefore affects a person’s quality of life [1]. Of the western adult population, approximately 15% are hearing impaired [2], and for these persons hearing aid use has proven to be effective [3, 4]. The growth of the elderly population has far-reaching implications for auditory health service delivery and expenditure since the prevalence increases heavily with age. As a result, increasingly more economic evaluations are undertaken on interventions such as hearing aid fitting. In economic evaluations health related quality of life (HRQoL) should be measured with a preference-based utility measure [5]. Until recently, the benefits of hearing aid fitting had not translated into a statistically significant improvement in health state utility [6–8]. In 2004 Barton et al [9] confirmed this for the EuroQol 5D (EQ-5D), but did find statistically significant utility gain after hearing aid fitting with the Health Utilities Index Mark III (HUI3). This difference may result from differences in the descriptive system, and the way the scoring function is derived. Also with regard to other conditions, a number of studies have found that different utility measures tend to lead to different utility scores [10–22]. In general there is need for head-to-head comparisons of utility measures, in order to assess the implications for the interpretation and comparability of economic analyses, especially in conditions where only subtle changes after treatment are expected [23, 24]. In these comparisons the comparison of different tariffs should also be incorporated. But most importantly, the impact of the differences between measures on incremental cost-effectiveness ratios (ICERs) should be made apparent. Surprisingly, the latter has not yet received much attention. The objective of this article is to compare the two most frequently used utility measures in economic analyses [25] in a Dutch population with hearing complaints. Utility scores derived with the EQ-5D UK tariff [26], the EQ-5D Dutch tariff [27], the Health Utilities Index Mark 2 (HUI2) and HUI3 [28] are compared to generate further insight into the differences between the measures, and the impact on the ICER for hearing aid fitting. In particular, this article considers:practicality of using the EuroQol and HUI in a population with hearing complaints;construct validity of the EQ-5D UK tariff, EQ-5D Dutch tariff, HUI2 and HUI3;agreement between the EQ-5D UK tariff, EQ-5D Dutch tariff, HUI2 and HUI3;responsiveness of the EQ-5D UK tariff, EQ-5D Dutch tariff, HUI2 and HUI3 after hearing aid fitting;and the impact of on the ICER for hearing aid fitting. Methods The EQ-5D, HUI2 and HUI3 The five questions of the EQ-5D descriptive system each represent one dimension of health-related quality of life (mobility, self-care, usual activities, pain/discomfort and anxiety/depression) [29]. Each question has three levels and the questions together classify persons into one of 243 health states. The commonly used scoring function is based on a British study (EQ-5D UK tariff) [26], with preferences derived with the time trade-off (TTO) method, in a representative sample of the UK population of 2,997 respondents. The scoring function is additive and the possible range of utility scores is −0.59 to 1.00. Recently, Lamers et al [27] developed a Dutch scoring function for the EQ-5D (EQ-5D Dutch tariff), based on TTO in a sample of 298 respondents, with utility scores ranging from −0.33 to 1.00. The 15 questions of the HUI descriptive system classify respondents into either HUI2 or HUI3 health states. The HUI system focuses on health ‘within the skin’, meaning that they purely focus on impairment and not on the social context of the impairment [28]. HUI2 consists of seven attributes (sensation, mobility, emotion, cognition, self-care, pain and fertility), with three to five levels, leading to 24,000 possible health states. The optional fertility attribute was not used in the present study. The multiplicative scoring function was derived using standard gamble (SG) and Visual Analogue Scale (VAS) in a random sample of 293 Canadian respondents and utility scores range from −0.03 to 1.00 [30]. HUI3 consists of eight attributes (vision, hearing, speech, ambulation, dexterity, emotion, cognition and pain), with five to six levels and leads to 972,000 possible health states. The multiplicative scoring function was derived from SG and VAS, in a random sample of the Canadian general population (n = 504), resulting in possible utility scores varying from −0.36 to 1.00 [31]. Study population and data collection Data were collected as part of a before-after study examining direct hearing aid provision versus provision by referral [32]. The study was carried out in 2004–2005 in three regions (Maastricht, Rotterdam and Amsterdam) in the Netherlands. Persons with hearing complaints were recruited from the participating Ear Nose and Throat (ENT) departments, audiological centres and hearing aid dispensers. Both EQ-5D and HUI were administered at the first visit to the ENT department. Questionnaires were completed at the department, and respondents were helped by a trained interviewer if requested. Patients who were fitted with a hearing aid were asked to attend the ENT department for a follow-up visit to evaluate the hearing aid fitting. During the evaluation at the ENT department this subset of patients completed both questionnaires for a second time. Again, questionnaires were completed at the department, and respondents were helped by a trained interviewer if requested. Practicality of the questionnaires Especially in an elderly population, an important aspect of a utility measure is the ease of completion. The practicality of using the EQ-5D and the HUI in a population with hearing complaints was therefore assessed by the completion rate, using a Wilcoxon Signed Ranks Test to test whether the completion rates were significantly different. Additionally we examined the item non-response. Construct validity In absence of a gold standard to measure health state utility, there is no clear technique to determine the construct validity of utility measures. A way to examine the construct validity is to examine whether utility scores are different for distinctive groups [33, 34]. Other studies have demonstrated differences of quality of life scores by sex [33, 35, 36] and age [35–38]. Comparisons were made between EQ-5D UK and Dutch tariff, HUI2 and HUI3 scores by age (above versus below median) and sex. It was expected that persons of a higher age and females have lower utility scores. In addition, the respondents were divided into five clinically distinctive groups, based on their hearing loss and hearing aid use. Hearing loss was defined as the better ear pure tone average (BEPTA) hearing loss for the frequencies 1000, 2000 and 4000 Hz. The groups were: persons who were not entitled to reimbursement of a hearing aid (BEPTA < 35 dB); persons who were entitled to reimbursement (BEPTA ≥ 35 dB) but did not apply for a hearing aid (non-applicants); first time hearing aid applicants; experienced hearing aid users who were about to have a new hearing aid fitted (re-applicants); and experienced hearing aid users who did not have a new hearing aid fitted. It was expected that persons with a BEPTA smaller than 35 dB would have a higher quality of life score than persons in the other four groups, because they are likely to experience less problems with hearing. It was also expected that non-applicants had a higher utility score than first time applicants, since the latter group is expected to experience more hearing complaints, resulting in fitting a hearing aid as a solution for their hearing problems. Descriptive summary statistics were provided and normality was tested for all data using the Kolmogorov–Smirnov test. Kruskal–Wallis one way ANOVA and pairwise comparison tests (Mann–Whitney U) were used to explore the differences between the groups. Agreement To assess agreement between the measures, a Wilcoxon Signed-Rank test and a Paired t-test were used to test whether the scores of the EQ-5D (UK and Dutch tariff), HUI2 and HUI3 had the same distribution and mean. Correlations (Kendall’s Tau) and the Intra-class Correlation Coefficient (ICC) were computed. The ICC was based on a two-way mixed effect model, such that the subject effect was random and the instrument effect was fixed, and computed at the individual patient level. An ICC below 0.75 implies poor to moderate agreement; above 0.75 implies good agreement [34]. Responsiveness Responsiveness was tested in the subpopulation with a valid score both before and after hearing aid fitting. Effect size and standardized response mean were calculated. Effect size is the change in score divided by the standard deviation of scores at baseline. Standardized response mean is the change in score divided by the standard deviation of the change in score. Both were interpreted using benchmarks for effect size: 0.20 through 0.49 is interpreted as small, 0.50 through 0.79 as moderate and ≥0.80 as large [39]. Also, a Wilcoxon Signed Ranks Test and a Paired t-test were conducted on the before and after scores. The change in score after hearing aid fitting was tested for differences between first time hearing aid applicants and re-applicants. It was expected that re-applicants had a smaller change in utility, since they already had a hearing aid and therefore expectedly less hearing problems at baseline. Impact on the ICER To illustrate the impact on the ICER, straightforward one year ICERs of hearing aid fitting versus no hearing aid fitting (doing nothing) were calculated. Quality Adjusted Life Years (QALYs) were calculated using the area under the curve method, for the EQ-5D (UK and Dutch tariff), HUI2 and HUI3 [5]. Baseline utility scores were used to estimate the effects of ‘no hearing aid fitting’. Costs of no hearing aid fitting were assumed to be zero. This was compared to the costs and effects after hearing aid fitting. The costs of hearing aid fitting were calculated prospectively for each respondent, including General Practitioner (GP) visit(s), ENT visit(s) and hearing aid(s). We used standard costs for medical consumption [40], and the actual price of the hearing aid. As costs and utility scores are generally not normally distributed, a non-parametric bootstrap sampling method was used to calculate the 95% confidence interval around the ICERs [41]. Cost-effectiveness acceptability curves for all measures were created to characterize the likelihood that hearing aid fitting will be deemed cost-effective based on the incremental costs and outcomes, for a range of ceiling ratios. Ceiling ratios reflect societies’ maximum willingness to pay for a unit of outcome. Results Practicality Of the 337 persons with hearing complaints included in the study, in total 315 (93%) fully completed both the EQ-5D and the HUI descriptive system at baseline. Each item on the EQ-5D had six missing values, except for mobility (n = 5). Regarding the HUI, all questions had six or seven missing values, except for the questions on hearing in a group conversation (n = 12), pain and discomfort (n = 10), and hearing in a conversation with one other person (n = 8). Although completion rates were high for both questionnaires, EQ-5D was fully completed by 328 persons (97%), which is significantly more than the 318 persons (94%) who completed the HUI (P-value 0.012). Of the 315 persons who completed both EQ-5D and HUI at baseline, 173 persons (55%) had a hearing aid fitted. Of them, 82 (47%) attended the ENT department for the follow-up visit after hearing aid fitting. Ninety-one respondents (53%) who had a hearing aid fitted did not show at the follow-up visit because they had not finished their hearing aid fitting before the end of the study (n = 37), or because they had a hearing aid fitted at a dispenser not participating in the study (n = 54), and were therefore lost to follow up. Of the 82 respondents who did attend the follow-up visit, 70 (85%) fully completed both the EQ-5D and HUI descriptive system. Each item on the EQ-5D had five missing values, except for pain/discomfort (n = 6). Regarding the HUI, the hearing questions both had four missing values, and the other questions had four to seven missing values. Seventy-six persons (93%) fully completed the EQ-5D, while 71 persons (87%) fully completed the HUI. This difference is not statistically significant (P-value 0.059). Construct validity Mean age of the respondents was 69.6 years (sd 8.9; median 70), and BEPTA was on average 42 dB. The respondents were divided into groups below 70 years old (n = 156) versus 70 years and older (n = 159) and male (n = 189) versus female (n = 126), see Table 1. The scores on the EQ-5D, HUI2 and HUI3 were not normally distributed (Kolmogorov–Smirnov test, P-value 0.000). Only the EQ-5D detected differences in reported health by both age and sex. HUI3 detected a difference in utility between the age groups, HUI2 did not detect any differences. Table 1Comparison of mean scores, median scores and interquartile range (IR) at baseline according to different characteristics of the respondentsAgeSexBEPTAEQ-5D UK tariffEQ-5D Dutch tariffHUI2HUI3NMean % MaleMean MeanMedian (IR) MeanMedian (IR) MeanMedian (IR) MeanMedian (IR)Age<7015661.964%37.40.860.94 (0.24) 0.880.95 (0.20)0.780.79 (0.15)0.640.65 (0.36)≥7015977.057%47.10.810.80 (0.27)0.840.84 (0.23)0.760.80 (0.17)0.580.59 (0.38) P-value*0.0100.0250.1340.034SexMale18969.3100%43.40.850.88 (0.26)0.870.90 (0.19)0.770.80 (0.13)0.610.62 (0.37)Female12669.90%40.50.810.80 (0.27)0.830.84 (0.23)0.760.78 (0.18)0.610.63 (0.39)P-value*0.0190.0120.5750.965Clinically distinctive groupsBEPTA <35 dB6961.954%21.30.840.81 (0.27)0.860.84 (0.20)0.800.80 (0.20)0.700.66 (0.37)Non-applicants4674.259%42.10.800.87 (0.27)0.830.90 (0.23)0.770.78 (0.15)0.620.60 (0.41)First time applicants10871.262%46.90.840.85 (0.27)0.860.87 (0.23)0.760.78 (0.16)0.580.61 (0.28)Re-applicants6569.666%53.60.850.85 (0.27)0.880.86 (0.20)0.770.79 (0.17)0.560.61 (0.41)P-value**0.9960.9930.2640.004* Mann–Whitney U test** Kruskal–Wallis test Furthermore, the respondents were divided into five clinically distinctive groups: BEPTA < 35 dB (n = 69), non-applicants (n = 46), first time hearing aid applicants (n = 108), re-applicants (n = 65), and experienced hearing aid users not applying for a new hearing aid (n = 12). The latter group was excluded from the analysis because of the small sample size. Fifteen persons could not be classified into a clinical group because they were lost to follow-up after the first visit. Based on the EQ-5D and HUI2, no distinction could be made between any of the clinically distinctive groups. A logarithmic transformation was performed on the EQ-5D and HUI2 data to compensate for skewness. Even after transformation, and also when correcting for age, sex and BEPTA, no differences were found between the groups. Only HUI3 scores demonstrated a significant difference between the clinically distinctive groups (Kruskall–Wallis; P-value 0.004). More specifically, HUI3 found significant differences between persons with a BEPTA < 35 dB and first time applicants (Mann–Whitney U; P-value 0.002), and between persons with a BEPTA < 35 dB and re-applicants (P-value 0.001). HUI3 did not confirm our expectation that non-applicants had significantly higher utility scores than first time applicants. As expected, non-applicants stated less problems on the hearing attribute than first time applicants, but they also stated more problems on the ambulation and pain attributes. Agreement A summary of the utility scores is presented in Table 2. Mean utility scores for the population with hearing complaints were higher for the EQ-5D Dutch tariff (mean 0.86; standard deviation 0.18) and EQ-5D UK tariff (0.83; sd 0.21) than for the HUI2 (0.77; sd 0.14) and HUI3 (0.61; sd 0.24). The differences in mean scores and distributions are statistically significant. The UK and Dutch tariff of the EQ-5D and HUI2 and HUI3 were found to have good agreement, since they were strongly correlated (0.90; 0.71) and had an ICC of 0.98 and 0.74, respectively. The scores on all other measures showed statistically significant, but low correlations and their agreement was moderate to poor (Table 3). Table 2Utility scores in baseline population with hearing complaints (n = 315)MeasureMinimumMaximumMedian*Interquartile rangeMean**Standard deviationEQ-5D UK tariff−0.251.000.850.270.830.21EQ-5D Dutch tariff−0.031.000.860.190.860.18HUI20.231.000.790.150.770.14HUI3−0.071.000.620.380.610.24* All statistically significantly different: Wilcoxon Signed-Rank test; P < 0.01** All statistically significantly different: Paired t-test; P < 0.01Table 3Agreement in the baseline population with hearing complaints (n = 315)Pairs of utility functionsKendall’s Tau*ICC(95% Confidence interval)EQ-5D UK tariff versus HUI2r = 0.410.51(0.42–0.59)EQ-5D UK tariff versus HUI3r = 0.370.47(0.38–0.55)EQ-5D UK versus Dutch tariffr = 0.900.98(0.97–0.98)HUI2 versus HUI3r = 0.710.74(0.68–0.78)HUI2 versus EQ-5D Dutch tariffr = 0.400.51(0.42–0.59)HUI3 versus EQ-5D Dutch tariffr = 0.360.44(0.35–0.53)* All statistically significant, P < 0.01 Ceiling effects were observed in the EQ-5D UK tariff (Figs. 1 and 2), results were similar for EQ-5D Dutch tariff. As measured with the EQ-5D (both UK and Dutch tariff), 44% of the respondents reported perfect health, despite their hearing complaints. Measured with the HUI2 or HUI3, less than 1% of the respondents reported perfect health. For respondents reporting perfect health on the EQ-5D, mean utility scores were 0.83 on the HUI2 (range 0.35–1.00) and 0.71 on the HUI3 (range 0.06–1.00). Fig. 1Scatterplot of utility scores derived with EQ-5D UK tariff and HUI2Fig. 2Scatterplot of utility scores derived with EQ-5D UK tariff and HUI3 Responsiveness Responsiveness was tested in the subpopulation with a valid score both before and after hearing aid fitting (n = 70). The effect size and standardized response mean of the EQ-5D UK (0.05; 0.05) and Dutch tariff (0.03; 0.02) were less than small. HUI2 and HUI3 were more responsive to change after hearing aid fitting, since both had a moderate effect size (0.64; 0.55) and standardized response mean (0.57; 0.66). Mean change in utility after hearing aid fitting (Table 4) was highest when measured with the HUI3 (mean 0.12; sd 0.18) and HUI2 (0.07; sd 0.13), while almost no change was measured with the EQ-5D (UK tariff 0.01, sd 0.13; Dutch tariff 0.00, sd 0.12). The change in utility measured with HUI2 and HUI3 is statistically significant (Paired t-test, P-values 0.005 and 0.000). No change was observed in any attribute of the EQ-5D (Fig. 3a). The change in HUI2 utility score after hearing aid fitting occurred in the sensation attribute (Fig. 3b) and in the HUI3 score in the hearing attribute (Fig. 3c). Almost no change was observed in any of the other attributes of the HUI2 and HUI3. Table 4Change in health state utility after hearing aid fitting and ICER with 95% Confidence Interval (CI) (n = 70)MeasureMeanStandard deviationMedianInterquartile rangeMinimumMaximumICERa €/QALY(95% CI) €/QALYEQ-5D UK tariff0.010.130.000.04−0.600.27286,866(inferiorb–47,082)EQ-5D Dutch tariff0.000.120.000.04−0.600.28647,209(inferiorb–61,934)HUI20.07*0.130.08**0.12−0.500.4025,337(19,356–38,012)HUI30.12*0.180.13**0.22−0.220.6015,811(11,664–24,654)* Statistically significant; Paired t-test; P < 0.01** Statistically significant: Wilcoxon Signed-Rank test; P < 0.01a ICER based on mean scoresb Inferior means higher costs and lower utilityFig. 3Percentage of responses for each level of each attribute before (b) and after (a) hearingaid fitting for EQ-5D, HUI2 and HUI3 The mean change in utility score after hearing aid fitting, when measured with HUI2 and HUI3, was higher for first-time hearing aid applicants (0.08; 0.13) than for re-applicants (0.06; 0.10). This outcome was in line with our expectations, but is not significantly different. Impact on the ICER Mean costs of doing nothing were zero. The mean costs of hearing aid fitting were €1,877. The latter consisted of GP visits (€37), ENT visits (€295) and hearing aid(s) (€1545). All hearing aids were digital, and hearing aids were bilaterally fitted in 83% of the respondents. This resulted in mean one-year incremental costs of hearing aid fitting versus doing nothing of €1,877. The mean utility gain of 0.01 (sd 0.13), measured with the EQ-5D UK tariff, resulted in a ratio of €286,866 per QALY, with a 95% confidence interval of inferior (higher costs, lower utility) to €47,082/QALY. There was a 36% probability that hearing aid fitting was both more costly and less effective (inferior). The mean utility gain of 0.003, measured with the EQ-5D Dutch tariff, resulted in an ICER of €647,209 per QALY (95% confidence interval: inferior to €61,934/QALY). There was a 42% probability that hearing aid fitting was inferior. Applying the HUI2 and HUI3, the ICER was €25,337 per QALY (95% confidence interval: €38,012/QALY to €19,356/QALY) and €15,811 per QALY (95% confidence interval: €24,654/QALY to €11,664/QALY) respectively. For both measures there was no probability that hearing aid fitting was inferior. The informal Dutch ceiling ratio of €20,000/QALY [42] implied that hearing aid fitting was only cost-effective when utility was measured with the HUI3. A cost-effectiveness plane with incremental cost and effect pairs for 1,000 bootstrap replications, for all measures, is shown in Fig. 4. Figure 5 shows the cost-effectiveness acceptability curves for all measures. Fig. 4Incremental cost-effectiveness ratio for hearing aid fitting. Incremental costs (euro) are displayed at the vertical line, incremental effects (QALYs) on the horizontal lineFig. 5Cost-effectiveness acceptability curves of EQ-5D UK and Dutch tariff, HUI2 and HUI3 Conclusion and discussion Hearing loss affects the exchange of information and therefore affects a person’s quality of life [1]. Hearing impaired persons can benefit from using a hearing aid, since hearing aid use has proven to be effective, it improves social, emotional and communication functions and reduces depression [3, 4]. The results of this study in a relatively healthy population with hearing complaints provide insight in the differences between two widely used utility measures, the EQ-5D and the HUI system. Regarding practicality, both questionnaires had high completion rates, with the EQ-5D having a higher completion rate than the HUI. With the EQ-5D, differences were detected in utility by age and sex, indicating construct validity. The HUI3 detected differences by age, but not by sex. Differences between clinically distinctive groups were only detected by HUI3. However, the HUI3 did not confirm our expectation that non-applicants would have higher utility values than first-time applicants. An explanation for this may be that non-applicants had more health problems other than their hearing, as reflected in the ambulation and pain dimensions of the HUI3. Overall, HUI2 and HUI3 scores were lower than EQ-5D scores and agreement was moderate to poor. Although these measures intend to assess the same construct, namely health state utility, this result was expected as the instruments differ in their underlying assumptions about what constitutes health state utility. As to responsiveness, only HUI2 and HUI3 measured statistically significant improvement after hearing aid fitting, the EQ-5D UK and Dutch tariff both were not able to capture this effect. Half of the patients (53%) who were fitted with a hearing aid were lost to follow up, either because they had not finished their hearing aid fitting before the end of the study, or because they had their hearing aid fitted at a dispenser not participating in our study. As these patients did not differ from the follow-up group in baseline utility, hearing loss and age, we did not expect this low response rate to influence the results. The HUI2 and HUI3 change scores resulted in smaller ICERs for hearing aid fitting. Although they were only illustrative, the different ICERs found in the present study clearly show that the choice of a utility instrument in the economic evaluation of hearing aid fitting may heavily influence the cost-effectiveness outcome. To calculate the ICER of hearing aid fitting we included the total population of respondents considered for hearing aid fitting, regardless of the type of hearing aid fitting. This makes the ICERs found in the present study representative for hearing aid fitting in general in the Netherlands. However, the one year ICERs calculated in the present study were merely illustrative of the impact of different utility scores on the ICER of hearing aid fitting, as the cost-effectiveness of hearing aid fitting has thoroughly been examined by Joore et al [8]. Our results confirm Barton et al [9, 43], who compared the EQ-5D UK tariff and HUI3 before and after hearing aid fitting in a UK sample. However, the change in utility after hearing aid fitting derived with the HUI3 in the present study was twice the change in utility derived with the HUI3 in the study of Barton et al [9]. Respondents in the latter study had a lower baseline utility score and a higher range in utility change after hearing aid fitting, which suggests a somewhat different study population. Our results confirm previous studies that concluded that the EQ-5D lacks sufficient sensitivity to measure the benefit of hearing aid fitting [6, 7, 44]. The lower utility scores and higher responsiveness of the HUI3 in relation to the EQ-5D were also found in a population with visual impairment [22]. Joore et al [8] calculated that hearing aid fitting costs €15,807 per QALY, using the EQ-5D as utility measure. The results of the present study indicate that using the HUI3 as utility measure probably had resulted in a more favourable ICER for hearing aid fitting. Vuorialho et al [45] recently concluded that counseling of hearing aid users to reduce the number of non-users is highly cost-effective, although they were unable to measure any change in utility. As they used the EQ-5D, it is possible that they would have been able to demonstrate favourable costs per QALY when they would have used the HUI3 to measure change in utility. Three questions arise from the results of the present study: can differences be explained by differences in the measures, are the differences observed between the measures important, and what are the implications of the findings for utility measurement and cost-utility analysis in populations with hearing complaints? First, differences in utility scores can be explained by differences in the descriptive system and the way the utility scoring function is derived. Regarding the descriptive system, the focus of the EQ-5D on physical, mental and social functioning [46] differs from the ‘within-the-skin’ perspective of HUI, which focuses on the underlying level of impairment. However, this does not explain why the EQ-5D does not measure change after hearing aid fitting, as previous studies have found that hearing aid fitting improves social and emotional functioning [3, 4]. Also, the measures differ in the content and number of attributes, items and levels used, and therefore differ both in the number and in the content of possible health states. It has already been suggested not to use the EQ-5D in relatively healthy populations, given the presence of a ceiling effect [15]. The ceiling effect of the EQ-5D found in the present study is likely to contribute to the differences in responsiveness. When, as found in our study, 44% of the respondents report perfect health at baseline, it is unlikely to find a considerable utility gain from any intervention. Furthermore, since the HUI descriptive system pays explicit attention to hearing abilities, it is to be expected that in a relatively healthy population with hearing complaints HUI and EQ-5D utility scores differ, and HUI is more responsive. As the HUI3 also pays explicit attention to visual abilities, this may explain why comparisons of EQ-5D and HUI3 in hearing and vision show similar results [22]. There are also differences in the utility scoring functions. Although in general SG (used for HUI) leads to higher scores than TTO (used for EQ-5D) [5], in the present study the EQ-5D scores were considerably higher than the HUI scores. Although different populations do not necessarily yield different results [5], the population sample in which the preferences are measured may also have impact on the differences. In the present study, differences between utilities derived with the UK and Dutch tariff were observed. This difference may be the result of differences in health valuation between people from the Netherlands and the UK, but may also result from the somewhat questionable representativeness of the population sample used to develop the Dutch tariff [27]. Differences also exist in the type of scoring function. EQ-5D uses an additive system, assuming no interaction for preferences among attributes at all. The HUI uses a multiplicative scoring function, with the effect that the loss of utility associated with a particular dimension is dependent on the level of impairment on other dimensions [5]. For example, Barton et al [43] illustrated that hearing impairment (‘unable to hear at all’) has a greater impact on HRQoL as measured with the HUI3 when one has no other health problems (−0.53), than when one also has moderate to severe pain and is unable to see at all (−0.05). It seems rational that persons find their hearing loss a less important aspect of their health state utility when they experience more comorbidity. The multiplicative scoring function of HUI takes this influence of comorbidity into account and seems to be more suitable for modeling utility scores. Are the differences observed between the measures important? The answer to this question is a clear ‘yes’. The impact of different utility measures on the ICER for hearing aid fitting is of a magnitude that can alter policy decisions and emphasizes that comparisons of QALYs across studies and interventions should be interpreted with caution [12]. The general purpose of a utility measure is to capture the health effects in terms of HRQoL of a policy or program for use in economic analyses. Health economic analyses are a tool to allocate resources in a way that maximizes health (or welfare). In order to use the outcomes of economic analyses for policy decisions, there should however be a clear notion of what should be maximized. The results of the present study show that the potential benefit of an intervention heavily depends on the assumptions of what constitutes health underlying the utility measure that is used. Beyond the apparent necessity of psychometric quality of a utility measure, an important area for future research is whether societies wish, or should, maximize life expectancy corrected for HRQoL from a more functional perspective as in the EQ-5D, or from a ‘within-the-skin’ perspective as in the HUI. Other studies that have thoroughly examined the psychometric differences between utility measures seem to pass over this important question [11, 13, 19, 21, 24]. What are the implications of the findings for utility measurement and cost-utility analysis in populations with hearing complaints? Generally, it has been recommended that the instrument that is most sensitive to the health states in which one is interested should be selected [5, 12, 47]. From clinical experience it is plain that hearing aid use is effective in alleviating hearing loss and does improve health-related quality of life, but the EQ-5D lacks the sensitivity to capture this improvement. In an otherwise healthy population, HUI3 has proven to be more responsive and therefore more appropriate for evaluating HRQoL in a population with hearing complaints, and is therefore the instrument of first choice in this population.

Eur_J_Health_Econ-_-_-1388079

The “Health Benefit Basket” in Denmark

Until 2007, when the new legislation on health care becomes effective, the right to receive free health care services in Denmark, or “health benefits,” are described in a comprehensive set of legislation, including laws, executive orders and legal guidelines. This contribution provides an overview of the current main legislation regulating the Danish “health benefit basket” and describes the regulatory mechanisms for the provision of curative care at Danish hospitals and primary health care offices. Although the services are both financed and planned by the counties, they differ substantially in the way that benefits are regulated. The Danish health care system offers a vast amount of benefits and entitlements to persons living in Denmark. The benefits are more or less specified health services provided at hospitals or in primary and tertiary health care institutions. Entitlements are rights to receive these services free of charge or against a fee covering a part of the costs. The benefits and entitlements altogether form the so-called “health benefit basket” or “basic package of health services” [1]. Although the notion of a benefit basket is not commonly used in Denmark, health benefits are indeed described in the legislation. The following pages provide an overview of the laws regulating health benefits and a description of the regulatory mechanism for the provision of curative care in hospitals and in primary care offices. Curative care services are financed and planned by the 14 counties and the municipal authorities of Copenhagen and Frederiksberg, and there are considerable differences between in- and outpatient services in the way in which benefits are regulated and services are provided and reimbursed. Table 1 provides an overview of the four most important laws regulating benefits in the Danish health sector:Table 1 The Danish “health benefit basket”: overview of main legislationHospital Act (and ancillary legislation)Public Health Insurance Act (and ancillary legislation)Medicines Act (and ancillary legislation)Social Services Act (and ancillary legislation)Other (specific) legislationHC.1.1 Inpatient curative care+–(+)––HC.1.2 Day-cases of curative care+–(+)––HC.1.3 Outpatient care(+)+(+)––HC 1.4 Curative home care++(+)––HC.2.1 Inpatient rehabilitative care+––(+)–HC.2.2 Day cases of rehabilitative care+(+)–(+)–HC 2.3 Outpatient rehabilitative care(+)+–+–HC 2.4 Services of rehabilitative home care(+)(+)–++HC.3.1 Inpatient long-term care+––+–HC.3.2 Day cases of long-term care+––+–HC 4.1 Clinical laboratory +++––HC.4.2 Diagnostic imaging+++––HC.4.3 Patient transport ++–+–HC.4.9 All other miscellaneous services++–(+)–HC.5.1 Pharmaceuticals, etc.–++––HC 5.2 Therapeutic devices etc.(+)(+)–+–HC.6.1 Maternal and child care++–(+)+HC 6.2 School health services––––+HC.6.3 Prevention of communicable diseases–(+)––+HC.6.4 Prevention of noncommunicable diseases–(+)––+HC.6.5 Occupational health services––––+HC.6.9 All other services––––+Legal statusLawLawLawLawLawsDecision makerCounty + physicianCounty + GPCounty, GP (physician)Municipality + countyCounty + municipality + GPPurposeEntitlements, delegation and defining responsibility of countiesEntitlements, reimbursement, copaymentRole of Danish Medicines Agency, process of approval of medicinesEntitlements, delegation and defining responsibility of counties and municipalitiesEntitlements, defining roles and responsibilities of actors etcPositive/negative definition of benefitsPositivePositivePositivePositivePositive Degree of explicitnessa11 (2 and 3)2 (and 3)1 (2 and 3)1, 2, 3If itemized: goods, procedures, indicationsNot itemizedProcedures (indications)(Goods)(Goods, indications)IndicationsUpdatingNoNoNoNoNoCriteria used for defining benefits  Need+++++  Costs–––––  Effectiveness––+––  Cost-effectiveness–––––  Budget+––––  Other–+–––a 1, “All necessary”; 2, areas of care are described; 3, services are linked to specific itemsThe Hospital Act establishes the regulatory framework for all types of inpatient care and day care, patient transport, maternity care, ambulatory and rehabilitative care provided at hospitals (Bekendtgørelse af lov om sygehusvaesenet, LBK 766, 28 March 2003). The law delegates the responsibility for the planning, providing, and financing of hospital services to the county council. A number of executive orders and directives further specify this law.The Public Health Insurance Act (PHIA) and its supplementary legislation provides the regulatory frame for entitlements to outpatient care and to some extent to rehabilitative care, ancillary services for outpatients, maternity and child care, and reimbursement of pharmaceuticals (Lov om offentlig sygesikring, LOV 509, 1 July 1998; og DSK 9926, 20 January 1999; og LOV 467 og 469, 31 May 2000; og LOV 495, 7 June 2001; og LOV 1031, 23 November 2000; og LOV 1045, 23 December 1998; og LOV 1118, 29 December 1999; og LOV 1207, 27 December 2003; og LOV 1431, 22 December 2004). Benefits and services are further specified in a number of executive orders as well as in the used fee schedule.The Medicines Act regulates the population’s access to pharmaceuticals and specifies the process of approval of medicines by the Danish Medicines Agency, marketing authorizations, etc. (Bekendtgørelse af lov om laegemidler, LBK 656, 28 July 1995).The Consolidated Social Services Act provides the regulatory frame for social services (Bekendtgørelse af lov om social service, LBK 708, 29 June 1998). The law delegates the planning, financing and provision responsibility to the counties for the specialized care and to municipalities for the less specialized care. This law covers the long-term inpatient, outpatient and home care. It also describes the provision responsibility for rehabilitative care not following a hospital treatment and also regulates the entitlement to therapeutic devices, nonacute patient transport, and other services. Danish legislation on health care furthermore contains a number of specific laws dealing with preventive services, for example, for children (Forebyggende sundhedsordninger for børn og unge, LOV 438, 14 June 1995), with explicit and specific definitions of benefits. The main decision makers in the Danish health sector at political and administrative levels are the local governing bodies responsible for the service, hence the elected politicians for the county and the municipal councils. At clinical level especially the general practitioner plays a very important gate-keeper role, but hospital physicians are also important in defining medical needs. Most benefits are explicitly but vaguely defined by the law, leaving it to decision makers to specify which services to offer to patients. The Hospital Act states that “the county shall provide hospital treatment free of charge for its citizens” (Sect. 5). The degree of explicitness is described as “all necessary” (1) by all laws; some laws further describe areas of care (2) and some are linked to items (3). The PHIA and the Social Services Act are to some extent more explicit, by linking benefits to items. This contribution is confined to curative health services, and the description of benefits in curative health services in Denmark is divided into three sections. First, the main actors along with their roles and responsibilities are outlined and differences between in- and outpatient services are emphasized. Second, benefits in the hospital sector are described and third, an outline of benefits under the Health Care Reimbursement Scheme (HCRS) for outpatient services is provided. Benefits are described with regard to the legislation and other regulatory mechanisms defining them, how they are classified, and how the fee schedule under the HCRS, a benefit “catalogue,” is structured. Main actors in defining benefits: their roles and responsibilities The Danish Parliament (Folketing) legislates on health care. The Minister for the Interior and Health is responsible for further specifying the law, setting rules in certain areas and initiating reforms and bills. Unless it is explicitly mentioned that the Minister is empowered to specify the law, it can only be changed by the Folketing. Regarding outpatient care the Minister must approve all negotiated agreements between the Health Care Reimbursement Negotiation Committee (HCRNC) and the health personnel trade unions. The National Board of Health (NBH) is a subdivision of the Ministry of the Interior and Health. The NBH is responsible for supervising and providing advice to “health care persons” (the Danish word sundhedspersoner used in the legislation refers to all types of health care professionals) and health administrative authorities, for elaborating and issuing guidelines and reference programs as well as for monitoring and documenting health services. Guidelines from the NBH define the current best practice within specific health care programs or interpret the law. The “health care persons” are for their part obliged to show clinical diligence and act according to “best practice.” According to the Hospital Act, the counties and the Copenhagen Hospital Corporation are responsible for the planning and financing of hospital services (Sect. 1) and for providing free hospital treatment for their citizens (Sect. 5). The providers of hospital services may be county-owned hospitals, private hospitals, or foreign hospitals (Sect. 3). Regarding outpatient care the county reimburses the services provided by private specialists and general practitioners (GPs). County representatives in a subcommittee under the HCRNC decide on the actual supply of services, for example, who is allowed to establish a private practice in which county. A county health plan coordinating health services in the county is to be elaborated every 4 years (Hospital Act, Sect. 11; PHIA Sect. 27e). The plan includes a description of the county’s population health status (premise of the plan), the physical capacity of the county hospitals, including how and where specialties are represented, and the number of beds and employees etc. in each specialty. Furthermore, preventive services, outpatient services, and cooperation between GPs, specialists, dentists, and the hospital sector, cooperation with municipalities, ambulance services, and local emergency services are coordinated and described in the plan. The NBH supervises and comments on the plan. However, the counties are not, in any legal sense, obliged to follow the advice provided by the NBH. Hospital and clinical department managers have considerable freedom within the legal, clinical, and economic framework set by their superiors at national and county level. The heads of department may therefore establish treatment protocols and indications for treatment for different patient categories as well as set up guidelines for prioritizing between patients on waiting lists. In general the Danish legislation is rather vague with regard to the specification of services and very often leaving decisions about benefits to medical judgement. Thus physicians play a certain role in the definition of benefits. Especially, the GP is important, as she/he is responsible for referring patients from general practice to practically all parts of the health sector. Definition of benefits Hospital care Benefits provided in the Danish hospital sector are defined at four levels of decision making: at national level by means of legislation, at county level by planning and supply control, and at hospital and clinical level by local guidelines and clinical decision making. According to the Hospital Act, citizens have the right to free hospital treatment (Sect. 5) at any hospital in the country (Sect. 5b) which they choose. If the waiting time for treatment exceeds 2 months, they have the right to be treated at a private hospital in Denmark or abroad at their county’s expense (Sect. 5 g). The Act includes inpatient care, hospital ambulatory (day) care and in some cases home care for somatic and mental patients. Benefits are explicitly but vaguely formulated by the law. In principle, patients are entitled to any treatment that is clinically indicated, but in practice limitations apply. At national level the Hospital Act and its amendments directly or indirectly specify which services to offer (positive definition of benefits) or not to offer (negative definitions of benefits) to whom and under which circumstances: Positive definition of benefits (examples):For persons suffering from a “life-threatening” disease, defined as patients with certain heart conditions or cancer patients, a maximum national waiting time is defined.The county offers a breast examination (mammography) to women aged 50–69 years every second year (Hospital Act; Sect. 5e). However, this benefit is accompanied by a clause stating that the Minister will decide when this service is to be implemented in the counties. The service is expected to be implemented by 2008. Today it is only offered in three counties.Patients with a clinically defined need for rehabilitation have the right to receive a plan from the hospital describing when and where rehabilitation is to take place at discharge, and who will coordinate between the county and municipal providers (Hospital Act; Bekendtgørelse om udarbejdelse af genoptraeningsplaner ved udskrivning fra sygehus, BEK 1009, 9 December 2003). Negative definition of benefits (examples):Induced abortions after week 12 of pregnancy are only allowed after permission from the Minister of Justice and upon medical or social indication (Bekendtgørelse af lov om svangerskabsafbrydelse og fosterreduktion, LBK 541, 16 June 2004, og BEK 540, 16 June 2004, og VEJ 57, 16 June 2004). Sterilization cannot be provided to persons below the age of 25 years (Bekendtgørelse af Lov om sterilisation og kastration, LBK 661, 12 July 1994, og BEK 1131, 13 December 1996).Unmarried women and women over the age of 45 years are exempt from in vitro fertilization treatment by a physician (specialist; Lov om Kunstig befrugtning i forbindelse med laegelig behandling, diagnostik, forskning mv, LOV 460, 10 June 1997; Bekendtgørelse om kunstig befrugtning, BEK 728, 17 September 1997).Free choice of private hospitals (in the case the waiting time for treatment at county hospitals exceeds the limit of 2 months) does not include transplantation, sterilization, reproductive health services, hearing aid, cosmetic surgery, sex change, psychiatric treatment, alternative treatment, or treatment characterized as research (Bekendtgørelse om amtskommuners betaling for sygehusbehandling ved en anden amtskommunes sygehusvaesen, BEK 594, 23 June 2003). In general, conventional treatments not included in the benefit package are not offered. For patients suffering from life-threatening cancer a referral needs to be approved by a board of specialists, a “Second Opinion Panel” established under the NBH (information available at: http://www.sst.dk/planlaegning_og_behandling/second opinion.aspx, accessed 4 August 2005). If the alternative such as conventional treatment is approved by the board, the expenses are covered by a special government grant. However, nonconventional treatment, such as a spa therapy, zone therapy, and homeopathy are not included in the benefit package [2]. Financial instruments Hospitals are financed by the counties through tax revenue and block grants from the state. Following the annual negotiations between the counties and the government on the level of taxation and size of block grant the government has introduced new financial measures as a means to influence the development of the health sector. Special block grants are earmarked to high priority areas such as heart surgery and cancer treatment. These grants accompany recommendations from the NBH on treatment improvements and expansion of the capacity, as described in the “Heart Plan” and the “Cancer Plan.” Furthermore, with the aim of shortening waiting lists, a provision of activity-based funding for additional production has been made available from 2002 for counties which can document an increase in their activity (Hospital Act; (Cirkulaer om udbetaling af statstilskud i 2004 til øget aktivitet på sygehusområdet CIR 6, 5 January 2004. Bekendtgørelse om økonomiske rammer for frit valg til private specialsygehuse, BEK 627, 20 June 2004.). Financial restrictions are imposed on the counties as their obligation to finance care at certain private hospitals including hospices, sclerosis hospitals, arthritis sanatoria, and brain rehabilitation hospitals is limited to a certain annual amount. Thus patients are either referred to a waiting list or must pay for the treatment themselves at a private clinic. At county level the counties determine the content and costs of hospital activity through the use of detailed budgets, enabling them to specify treatments offered, technologies used, service standards, and capacity available. There is an indirect regulation of benefits available to county inhabitants through the control of the supply. In principle, the county can decide not to offer a certain treatment, for example, in vitro fertilization treatment to women who have already had one child by this means, or an expensive cancer treatment. However, the patient can choose to be treated in another county due to the right of free choice of hospital with the home county having to pay for it, which actually displays certain limitations to the county’s planning freedom. In the hospital sector there are no benefit “catalogues” as such. However, a wide number of classifications are used to document the activity and prices based on diagnosis-related groups (DRG). DRGs are used for the remuneration of hospital services provided in other counties or at private hospitals. Outpatient services: the Health Care Reimbursement Scheme Unlike hospital services, benefits regarding outpatient services are under the HCRS explicitly and specifically detailed by the law and in a fee schedule negotiated on services and prices between the private providers and the HCRNC. The PHIA states that primary health care is available to (almost) everyone with a permanent address in Denmark. All citizens must choose between insurance groups I and II with a trade-off in terms of free choice of the provider and the size of reimbursement. Persons in group I, more than 98% of the population, are entitled to free services from GPs and to partial reimbursement of dentist services, physiotherapy treatment, etc. With a few exceptions public reimbursement of specialist and paramedic services are subject to GP referral. Persons in group II, fewer than 2% of the population, have a wider choice of providers, but are less entitled to reimbursement. Everyone receives a personal card clearly stating the entitlements to the services (PHIA; Bekendtgørelse om valgfri indplacering i sygesikringsgrupper, udstedelse af sygesikringsbevis mv, BEK 198, 21 March 2003; Bekendtgørelse om begraensning i adgang til ydelser efter sygesikringsloven for visse persongrupper, BEK 119, 17 March 1976, og BEK 115 af 21 February 1990). A large number of executive orders and the fee schedule for reimbursement of private providers further specify entitlements and benefits laid down by the PHIA (PHIA; Bekendtgørelse om befordringsgodtgørelse efter sygesikringsloven, BEK 3, 3 January 2001; Bekendtgørelse om tilskud efter sygesikringsloven til fodbehandling, BEK 129, 18 March 2003; Bekendtgørelse om tilskud efter sygesikringsloven til kiropraktisk behandling, BEK 181, 22 March 2004; Bekendtgørelse om tilskud efter sygesikringsloven til fysioterapeutisk behandling, BEK 405, 18 May 2001; Bekendtgørelse om tilskud efter sygesikringsloven til psykologbehandling for saerligt udsatte grupper, BEK 472, 18 June 2002; Bekendtgørelse om tilskud efter sygesikringsloven til sondeernaering og andre ernaeringspraeparater, BEK 531, 18 June 2003; Bekendtgørelse om tilskud efter den offentlige sygesikring til betaling af briller til børn under 16 år, BEK 543, 8 December 1980; Bekendtgørelse af lov om tandpleje, LBK 1261, 15 December 2003, og BEK 1073, 11 December 2003; Bekendtgørelse om tilskud efter sygesikringsloven til tandpleje, BEK 147, 6 March 2004; Bekendtgørelse om adgang til laegehjaelp efter sygesikringsloven, BEK 180, 18 March 2003). Some benefits are defined very explicitly, e.g., a specific maximum amount in Danish crowns to be covered for spectacle lenses and spectacle frames; (BEK 543, 8 Decemberr 1980, see above), while others are defined in more general or vague terms, e.g., proportional reimbursement of negotiated fee for physiotherapy treatment (PHIA; BEK 405, 18 May 2001, see above), or proportional reimbursement of negotiated rates for dental treatment (LBK 1261, 15 December 2003; og BEK 1073, 11 December 2003, see above; BEK 147, 6 March 2004, see above). There are positively defined benefits (e.g., the right to spectacle lenses and frames) and exclusions through negatively defined benefits, e.g., a service excluded from reimbursement from the public travel insurance benefits (Bekendtgørelse om befordringsgodtgørelse efter sygesikringsloven, BEK 3, 3 January 2001). While, for example, GP services, ear nose and throat medicine, psychiatric, and oculist services are free of charge for patients, dentist services are only partly covered by public funds. Some services are regulated by legislation but not included in the benefit basket, unless certain specific criteria are fulfilled (e.g., psychologist counseling, chiropody, and chiropractic services (BEK 129, af 18 March 2003, see above; BEK 181, af 22 March 2004, see above; BEK 472, 18 June 2002, see above). Table 2 provides an overview of the main services under the HCRS, showing the extent and criteria for coverage and whether the services are subject to a referral from the GP.Table 2 Benefits regarding outpatient services according to the Health Care Reimbursement Scheme. Services fully, partially or not covered by public fundsType of service/providerCoverage by public fundsCoverage for special groupsReferral from GP required (Group I insured)General practiceFull––Dentist servicesPartialFull coverage for children at school age, disabled, low income elderlyNoTube feeding/nutritional preparationsFull–YesPhysiotherapyPartialFull coverage for patients with specific diagnoses (muscular-skeletal) YesPsychiatristFull–YesPsychologistNone/(partial)Partial coverage for persons exposed to traumatic incidentsYesChiropodistPartialPartial coverage for diabetics and low-income elderlyYesChiropractorPartial–YesOpthalmologistsFull–NoEar-nose-throat physiciansFull–NoDieticiansNone–NoOther specialistsFull–Yes “Other specialists” whose services are fully part of the benefit basket upon GP referral include: anesthetists, specialists in dermatology and venereal diseases, specialists in diagnostic radiology, gynecologists, obstetricians, internists, surgeons, biochemists, neurologists, orthopedics, specialists in pathological anatomy or plastic surgery, pediatricians, rheumatologists, and specialists in tropical medicine or laboratory examinations. The fee schedule Whereas the laws describe mainly the extent to which outpatient services are included in the benefit basket, for example, the proportion of public funding and the criteria for coverage, the fee schedule negotiated between the HCRNC and the health personnel trade unions specifies the services for which the providers are paid. Since the schedule lists services which are reimbursed from public funds and are therefore provided free of charge to the patients, the fee schedule may—in an indirect way—be regarded as a benefit catalogue (Sygesikringens Forhandlingsudvalg, takstkatalog, April 2005). Items in the fee schedule are negotiated every 3 years while the tariffs are revised twice a year. Health technology assessments, cost-effectiveness analyses, and clinical studies are adduced by the professional committees and included in the negotiation but they are not mandatory. Criteria for the decisions concerning the inclusion or the exclusion of an item are neither systematic nor public. An example of an argument over the inclusion of services regards progress in treatment methods (substitution of outdated services with more up-to-date procedures). There is usually a reimbursement rate when benefits such as those for contraceptive advice from the GP are specifically stated by the law (A. Bonne, HCRNC, personal communication). The catalogue is divided into specialties with a subdivision into type of services offered (items). Items may be directly linked to the legislation that specifies the benefits, for example, antenatal care consultation and first child health check-up at the age of 5 weeks (general practice). They are also directly linked to the logic applied by the legislation, for example, following certain age groups, certain timing (week 12, first and second consultations, etc.), or specific agreement, such as on working hours or transport fees for the professional group involved. When the legislation does not specify the benefits, items are either linked to consultation type (visits), goods (e.g., laboratory tests) procedures (e.g., spirometry) or even indications (e.g., podiatry for diabetics) in some rare cases. There is generally no specification of the technology to be used with a few exceptions. Concluding remarks The Danish benefit basket for curative services is regulated directly by law and influenced indirectly by more “soft” regulatory mechanisms. In the hospital sector which is the main part of the health sector in terms of expenditure, the law delegates the responsibility for providing, financing, organizing, and planning services to the 14 counties. Thus there may be regional variations in the services offered. Benefits are explicitly defined but not specified by the law. In principle, all clinically indicated hospital services are included in the benefit basket. However, central initiatives, such as legislation on the patient’s free choice of hospital, waiting time guarantees for some patient groups and financial constraints influence the counties’ decisions and narrow the room for regional variation. Regarding outpatient services the picture differs slightly as benefits in the HCRS are regulated at central level through explicit and specific legislation while being further specified in agreements about items and tariffs between county and provider representatives. In 2007 the Danish health sector will be subject to a structural reform which will significantly influence the way in which health services are financed and organized. The current political layer of 14 counties will be abolished and be replaced by five regions responsible for providing hospital services but without the taxation rights that they have today. The municipalities will merge into fewer units, assume financing responsibilities for more health services than today and will purchase hospital services from the regions. The reform’s immediate impact on health benefits is hard to predict, but there will be some modification as to legislation at least. A law, “the Health Act” (Sundhedsloven, LOV nr. 546, 24 June 2005), recently passed by the Folketing, collects the previous vast amount of legislation on health care into a single document. Among other things the goals of Danish health care services, for example, that citizens should have easy and equal access to health care, have now been made explicit. The rights to health care services do not change fundamentally, nor do decision makers and their roles and responsibilities. The Act contains many clauses empowering the Minister for the Interior and Health to set specific rules, thus giving the Minister much more decision-making power than today. However, the Health Act provides a regulatory framework with explicitly mentioned rights to health services not being different from what is outlined here. The specification of benefits will still take place at the lower levels of decision making, by specifications of the law by the Minister, by the National Board of Health, in agreements between regions and municipalities, through a payment schedule, or at clinical level.

J_Comput_Aided_Mol_Des-4-1-2270916

How to do an evaluation: pitfalls and traps

The recent literature is replete with papers evaluating computational tools (often those operating on 3D structures) for their performance in a certain set of tasks. Most commonly these papers compare a number of docking tools for their performance in cognate re-docking (pose prediction) and/or virtual screening. Related papers have been published on ligand-based tools: pose prediction by conformer generators and virtual screening using a variety of ligand-based approaches. The reliability of these comparisons is critically affected by a number of factors usually ignored by the authors, including bias in the datasets used in virtual screening, the metrics used to assess performance in virtual screening and pose prediction and errors in crystal structures used. Introduction Based on the large number of papers recently published, it has become obvious that a large proportion of the computational chemistry community, both in academia and in industry, is very interested in evaluating and comparing software for a number of different purposes. A large number of publications have appeared over the last 5 years or so that are focused on the evaluation of docking tools for pose prediction [1], virtual screening [2] and affinity prediction [3]. There have also been a number of recent publications examining the performance of ligand-based tools in similar tasks. The ligand-based tools have also been evaluated in the areas of pose reproduction (by conformer generators [4–6]), virtual screening [7] and affinity prediction [8]. In the following sections some issues with studies on pose prediction and virtual screening will be discussed. Pose prediction A common method of evaluating a docking program is to gauge its performance in cognate re-docking or self-docking. In this process a ligand is extracted from a co-crystal structure with its target protein and the program is challenged to pose the ligand as closely as possible to its experimentally identified structure. It may be argued that cognate re-docking is not a task commonly faced in the normal use of docking tools, since cross-docking (docking of a ligand into a structure with which it was not crystallised) is the actual application of a docking tool [9]. However, the exercise remains popular, doubtless in part due to the relative ease of execution of a self-docking study and partly for comparison with previous studies. As has been pointed out previously [10], comparing docking programs for their ability to predict the bioactive pose of a ligand is difficult for a number of reasons, some of which are obvious while others are subtler. However, the operational difficulties in comparing programs in a robust way should not cause other sources of error to be ignored. It is a truism that it is meaningless to compute a property with greater precision than the accuracy of the experiment that measures that property. Unfortunately, as will be seen in subsequent sections, this is often ignored by authors of papers in the area of pose prediction. When comparing tools for pose prediction, the heavy atom root mean square deviation (RMSD) between the computed and experimental poses is the de facto standard. A regrettably common method to compare docking tools for pose prediction success––illustrated, for example, in the papers presenting results from the MolDock program [11] or the Glide XP evaluation [8]––is to compare the average RMSDs across a set of structures. The use of the average RMSD admits of many possible problems of interpretation, not least of which is the biasing of the average by a few very large or very small numbers. In the case of the MolDock results, the mean RMSD across a set of structures was used to suggest that the performance of MolDock was comparable to that of GLIDE and superior to that of Surflex (see Table 1). However, the use of the median RMSD, which is far less biased by a few extrema, suggests a different conclusion—that MolDock is in fact somewhat superior to Surflex, but not as good as GLIDE. For a similar analysis on the drawbacks of using average RMSD as a comparator, see Cole et al. [10]. A number that is noticeably absent from this, and all other comparisons of docking tools using RMSD, is an error bar on the average RMSD (which can be calculated by bootstrapping). Without such an error bar it is impossible to assert that any tool compared in this study is actually better than any other. Table 1Comparison of RMSD results from a set of docking enginesMolDockGLIDESurflexMean1.381.381.86SD1.491.742.02Median0.920.691.10 A separate issue with these two experiments involves contamination of the dataset used to evaluate the performance of the tools. In the case of the Glide results the RMSD’s given are not to the deposited crystallographic pose but rather to one that results from a pre-processing step, as noted by the authors of the MolDock paper [11]. As such this is not an “apples to apples” comparison, since the Glide pre-processing step optimizes protein and ligand coordinates using the force-field component of the Glide scoring function, which necessarily introduces bias in the structure. In the MolDock case the authors have essentially trained the MolDock fitness function on the 77 complexes that they use to evaluate its performance. As such the reported results give no indication of the likelihood of success in predicting a pose for a system upon which MolDock was not trained. The RMSD between two poses is a geometric measure, comparing the atomic positions between the experimental structure and the docked or predicted structure. Other metrics based on comparing the geometry of the experimental and the computed pose have been developed, such as relative displacement error (RDE) [12]. These, and all other atom-based metrics, suffer from the drawbacks pointed out by Cole et al. [10]. A more serious problem for metrics like RMSD and RDE is that they attempt to indicate the quality of reproduction of a model for the data, not the crystallographic data itself, i.e., the electron density. This disconnection between RMSD as a metric of quality for pose prediction and the original crystallographic data has been a cause for concern. Some attempts have been made to arrive at metrics that better reflect the reproduction of the actual crystallographic data, in particular real-space refinement (RSR) [13]. Unfortunately, RSR has not been widely used, possibly because it is more difficult to calculate than atom-based metrics like RMSD or RDE. Other metrics that are not based purely on atom position, such as interaction-based accuracy classification (IBAC) [14], try to reflect the ultimate use of the predicted pose, i.e., determining the nature of the interactions that the ligand makes with the protein. While the IBAC approach has value in that it assesses a computed pose by its interactions with the protein, it is not amenable to automation; it is therefore tedious to assemble sufficient data to make statistically robust comparisons between tools based on IBAC. However, the problem of choosing which metric to use to compare pose prediction studies is dwarfed by the difficulty in choosing a dataset of protein–ligand co-complexes upon which to perform the comparison. A widespread tendency in conformer reproduction and pose prediction studies is to ignore even the possibility of error in the crystal structures that are being reproduced. Crystal structures are often treated as perfect, infinitely precise and accurate representations of the atomic details of a protein–ligand complex. There are a number of reasons why this is not so; a few will be discussed in the following paragraphs. Crystal structures are models This statement makes up the warp and weft of crystallography, yet in the transition of crystallographic data from crystallographer to computational chemist the distinction between the actual data and a model for that data is often lost. The actual data in crystallography is, of course, diffraction data leading to electron density. The atom positions that make up a crystal structure are a model that attempts to explain this data in as complete a fashion as possible. This is illustrated in Fig. 1. Fig. 1Fitting atoms into electron density produces a crystallographic model The process of fitting or modeling the atoms (of a ligand or of a protein) into the electron density is not always straightforward, and the problems are often more serious when the fitting of a ligand into density is being carried out. The sources of these problems include:Incomplete or fragmentary density;The electron density not defining the positions of all atoms unambiguously;Poor structural parameters are used for the fitting process, which can give inappropriate conformations (particularly of ligands);Errors by the users, arising from careless treatment of the data or lack of expertise with small molecules. A variety of metrics can be calculated for a given crystal structure that attempts to give an indication of the quality of the model. While no single number can encompass the quality of a structure, some of the metrics of quality are more useful than others. The most commonly used is the nominal resolution. This is a measure only of the quantity (of data collected) and not of the quality of the data nor, most especially, of the quality of the model fitted to that data. (For a discussion on the problems with resolution, see Ref. [15]) It is therefore unfortunate that nominal resolution is often the only criterion used to select a protein structure, and that its meaning is often misconstrued. For example, in a paper [16] by Nissink et al. the following incorrect statement is made: “The resolution of a protein structure is directly related to the accuracy of the data.” Not only is this statement not entirely true, it also confuses what the nominal resolution means (quantity of data collected) with the accuracy of a model, which are two very different things. A metric that does attempt to assess the quality of the fit of the crystallographic model to the source data is the so-called Rfree, introduced by Brunger [17]. Rfree is an indication of how well an atomic model explains a small percentage of the density data that was omitted during the fitting process, and is thus an unbiased metric that can be reliably used to distinguish a well fitted model from a poorly fitted model. Unfortunately, Rfree is infrequently used as a metric of quality when selecting crystal structures for docking studies or other purposes (see Ref. [18] for the use of Rfree as a criterion for selection of structures). More frequent use of Rfree as one of a set of metrics for selection of crystal structures could help to avoid the selection of poorly fitted models for pose reproduction. Crystal structures have unavoidable imprecision As already discussed, nominal resolution is not a metric of quality for a structure and although Rfree indicates the quality of the fit of the model to the original data it can provide no estimate of the uncertainty in the atomic positions within that model. It is often assumed that the experimental uncertainty in the atomic positions in a crystallographic model can be estimated by use of the isotropic B-factor, which is supposedly a measure of the thermally driven fluctuation in atomic positions. However B-factors are a refined parameter and so they cannot be compared between structures without detailed knowledge of the restraints used [19]. It is therefore impermissible to use low B-factors as a sure indication of low positional uncertainty. Nor is it possible that there exists a uniform cut-off for B-factors that indicates low positional uncertainty in all structures, though such an assumption is frequently made [20]. Another potential problem with B-factors is that they can be over-refined in the pursuit of better quality metrics for a structure. An example is given in Fig. 2, which shows the ligand (leucinol) from the structure 5ER1 from the RCSB [21] annotated with its B-factors. The B-factors for every atom in the ligand are <1. In contrast, B-factors even for well-located atoms in high quality models, are generally not <5 [22]. In this instance the crystallographer has over-refined the B-factors of the ligand atoms in order to improve the quality of the overall model. In this model the B-factors are unphysically low only as a consequence of a pathology of the refinement process. Accordingly, B-factors as an indication of local mobility in a protein structure should be treated with caution, as they are not simply an experimentally derived quantity but are free parameters in the refinement process. Fig. 2B-factors for the ligand in the 5ER1 crystal structure An alternative to using the average B-factor for estimating the uncertainty of the position of atoms an atom in a structure is the average coordinate precision (or diffraction-component precision index, DPI). The DPI expresses the average precision for the atomic coordinates in a protein structure [23] and as such can be used as a measure of the experimental precision of the atomic positions in that structure. The original formulation of DPI was very complex and has recently been recast in a more easily calculable way by Blow [24]. The use of DPI as a metric of quality for crystal structures used for docking studies was introduced to the computational community by Goto et al. [18], and their formula is given in Eq. 11. In this treatment the standard error of position, σ(r), is related to the number of atoms in the unit cell, Natoms, the volume of the unit cell, Va, the number of crystallographic observations, nobs and the Rfree. It should be noted that the formula presented by Goto et al. is not precisely the same formula that Blow derives in his paper. In Eq. 1 the prefactor is given as 2.2, while in the original work by Blow the prefactor is given as 1.28. This is because Blow is calculating coordinate error for a particular axis, σ(x, y, or z), while Goto et al. are calculating the error in the distance, σ(r), giving rise to the √3 difference between the two prefactors. As such it is appropriate to consider the error in the coordinates, σ(x, y, z) as a measure of the uncertainty in the atomic positions, and the error in inter-atomic distances, σ(r), when comparing a computed and an experimental structure. The resolution of a crystallographic model, as has been mentioned, is often used to select protein structures for pose prediction by docking or conformer generation studies, on the assumption that resolution imparts some information on the quality and precision of the model. The DPI is a much more direct estimate of the reliability of crystallographic models when it comes to comparing experimental and computed atom positions (as is done in conformer reproduction or pose prediction). It is therefore of interest to compare the nominal resolution for a large number of “good quality” crystal structures with the DPI (σ(r)) for the same structures. A good dataset for this comparison is provided by the extensive investigations performed by Kirchmair et al. [4]. Here 776 co-crystal structures were used to provide experimental ligand structures that were then compared to sets of computed conformations from conformer generation applications. For 556 of these crystal structures there exists sufficient data to allow the DPI to be calculated and the relationship between the nominal resolution for these structures and their DPI is shown in Fig. 3. It is obvious from Fig. 3 that the statement by Kirchmair that “0.5 Å approximately represents the accuracy of protein X-ray crystallography” is not supported by the actual properties of the crystal structures they studied. In fact, in those cases where the DPI can be calculated, almost 56% of the structures from their paper have DPIs > 0.5 Å.Fig. 3Coordinate error for 556 structures from the paper by Kirchmair et al. [4] Figure 3 illustrates a number of other interesting points. While the expectation that greater coordinate precision will arise from structures with better nominal resolution is generally borne out by the data, there are many exceptions. Table 2 shows some examples of structures where the nominal resolution gives an unexpected estimation of coordinate precision. In the top half of the table are structures with good nominal resolution but unexpectedly high DPI, while the lower half of the table shows some structures with low nominal resolution and either unexpectedly low or unexpectedly high DPIs. Accordingly, simply using nominal resolution as a metric of quality for structures to be used in a pose prediction or conformer generation study is insufficient to guarantee that structures of appropriate quality will be used.Table 2Resolution and DPI for selected structures from the Kirchmair datasetPDB codeResolution (Å)DPI (Å)1FC71.380.691FDO1.380.601JJT1.81.371JJE1.81.251CIB2.55.541ILH2.760.141C8M2.80.181QJX2.80.25 With the DPI for a structure in hand one can set a lower limit on the precision with which a computed conformation can reproduce an experimental one—the RMSD between the two conformations cannot be less than the DPI for the experimental structure. It can be seen that over half (55.6%) of the structures in this set has DPIs > 0.5 Å, while Kirchmair et al. report pose reproduction statistics both at <0.1 Å and at <0.5 Å RMSD. Since in over half of the structures in this dataset the DPI is >0.5 Å, Kirchmair et al. report pose reproductions at <0.5 Å RMSD that are more precise than the accuracy of the source data allows. This analysis is made on the conservative assumption that the error in the atomic positions in the computed pose is zero. In the Goto et al. paper [18] the assumption is made that the errors in the computed pose are the same as for the experimental pose. In this analysis a computed and experimental pose must be different by an RMSD of √2 × DPI for the difference to be significant, which would mean that an even higher proportion of the poses in the Kirchmair set have been reproduced with a greater precision than the experimental accuracy. The same tendency to reproduce experimental data with a precision greater than the experimental accuracy is seen frequently in pose prediction experiments with docking engines. In Fig. 4 nominal resolution is plotted against the DPI for crystal structures from two well known docking validation sets, those for GOLD [25] and GLIDE [3]. The two graphs are plotted on the same scale to allow direct comparison. On each graph is also plotted an estimate of the theoretical lower limit for the atomic precision (the pink line). A formula describing the relationship between nominal resolution and DPI is given in Eq. 22 below, based on a derivation by Blow [24].Fig. 4The nominal resolution versus the coordinate error for a subset of the Gold (structures with resolution <2.5 Å) and the Glide data sets The variables found in Eq. 2 are as follows: s is the percent solvent present in the crystal, Vm is the asymmetric unit volume to molecular weight ratio, C is the completeness of the data, and dmin is the nominal resolution of the structure. The ideal lines shown in Fig. 4 were calculated using Eq. 2 and assuming an s of 0.0, a Vm of 2.5, a C of 100%, and that Rfree is equal to the resolution/10. Inspection of Fig. 4 shows that the GOLD set contains one structure (1YEE) whose calculated DPI lies below the theoretical lower limit. This is probably due to a mistranscription error in the PDB file. The PDB file for the 1YEE structure gives the number of reflections as 77209, while the number of unique reflections is 21342—a disparity that cannot be reconciled by reference to the data redundancy for this structure. Accordingly the calculated DPI for 1YEE is too low because the reported number of reflections is erroneously high. In the GLIDE set all structures have DPIs higher than our predicted theoretical minimum. In fact, in this set 52% of the structures have a DPI > 0.5 Å, and in 31% of the cases the reported RMSD for redocking is less than the DPI (so that the prediction is more precise than experimental accuracy allows). This is only one example of a publication in which protein structures are assumed to be free of uncertainty in their atomic positions; the literature abounds with others. The combination of Rfree and the DPI for a structure can give a good overall picture of the quality of a model and the reliability of the atom positions within that model. In spite of the availability of these and other measures of quality, there are a number of test sets of protein co-crystal structures used for evaluating docking engines that appear to have been selected by other criteria that do not relate in any way to their quality [26]. Accordingly, the results from these studies should be treated with some caution. Crystal structures have avoidable errors The Rfree and the DPI are global measures of quality; other, local, measures of quality are also useful. For example, if a small portion of the atoms in a structure have been poorly fitted to the density, this will not be revealed by any global measure of fit. Only a local measure would reveal the error. A relatively common problem in co-crystal structures in the PDB is poor fitting of small molecules to the density, giving unrealistic ligand structures. These poorly fitted ligand structures are then used as a “standard” to judge the quality of a docking program’s or conformer generator’s performance. While the poor quality of some ligand structures in PDB models has been known for some time [27], these reports have been anecdotal and few systematic attempts to avoid such poorly solved structures (other than visual inspection) appear to have been undertaken. If appropriate attention is not paid to selecting good quality ligand structures, then a dataset could be constructed that contains ligand structures that have significant errors. For pose prediction studies, poorly solved ligand structures in the dataset must be avoided. It is senseless to try to computationally reproduce an “experimental” ligand structure that has been incorrectly fitted to the electron density. Examples of some possible errors in ligand structures that should result in structures that are not computationally reproducible are shown in Figs. 5 and 6.Fig. 5Ligand conformation from 1A8T structure. The conformation has two serious atom–atom clashesFig. 6Ligand conformation from the 1A4K structure. The cis-amide group is an error of fitting In the case of the 1A8T structure (Fig. 5), the deposited ligand coordinates contain two serious atom–atom clashes that give the resulting conformation very high energy (28.4 kcal/mol above a refitted structure by the MMFF94 forcefield [28]). The deposited coordinates are clearly in error in this case and no docking engine or conformer generator should be expected to reproduce such a structure. The error shown in Fig. 6 for the 1A4k structure is of a more subtle nature. Here the crystallographer has fitted a highly strained cis-amide into the electron density with no compelling reason from the electron density to do so. The amide group is packing against a tyrosine residue from a symmetry mate in the unit cell, and makes no polar interaction with it. The corresponding trans-amide, which fits the experimental density just as well, is 15.5 kcal/mol lower in energy (using the MMFF94 force-field). Also this trans amide is able to make a hydrogen bond (with a backbone carbonyl group) that is not available to the cis conformation. Once again, this model should not be reproduced by a docking engine or conformer generator, as it is obviously not the correct solution. One of the only studies that bears on the issue of ligand strain was published by Perola and Charifson in 2004 [29]. The authors examined ligand strain in a number of public structures from the RCSB database and proprietary structures from Vertex internal collection and found that 10% of the ligand structures examined had high strain (10 kcal/mol or greater above the global minimum). The two examples discussed above, 1A8T and 1A4K, are both part of the Perola dataset. Clearly these two ligand structures have high strain energies due to errors in fitting and not due to a fundamental property of the ligand’s conformation in complex with the protein. Close examination of the rest of the models for the ligands (where structure factors are available) showed that a number of them are incorrect, and re-solving them provided lower energy structures in about 85% of the cases [30]. Recent re-examination of the Perola dataset by the Snyder group has provided further insights into the large strain energies originally reported [31]. It is therefore most likely that the vast majority of ligand conformations in complex with a protein show strain energies less than 6 kcal/mol, unless the ligand is rather large [32]. Those ligand conformations with strain energies higher than 10 kcal/mol are almost certainly incorrect. As such low ligand strain should be among the criteria for selection of structures for pose prediction, as structures with high strain are very likely to arise from errors in the fitting process. It is clear from the foregoing discussion that, while crystal structures are an invaluable source of information on protein–ligand binding, these structures are not without many sources of confounding errors. These errors, those inherent to the data and the process of fitting, as well as those introduced by human error or the insufficiencies of the fitting program, should be borne in mind before using crystal structure data. The selection of a reliable set of structures for pose prediction is a therefore not a trivial task. An excellent study considering these issues can be found in Hartshorn et al. [33]. Virtual screening Virtual screening can be defined as any method that ranks a set of compounds by some score. Successful virtual screening relies on having a scoring method that assigns good scores to interesting molecules (usually defined as active against a target protein of interest) and worse scores to uninteresting (inactive) molecules. Accordingly a successful virtual screen will provide, from the top of this ranked list, a set of compounds for experimental screening that is highly enriched in active molecules. This topic has been of great interest both in academia and in the pharmaceutical industry in recent years, and a large number of publications have appeared on the subject. While a few of the publications have investigated virtual screening conducted prospectively [34–36], the vast majority have been concentrated in the area of retrospective virtual screening. In the rest of this article we shall concern ourselves solely with the retrospective experiments. The goal of such experiments is often to identify an application that performs well on a given target, or across a wide range of targets, with a view to utilizing this application in prospective virtual screens. There are a number of approaches to quantitating the success of a particular tool for virtual screening. The most often used, and simplest to calculate, is enrichment at a given percentage of the database screened. Enrichment (EF) is defined according to Eq. 33 (Hitssampledx% = number of hits found at x% of the database screened, Nsampledx% = number of compounds screened at x% of the database, Hitstotal = number of actives in entire database, Ntotal = number of compounds in entire database). Enrichment appears to measure the quantity of most interest to those performing virtual screening: the ability of a tool to place a large proportion of the active compounds at the top of the ranked list. Enrichment is also simple to calculate and understand, so it seems the ideal metric with which to compare tools for their virtual screening performance. However, enrichment suffers from a number of significant drawbacks, especially when comparing results between studies or using enrichment to predict future performance:It is dependent on the structure of the dataset, in that datasets with larger proportions of actives will have a narrower range of possible enrichments.It penalizes ranking one active compound above another.It exhibits pernicious behaviour at the cut-off at which the enrichment is calculated.It gives no weight to where in the ranked list a known active compound appears. Thus to calculate enrichment at 1% in a virtual screen of 10,000 compounds, the number of actives (N) in the top ranked 100 compounds is needed. However the enrichment at 1% is the same whether the N active compounds are ranked at the very top of the list or at the very bottom of the top ranked 100.It is difficult to calculate analytically errors in enrichment, and there is no available literature for such a calculation. With regard to point (i), experiments performed on different datasets cannot be compared when using enrichment, as the dynamic range of enrichment will be different for different datasets, but there are several cases where this has been done [37]. Early enrichment can also suggest overly impressive performance for tools that rank a small fraction of the active molecules very early in the list, but fail to give good ranks to the majority of the actives [37]. It has been shown by Seifert [38] that enrichment does not detect significant pathologies in the ranking function. Other metrics have been developed by some groups specifically to address some of these problems: RIE by Merck [39], cumulative probability by Molsoft [40] and average number of outranking decoys by Schrodinger [3]. These metrics have, historically, only been used by the groups that invented them and are thus not useful for comparison to studies conducted by other groups. This lack of direct comparability across studies is compounded by the fact that these metrics cannot be converted into a more commonly used metric that could be used to compare results. Also, as with enrichment, it is unknown how to estimate analytically errors in any of these metrics. A metric used to determine success in detecting a signal in a background of noise is the receiver operator characteristic (ROC). The ROC curve is derived by plotting noise (fraction of false positives) on the x-axis versus signal (fraction of true positives) on the y-axis. The area under the ROC curve (AUC) is a widely used measure in a variety of fields including medical statistics, criminology and bioinformatics [41, 42]. When applied to virtual screening the ROC illustrates success in ranking actives (signal) above decoys (noise). The AUC for the ROC curve shows performance of a given tool when screening across the entire database is examined, not just at fixed, early points in the screen as enrichment does. The theoretically perfect performance of a virtual screening application gives the maximum area under a ROC curve (1.0), while random performance of a tool gives an AUC of 0.5. Areas under the curve of less than 0.5 imply a systematic ranking of decoys higher than known actives. For recent applications of the ROC curve in virtual screening, see [2, 43]. The AUC for the ROC curve is also known as the ‘discrimination’. Discrimination is defined as the fraction of occurrences that a randomly chosen true positive (active) is given a better score than a randomly chosen true negative (decoy). This number then allows prediction of the likely effectiveness of a tool in experiments that have not yet been conducted. This predictive ability is not provided by metrics such as enrichment, cumulative probability and average number of outranking decoys, because while the ROC describes a property of the application studied, the other metrics essentially describe a property of the experiment. The AUC assesses virtual screening performance across the entire database and many practitioners of virtual screening are, rightly, most concerned about early performance of the tools they use. This is one reason why enrichment is so commonly used to measure success. The metric of early performance based on the ROC curve is the true positive rate at fixed false positive rates. The true positive rate at a false positive rate of, for example, 1% is a much more robust measure than the enrichment at 1% and provides similar information about the early performance of a tool. The AUC also offers the advantage that a statistically robust estimate of its errors can be estimated analytically from the AUC itself, using a method developed by Hanley [41]. This is not a property possessed by enrichment, average number of outranking decoys etc. For these metrics errors can only be estimated (by bootstrapping or other approaches) from the raw data, which is rarely provided. The error in the AUC, as with other metrics, is reduced by increasing the number of positives (active compounds) and by increasing the number of negatives (inactives). The Hanley treatment shows that the error in the AUC is most significantly reduced by increasing the number of actives, while increasing the number of decoys has a much smaller impact on the error. Therefore virtual screening datasets with high proportions of active compounds will provide results with lower error bars for the AUC. With the error for an AUC available, meaningful comparisons can be made between two or more different tools. However the other metrics used for virtual screening mentioned above do not allow an analytical estimation of their errors. Comparisons designed to determine which tool is superior for a given purpose, that are based on metrics assumed to be free of errors, are fraught with difficulty. While the choice of metric may affect the relative ranking of tools compared on the same datasets, the composition of those datasets has a very profound effect on the results generated. Until very recently it has been common practice to assemble a dataset for virtual screening by seeding a set of active molecules against a target of interest into a background of compounds (decoys) chosen essentially at random. These decoys compounds were often drawn from public sources such as vendor catalogues. An example of this approach is the seminal paper on virtual screening by docking from the Rognan lab where decoy compounds were selected at random from the publicly available compounds [44]. This set of decoys, or a subset thereof, has been used extensively since its publication in 2000 [2 and references therein] so that this same set of decoy compounds has been used in more than 45 different published virtual screening experiments on a wide variety of target systems. Given the huge variety in the types of active molecules camouflaged in this same set of decoys it seems intuitively obvious that in some cases the active compounds for a given target will be very easily discriminated from these decoys. As such a number of the virtual screening experiments performed using this set of decoy compounds have given good results purely due to differences in simple properties between the actives and the decoys (vide infra). In the Rognan dataset there were 10 active compounds for each of the targets studied and 990 randomly selected decoys. There are two problems arising from constructing a dataset in this way. The first is that the small number of active compounds means that the errors can be very high (vide supra). The second is that trivial property differences between the active compounds and the decoys can result in undeservedly good performance. The first issue, low prevalence, is still widespread in retrospective virtual screening studies. Although low prevalence reduces the reliability of the results, many virtual screening experiments are still conducted using very low numbers of active compounds. Reasons for this could include a desire to mimic “real” HTS experiments, where hit rates are often on the order of 0.01–0.1% [45] or to allow enrichment, the most commonly used metric for success in these studies, the maximal dynamic range. Tribelleau et al. [43] show that the dynamic range of enrichment, the difference between random and maximal performance, decreases as the proportion of active molecules in a dataset rises. For a recent example of a retrospective virtual screening study with deliberately large numbers of actives, providing high statistical power and small error bars, see Ref. [37]. The second issue, systematic differences in simple properties between decoys and actives in retrospective virtual screening experiments, is much more serious. As has been pointed out in a number of publications, scoring functions in docking programs, which are almost always additive, are sensitive to molecular size or heavy atom count, the number of hydrogen bonds that the molecule can make etc. Accordingly, systematic differences in these simple molecular properties between actives and decoys will cause systematic differences in ranking. For example, active compounds with higher average heavy atom counts will tend to rank better than the decoys when scored by a function that is sensitive to heavy atom count. For a fuller discussion of this issue see Verdonk et al. [46]. An example of the influence of the selection methods for decoys on virtual screening performance is shown in Fig. 7. In this figure, two retrospective virtual screening studies against CDK-2 using the docking tool FRED [47] are compared using ROC curves. In both cases the same actives were docked against the same co-crystal structure, while different decoy compounds were used. In one case (the green line), the decoys were chosen at random from the Maybridge compound collection [48], in the other (the red line), decoys were chosen to match the properties of the actives based on simple 1D properties. The striking difference, especially in early performance, is obvious. Clearly the performance of FRED is heavily affected by the nature of the decoy compounds, and to obtain a predictive indication of the utility of FRED in virtual screening decoy sets similar to those giving the red line should be used. It is worthy of note that the AUC’s for the two experiments shown are different by more than their respective 95% confidence intervals, so that this difference is indeed statistically significant. Fig. 7Effect of decoy selection method on virtual screening by docking Another example of how much of the signal that separates active from inactive compounds arises from systematic differences between their properties is shown in Fig. 8. In these eight examples, drawn from the Surflex-Dock dataset [2], the performance of three 3D virtual screening methods––Surflex-Dock, ROCS [49] and FRED––is compared to a simple 1D method. In this 1D approach [46, 50], compounds are ranked by distance in an Euclidean property space to the centre of the space defined by the active compounds. The Euclidean space is defined by five simple molecular properties: number of donors, number of acceptors, number of rotatable bonds, XlogP and 0.01 × molecular weight. This concept of ranking compounds by distance in a high dimensional Euclidean property space has recently been published as a virtual screening method, known as DACCS [51]. Fig. 8AUCs for various virtual screening methods on part of the Surflex-Dock validation set It is clear from Fig. 8 that in four of the eight cases (OPPA, HIV-PR, TK and PARP) the active compounds are very dissimilar from the background set, as the 1D ranking method gives very good virtual screening performance. In a fifth case (TS) the performance of 1D method is as good as any of the 3D methods, although none of the tools perform particularly well. Accordingly, judging virtual screening performance for any tool using such datasets is unlikely to be productive as most of the “signal” separating actives from decoys lies purely in differences in simple molecular properties. A further confounding issue with these datasets is that a large number of the active compounds in these sets are close structural analogues of one another. For ligand-based methods this high structural similarity amongst the actives can cause the actives to be very easily discriminated from the decoys, while a structure-based method may have more difficulty. Accordingly, while property bias is an important consideration in constructing decoy sets, analogue bias should be carefully considered when selecting sets of active compounds. A recent effort to avoid some pathologies arising from poorly selected decoy sets has come from the work of Huang et al. [52] with the Database of Useful Decoys, or DUD. In this work decoys were selected to match the same simple molecular properties of the active compounds using a similar approach to that mentioned earlier, so that the decoys are not trivially separable from the actives. DUD represents a very wide-ranging dataset (actives against 40 target proteins) that has been designed to evaluate the underlying performance of docking tools, and not the sensitivity of that tool’s scoring metric to differences in simple molecular properties. Note that the DUD decoy selection approach uses a discontinuous representation of the molecular properties, while the approach mentioned above uses a continuous representation. The similarity between these two approaches for decoy selection and the DACCS approach for active selection is striking. It is a topic of further investigation whether much of the reported success of the DACCS method is due to over-training/poor active compound selection. The work of Huang et al. with the DUD dataset showed that, at least for DOCK, this approach of specifically matching the properties of the decoys to those of the actives can produce more difficult decoy sets than those chosen purely for drug-likeness in a number of cases. It should be noted that the differences in performance reported in the DUD paper are essentially anecdotal, since no error bars are reported. With that caveat in mind inspection of Fig. 4 of the DUD paper, in which comparisons of performance of DOCK between the DUD “own” decoy sets and some commonly available agnostic decoys are shown, is instructive. In 10 of the 12 cases presented the DUD “own” decoy set is the most challenging of the four decoy sets compared, implying that a property-matched decoy set can provide a more difficult background set than an “agnostic” or general decoy set. For a drug-like decoy set chosen without specific reference to the active compounds being screened for, which is therefore to be expected to be less challenging than decoy sets designed with specific reference to the actives being searched for, see Ref [53]. The DUD datasets also have relatively high prevalence (the design goal was to achieve a prevalence close to 3%, though this varies slightly from target to target), giving the results generated with DUD reasonably low errors (except for those cases where the number of actives is small). This is not to say that DUD is perfectly constructed––there are still some large differences between the properties of the actives and the decoys in some cases and in some cases there are so few active molecules that statistically robust results cannot be generated. An important property not considered in the selection of the DUD decoys is formal charge. As such, there are some sets of the DUD actives that are easily discriminated from the decoys based on formal charge. For example, the mean formal charge on the neuraminidase ligands is +1.76, while the mean charge on the decoys is +0.76. For acetylcholinesterase, the mean charge on the actives is −1.68, while the mean charge on the decoys is −0.76. As with the datasets illustrated in Fig. 8, the DUD active sets were not chosen with a view to structural diversity and some of the active sets consist entirely of closely analogous compounds. Very recently the original DUD dataset has been extended by adding more active compounds and by clustering the actives to remove trivially graph similar actives from the set [54]. This makes “DUD 2.0” a suitable dataset not only for docking approaches but also ligand-based techniques. It should be noted that there is a limit to the acceptable level of similarity between actives and presumptive decoys. When the decoys are too similar to the actives the assumption that the decoys are inactive becomes increasingly untenable, giving rise to large numbers of “false false positives”. Accordingly the problem of decoy selection is not yet completely solved, and may not admit of a single solution for all problems or tools. However, since in retrospective work the point is purely to gain a measure of the expectation of performance in as yet unperformed studies, the use of carefully designed decoy sets is mandatory. It is unfortunate that the docking targets in DUD (39 crystal structures and 1 homology model) were not selected with as much care as the small molecule datasets. In 6 of the 38 co-crystal structures in DUD (there is one apo structure in the set), the DPIs are 1.5 Å or more, resulting in significant uncertainty in the positioning of any atom in these structures. These structures are ALR2 (1AH3), COX-2 (1CX2), EGFR (1M17), GR (1M2XZ), InhA (1P44) and p38 (1KV2). Accordingly docking results from these structures should be interpreted with great care. Conclusions Large numbers of evaluations and comparisons of tools for pose prediction and virtual screening have been published in recent years, an indication of significant interest in identifying tools that will have robust performance in one or both of these areas. Unfortunately the vast majority of these studies have been invalidated by poor choice of datasets, lack of consideration of error in source data and use of metrics that do not permit robust comparisons. For those papers using crystal structure data, too little account is taken both of the unavoidable imprecision in these structures and of the errors of fitting that are regrettably frequently seen in structures in the RCSB. In many cases nominal resolution, a measure of the quantity of data gathered, is confused with a measure of quality for the structure and other metrics indicating quality and reliability (DPI and Rfree) are ignored. When performing pose prediction geometric measures such as RMSD are almost always used to compare the experimental and predicted pose. These measures are uniformly used without taking into account either the inevitable imprecision in the atomic positions in crystal structures or the fact that using geometric measures necessarily implies comparing a model for the source data with a computed pose. In almost no cases are crystal structures inspected for errors in fitting. Without taking all these sources of error into account the results of any publication that uses crystal structure data will be suspect and of little use in deciding what tools are the most suitable for the task at hand. In papers concerned with virtual screening there has been, until recently, too little focus on eliminating trivial reasons for good performance from a given tool. The DUD dataset [52] has illustrated ways in which challenging virtual screening datasets can be constructed and, since it is publicly available, DUD offers the opportunity for a common benchmark upon which a wide variety of tools can be compared. The plethora of metrics used to judge and compare virtual screening performance serves merely to confuse the field rather then to clarify it. The lack of confidence intervals on metrics for success makes meaningful comparisons between tools almost impossible to interpret. The AUC for ROC offers great promise as a metric for virtual screening, as it offers the possibility of predictive value along with robust errors. For an exemplary use of ROC in virtual screening tests see a recent paper by Jain [55]. It is hoped that the field will soon converge to a single metric of virtual screening performance, such as the ROC, that will allow robust and direct comparisons between tools and between studies.

Childs_Nerv_Syst-4-1-2413122

Secondary meningioma in a long-term survivor of atypical teratoid/rhabdoid tumour with a germline INI1 mutation

Objective We report on a patient who developed a meningioma more than two decades after removal at a young age of an atypical teratoid/rhabdoid tumour (AT/RT), which was due to a germline INI1 mutation, and radio- and chemotherapy. Introduction We previously reported on a family in which four cousins developed an atypical teratoid/rhabdoid tumour (AT/RT) at young age, due to inheritance of a germline G>A mutation in the donor splice site of exon 4 of the tumour suppressor gene INI1 [2]. More than 20 years after resection of the AT/RT, the oldest of the four cousins developed an intracranial meningioma and, after an additional year, a myoepithelioma of the lip. In addition to the constitutional INI1 mutation and loss of the normal INI1 allele, we detected an identical rearrangement in the NF2 gene, not affecting its coding regions, in the AT/RT and myoepithelioma of the patient and concluded that they both originated from a common precursor cell. However, the rearrangement in NF2 and loss of the normal INI1 allele could not be demonstrated in the meningioma. Despite this, the development of the meningioma may be the consequence of the presence of the constitutional INI1 mutation, because mutations in this gene have been identified in meningiomas [8]. Alternatively, the meningioma could be induced by the radiation therapy, which our patient received after removal of the AT/RT and which is a well-known consequence of this treatment [1]. We investigated the meningioma for the presence of genetic changes reported in radiation-induced meningiomas and performed a comparative loss of heterozygosity (LOH) analysis to demonstrate that such radiation-induced genetic changes are absent in the AT/RT and myoepithelioma of our patient. Materials and methods Tumour samples and DNA isolation DNA from the AT/RT and the myoepithelioma was isolated from formalin-fixed and paraffin-embedded tumour material, and DNA from the meningioma was isolated from a fresh frozen tumour sample using commercially available kits (Qiagen, Venlo, The Netherlands). SNP analysis A genome-wide single nucleotide polymorphism (SNP) analysis of normal and meningioma DNA was conducted with the Affymetrix GeneChip Human Mapping 10K Array Xba 142 2.0. The array contains 10,204 SNPs distributed across the genome. The GeneChips were processed by an authorised Affymetrix Service Provider using standard Affymetrix protocols (Service XS, Leiden, The Netherlands). LOH regions were identified by comparing the SNP profiles of normal and meningioma DNA and scoring heterozygous SNP markers (AB) in the normal DNA that were reduced to homozygosity (AA or BB) in the meningioma DNA. Intervals were determined on the basis of locations of SNP markers in the UCSC Genome Browser, March 2006 Build, at http://genome.ucsc.edu/. LOH analysis Loss of heterozygosity (LOH) analysis was performed with microsatellite markers as described previously [3] using primer sequences and conditions for polymerase chain reaction given by The Genome Database at http://www.gdb.org/. Because of rather extensive degradation of DNA extracted from the AT/RT, only markers generating PCR fragments smaller than 150 bp were used. Case report History Our patient was diagnosed with a tumour in the fourth ventricle and right cerebellar hemisphere at the age of 4.5 years. Resection of the tumour was incomplete. The tumour was originally classified as anaplastic ependymoma WHO grade III, but upon extensive genetic analysis and histopathological re-evaluation, reclassified as AT/RT [2]. He received adjuvant chemotherapy (methotrexate, vincristine and prednisolone) followed by craniospinal radiotherapy (3,300 cGy in 22 doses) with a boost of 2,100 cGy on the location of the tumour. Fifteen years later, a brain CT without contrast showed no signs of recurrent disease or of other new pathology. Almost 6 years later, 20.5 years after initial diagnosis, he presented again with bifrontal headaches, increasing ataxic gait disturbances and fatigue. CT examination showed a large isodens space-occupying lesion in the left temporal region with oedema, midline shift and contrast enhancement. One year after resection of this tumour, he developed a myoepithelioma of the upper lip. The patient is alive at 29 years of age. Histological findings Histopathological examination of the left temporal tumour revealed an atypical meningothelial meningioma WHO grade II with many vessels and up to six mitoses per 2 mm2. SNP and LOH analysis Radiation-induced meningiomas usually display a complex karyotype [1]. To determine the genetic changes, we performed a genome-wide scan by SNP analysis of the meningioma and corresponding normal DNA. The identified chromosomal regions displaying LOH are listed in Table 1. Next, we tested several markers from within each of the five largest deleted regions by conventional microsatellite analysis to determine their LOH status in the meningioma and in the AT/RT and myoepithelioma. Presence of LOH in the meningioma, but absence of LOH in AT/RT and myoepithelioma was found for D1S199 in 1pter-p34.3 (Fig. 1a), D2S327 in 2pter-p12, D3S3727 in 3p26.1-cen (Fig. 1b), D16S3020 in 16pter-p12.1 and D17S1872 in 17qcen-q21.31 (Fig. 1c). In contrast, as determined previously [2] and exemplified by marker D22S430 in Fig. 1d, there was LOH for chromosome 22 markers in the AT/RT and myoepithelioma, but not in the meningioma. Fig. 1LOH analysis using markers from chromosome regions a 1pter-p34.3 (D1S199), b 3p26.1-cen (D3S3727), c 17qcen-q21.31 (D17S1872) and d 22q (D22S430) of AT/RT (A), meningioma (M) and myoepithelioma (My) DNAs of the patient. N denotes normal DNA derived from the patient’s blood leucocytes. Arrowhead indicates position of lost alleleTable 1LOH regions in the meningioma, as determined by SNP analysisChromosome regionExact interval (Mb)1pter-p34.30–38.901p33-p3346.99–51.091p32.2-1p31.358.19–65.502pter-p120–79.283p26.1-cen6.11–88.847q36.1-qter150.04–158.6616pter-p12.10–22.7516q23.1-q23.373.46–81.3317cen-q21.3123.09–40.3519pter-p13.20–8.83 Discussion In a previous study, we provided evidence for the involvement of INI1, by constitutional mutation and independent somatic loss, in the development of both the AT/RT and the myoepithelioma of our patient. Moreover, we could demonstrate that these tumours originated from an identical precursor cell [2]. Although involvement of INI1 has been described in sporadic meningiomas [8], until now meningiomas have not been reported to occur as a consequence of the inheritance of an INI1 mutation. However, this might be due to the extreme rarity of families with an inherited INI1 mutation, with only four published cases so far [2, 5, 6, 11]. In this context, it is interesting to note that (as demonstrated before and exemplified by marker D22S430 in Fig. 1d) the meningioma of our patient has no LOH 22 and the retained INI1 allele carries no additional inactivating mutation [2], indicating that this tumour is not the result of inheritance of the mutated INI1 gene. In contrast, the meningioma displayed genetic changes that are characteristic for radiation-induced meningiomas [1, 7, 9], such as a complex karyotype, as evidenced by the SNP analysis (Table 1), including extensive losses on chromosome arm 1p, but no loss of chromosome 22, and no mutation in the NF2 gene [2]. The genetic changes found in the meningioma proved not to be present in the AT/RT or myoepithelioma (Fig. 1a–c, and data not shown), demonstrating that the meningioma is genetically not related to these tumours and should not be considered as a late and more differentiated recurrence or metastasis of the AT/RT. In conclusion, the clinical and genetic characteristics of the meningioma clearly indicate that the tumour developed as a consequence of the radiation therapy. Recent evidence suggests that long-term survival of patients with a childhood AT/RT can occur especially after aggressive treatment including radiotherapy [10, 12]. The possible late effects of radiation therapy have been considered as serious drawbacks of this treatment. This is the first case in literature in which such a possible consequence in the form of a radiation-induced meningioma is actually documented in a long-term survivor of an AT/RT. Compared to sporadic meningiomas, radiation-induced meningiomas are more aggressive and have a higher chance to recur [1]. Because of this unfavourable prognosis, continuous medical surveillance of our patient is warranted.

Diabetologia-4-1-2362134

Increased osteoclastic activity in acute Charcot’s osteoarthopathy: the role of receptor activator of nuclear factor-kappaB ligand

Aims/hypothesis Our aims were to compare osteoclastic activity between patients with acute Charcot’s osteoarthropathy and diabetic and healthy controls, and to determine the effect of the receptor activator of nuclear factor-kappaB ligand (RANKL) and its decoy receptor osteoprotegerin (OPG). Introduction Although Charcot’s osteoarthropathy is characterised by increased local bone resorption [1], the exact cellular mechanisms contributing to the pathogenesis of this condition remain unresolved. Osteoclasts have been shown to be the principal cell type responsible for bone resorption [2]. These cells originate from the haemopoietic lineage and are known to undergo various stages of proliferation, fusion and differentiation before they are fully functionally active, mature osteoclasts. Recently, receptor activator of nuclear factor-kappaB (RANK) ligand (RANKL) has been identified as an essential mediator of osteoclast formation and activation [3]. RANKL is expressed on a variety of cell types such as bone forming osteoblasts, T lymphocytes, dendritic cells, endothelial cells and fibroblasts. RANKL mediates the process of osteoclastogenesis by binding to its RANK, which is expressed on mononuclear osteoclast precursors. The effects of RANKL–RANK interaction are physiologically counterbalanced by osteoprotegerin (OPG), which acts as a soluble receptor decoy for RANKL and blocks the interaction of RANKL with RANK. The ratio of RANKL to OPG has been suggested to regulate the extent of osteoclast formation and resorption. Therefore, any alteration in the RANKL/OPG ratio could be critical in the pathogenesis of osteolytic bone disorders [4]. Recently, Jeffcoate hypothesised that the RANK/RANKL/OPG pathway may play an important role in the osteolysis seen in acute Charcot’s osteoarthropathy [5]. Using an in vitro technique to generate functional human osteoclasts from peripheral blood monocytes (PBMCs) [6] in the presence of macrophage-colony stimulating factor (M-CSF) [7] and soluble RANKL, it is possible to determine the cellular mechanisms involved in the process of osteoclast formation and resorption in physiological and pathological conditions. To our knowledge, this technique has not yet been studied in patients with Charcot’s osteoarthropathy. The aims of this study were: (1) to generate functional human osteoclasts in vitro from diabetic patients with acute Charcot’s osteoarthropathy and from healthy and diabetic control participants; (2) to compare the extent of osteoclast formation and resorption; and (3) to determine the role of the RANK/RANKL/OPG pathway in osteoclastic activity in Charcot’s osteoarthropathy. Methods Patients We studied nine consecutive diabetic patients with recent onset of acute Charcot’s osteoarthropathy (five men, four women; five type 1, four type 2 diabetes), eight diabetic patients with no previous history of Charcot’s osteoarthropathy (five men, three women; four type 1, four type 2) and eight healthy control participants (five men, three women). Patients with acute Charcot’s osteoarthropathy were matched for age and duration of diabetes with the diabetic control patients and for age with the healthy control participants. The mean age was similar between patients with Charcot’s osteoarthropathy and diabetic control patients (53 ± 2.8 versus 59 ± 2.9 years [mean ± SEM], p = 0.167) as was the mean age between the former and healthy control participants (53 ± 2.8 versus 47 ± 2.7 years, p = 0.114). The mean duration of diabetes was similar in both groups with diabetes (31 ± 5.1 [Charcot patients] versus 27 ± 4.6 years, p = 0.606). Diabetes control as indicated by glycated Hb was also similar in the two diabetes groups (7.7 ± 0.6 [Charcot’s] versus 7.8 ± 0.4%, p = 0.743). Diagnosis of Charcot’s osteoarthropathy was made on the presentation of a hot swollen foot, with skin foot temperature 2°C greater than the corresponding site on the contralateral foot and with typical radiological changes of subluxation, dislocation or fragmentation of bone on standard foot radiographs [8]. All patients had intact feet and no evidence of foot infection or ulceration. Ethical permission for this study was obtained from the King’s College Hospital Research Ethics Committee and all participants gave written informed consent. Isolation and culture of monocytes Peripheral blood mononuclear cells were isolated as previously described [6]. Briefly, blood was diluted 1:1 in α-minimum essential medium (MEM; Invitrogen, Paisley, UK), layered over Histopaque and centrifuged (693 g) for 20 min. The interface layer was resuspended in MEM, then centrifuged (600 g) for a further 10 min. The resultant cells were resuspended in MEM with 10% heat-inactivated FCS and counted in a haemocytometer following lysis of erythrocytes by a 5% (vol./vol.) acetic acid solution. To assess the extent of osteoclast formation and resorption, PBMCs were cultured on glass coverslips and dentine slices. Initially, 5 × 105 PBMCs were added to 6-mm diameter glass coverslips and 4-mm diameter dentine slices in MEM containing 100 UI/ml penicillin, 100 μg/ml streptomycin and 10% FCS (Gibco, Paisley, UK). After 2 h incubation, coverslips and dentine slices were vigorously rinsed in medium to remove non-adherent cells. The cultures were maintained in MEM/FCS under three different culture conditions: (1) human M-CSF (R&D Systems Europe, Abingdon, UK) alone at 25 ng/ml; (2) M-CSF plus 100 ng/ml human soluble RANKL (sRANKL; Peprotech, London, UK) (a concentration known to facilitate differentiation of osteoclast precursors to active bone-resorbing osteoclasts in vitro); and (3) M-CSF plus sRANKL plus 250 ng/ml human OPG (R&D Systems Europe). Coverslips and dentine slices were cultured at 37°C in 5% CO2 for 14 and 21 days respectively. Osteoclast formation After 14 days, the coverslips were examined histochemically for the expression of tartrate-resistant acid phosphatase (TRAcP), an osteoclast marker. Coverslips with newly formed osteoclasts were collected and rinsed in PBS buffer, fixed with formalin (10% [vol./vol.] in PBS buffer) for 10 min and rinsed in distilled water. TRAcP was histochemically revealed by a simultaneous coupling reaction using Naphtol AS-BI-phosphate as substrate and Fast violet B as the diazonium salt. The coverslips were incubated for 90 min at 37°C in a dark room, rinsed three times in distilled water and the residual activity was inhibited by 4% NaF (wt/wt) for 30 min. Coverslips were then rinsed in distilled water, counterstained with DAPI for 20 min and allowed to dry before mounting, using an aqueous medium. TRAcP-positive cells with more than three nuclei were identified as osteoclasts. The number of newly generated osteoclasts was assessed using a light microscope examination. Osteoclast resorption After 21 days, the dentine slices were removed from the culture wells, placed in NH4OH (1 mol/l) for 30 min and sonicated for 5 min to remove any adherent cells. They were then rinsed in distilled water and stained with 0.5% (vol./vol.) toluidine blue prior to examination by light microscopy. The surface of each dentine slice was examined for evidence of lacunar resorption and the extent of eroded surface on each dentine slice was determined using image analysis and expressed as the percentage of surface area resorbed. Statistical analyses Data were expressed as a mean ± SEM. Initially the difference within the three study groups (Charcot patients, healthy and diabetic controls) was assessed with the non-parametric Kruskall–Wallis test. Then the differences between Charcot and diabetic patients, and Charcot patients and healthy controls were assessed by the non-parametric Mann–Whitney U test. In each patient group, the differences between the various culture conditions were also assessed using the Mann–Whitney U test. Differences were considered significant at p < 0.05. Results Osteoclast cultures in the presence of M-CSF Osteoclast formation The mean number of newly formed TRAcP-positive multinucleated osteoclasts in the presence of M-CSF alone was significantly greater in the patients with acute Charcot’s osteoarthropathy (48.6 ± 18.2) than in diabetic (6.8 ± 2.7) and healthy control participants (5.0 ± 0.7) (p = 0.008). The number of TRAcP-positive multinucleated osteoclasts formed in acute Charcot’s osteoarthropathy was 7.2 and 9.7 times greater than those formed in diabetic (p = 0.010) and healthy control groups (p = 0.003), respectively. Osteoclast resorption The newly formed osteoclasts exhibited increased functional activity as demonstrated by the extent of resorption on dentine slices, with percentage area resorption significantly elevated in the patients with acute Charcot’s osteoarthropathy (0.264 ± 0.06%) compared with diabetic (0.000 ± 0.00%) and healthy control groups (0.004 ± 0.01) (p < 0.0001). The percentage of resorption was significantly greater in the Charcot patients than in the diabetic (p = 0.001) and healthy control groups (p = 0.001). Osteoclast cultures in the presence of M-CSF and sRANKL Osteoclast formation The addition of sRANKL led to an increase in the number of TRAcP-positive multinucleated osteoclasts in all three groups of patients. The mean number of these osteoclasts in patients with acute Charcot’s osteoarthropathy was 96.0 ± 21.6, which was significantly greater than that in the diabetic (56.5 ± 11.5) and healthy (29.0 ± 5.1) control groups (p = 0.010; Fig. 1a,c,e). The number of TRAcP-positive multinucleated osteoclasts in the patients with acute Charcot’s osteoarthropathy was 1.7 times higher than in diabetic control patients, but this finding did not reach significance (p = 0.105). However, the number of these osteoclasts in the acute Charcot group was 3.3 times (and significantly) higher than in the healthy control group (p = 0.005). When the number of cells in the cultures with M-CSF alone was compared with that after the addition of sRANKL, there was a significant increase in the diabetic control patients (from 6.8 ± 2.7 to 56.5 ± 11.5, p = 0.003) and in the healthy participants (from 5.0 ± 0.7 to 29.0 ± 5.1, p = 0.002), while the increase in the number of TRAcP-positive multinucleated osteoclasts in the acute Charcot group failed to reach significance (increase from 48.6 ± 18.2 to 96.0 ± 21.6, p = 0.059; Fig. 2a). Fig. 1Multinucleated TRAcP-positive cells were formed on glass coverslips (a, c, e) capable of lacunar resorption (b, d, f) after 14 and 21 days incubation, respectively, in the presence of 25 ng/ml human M-CSF and 100 ng/ml sRANKL. Newly formed osteoclasts were numerous and highly active in Charcot’s patients (a, b) compared with diabetic (c, d) and healthy control (e, f) participants. Scale bars, 10 μmFig. 2a Quantitative comparison between the number (n) of TRAcP-positive cells formed in cultures with M-CSF alone (white bars) or with M-CSF and sRANKL (black bars) in patients with Charcot’s osteoarthropathy and diabetic and healthy control participants. b Quantitative comparison between the percentage area resorption in the same cultures and patient groups. Statistical differences between the groups were determined using the Mann–Whitney U test, with significance as follows: a Charcot’s p = 0.059, diabetic control p = 0.003, healthy control p = 0.002; b Charcot’s p < 0.0001, diabetic control p < 0.0001, healthy control p < 0.0001 Osteoclast resorption The percentage area resorption on dentine slices with M-CSF and sRANKL was significantly increased in the acute Charcot group (41.6 ± 8.1%) compared with that in the diabetic (14.2 ± 16.5%) and healthy control groups (10.5 ± 1.9%; p = 0.005). Resorption in the Charcot patients was 2.9 times higher than in diabetic control patients (p = 0.008) and four times higher than in healthy participants (p = 0.005; Fig. 1b,d,f). The addition of sRANKL to the cultures with M-CSF led to the following rises in the percentage area resorption when compared with M-CSF alone: Charcot’s 0.264 ± 0.06% to 41.6 ± 8.1%, p < 0.0001; diabetic control 0.000 ± 0.00% to 14.2 ± 16.5%, p < 0.0001; healthy control 0.004 ± 0.01% to 10.5 ± 1.9%, p < 0.0001 (Fig. 2b). Osteoclast cultures in the presence of M-CSF, sRANKL and excess concentrations of OPG Osteoclast formation The addition of excess concentrations of OPG led to a reduction in the number of TRAcP-positive multinucleated osteoclasts in the cultures with M-CSF, sRANKL and OPG in all the three groups of patients. However, after the addition of OPG, the number of TRAcP-positive multinucleated osteoclasts was still significantly increased in the Charcot group (54.4 ± 17.6), as compared with diabetic (8.8 ± 5.3) and healthy control participants (4.4 ± 1.2; p = 0.003). In the cultures with M-CSF, sRANKL and OPG, the number of TRAcP-positive multinucleated osteoclasts was greater in the Charcot patients than in the diabetic (p = 0.005) and healthy control groups (p = 0.001). When OPG was added to the cultures with M-CSF and sRANKL, the reduction in the number of TRAcP-positive cells in Charcot patients was not significant (96.0 ± 21.6 versus 54.4 ± 17.6, p = 0.189). OPG on the other hand significantly inhibited the number of TRAcP-positive cells in M-CSF and RANKL-mediated cultures from diabetic (reduced from 56.5 ± 11.5 to 8.8 ± 5.3, p = 0.005) and healthy control participants (29.0 ± 5.1 to 4.4 ± 1.2, p = 0.003; Fig. 3a). Fig. 3a Comparison between the number (n) of TRAcP-positive cells formed in cultures with M-CSF and sRANKL (black bars) or with M-CSF, sRANKL and excess concentrations of OPG (250 ng/ml) (grey bars) in patients with Charcot’s osteoarthropathy and diabetic and healthy control participants. b Comparison between the percentage area resorption in the same cultures and patient groups. Statistical differences between the groups were determined using the Mann–Whitney U test, with significance as follows: a Charcot’s p = 0.189, diabetic control p = 0.005, healthy control p = 0.003; b Charcot's p = 0.001, diabetic control p = 0.001, healthy control p < 0.0001 Osteoclast resorption The addition of OPG led to a marked reduction of the percentage area resorption on dentine slices in Charcot patients (from 41.6 ± 8.1% to 5.9 ± 2.4%, p = 0.001) and also in diabetic (14.2 ± 16.5% to 0.45 ± 0.31%, p = 0.001) and healthy control (from 10.5 ± 1.9% to 0.00 ± 0.00%, p < 0.0001) participants (Fig. 3b). However, the percentage area resorption on the dentine slices was still greater in the cultures with M-CSF, RANKL and OPG from the patients with acute Charcot’s osteoarthropathy (5.9 ± 2.4%) than in those from diabetic (0.45 ± 0.31%) and healthy control (0.00 ± 0.00%) participants (p = 0.003). Resorption on the dentine slices was greater in the Charcot patients than in diabetic (p = 0.005) and healthy control (p = 0.003) groups. Discussion This study shows that monocytes from patients with acute Charcot’s osteoarthropathy cultured in the presence of M-CSF alone were capable of differentiating into mature osteoclasts that exhibited increased resorption compared with diabetic and healthy control participants. Furthermore, osteoclasts generated after the addition of sRANKL were functionally more aggressive, exhibiting a considerable increase in the extent of resorbing activity in patients with acute Charcot’s osteoarthropathy. This resorption was partially blocked by the addition of excess concentrations of OPG, a soluble receptor decoy for RANKL. This suggests that the increased osteoclastic activity in patients with acute Charcot’s osteoarthropathy is mediated through both a RANKL-dependent and a RANKL-independent pathway. Cultures from the patients with Charcot’s osteoarthropathy showed increased osteoclast formation and resorption when cultured with M-CSF alone. Although M-CSF is an essential factor for proliferation, differentiation and survival of the monocyte-macrophage lineage [9, 10], it is not an osteoclastogenic factor and it is unusual to detect osteoclast formation and resorption in cultures with M-CSF, as was seen in the diabetic and healthy controls. This observation suggests that in acute Charcot’s osteoarthropathy there may be increased levels of other circulating pro-inflammatory factors such as TNF-α [11, 12], IL-6 [13], IL-8 [14] and LIGHT (homologous to lymphotoxins exhibiting inducible expression and competing with herpes simplex virus glycoprotein D for herpes virus entry mediator [HVEM], a receptor expressed by T lymphocytes) [15], which have been previously shown to stimulate osteoclastogenesis independently of RANK/RANKL mechanisms. The concentrations of these circulating factors in diabetic and healthy control participants may not be sufficient to induce the formation and differentiation of active osteoclasts in the presence of M-CSF alone. After the addition of sRANKL to M-CSF cultures, the newly formed osteoclasts exhibited markedly increased resorption in the patients with Charcot’s osteoarthropathy, although the number of osteoclasts did not significantly increase in these patients compared with cultures with M-CSF alone. These observations may not be unique to Charcot’s osteoarthropathy, and indeed similar observations have been reported in other conditions associated with increased bone resorption, such as rheumatoid arthritis where the addition of sRANKL resulted in a significant increase in lacunar resorption, but did not lead to a significant increase in the number of TRAcP-positive cells [16]. Overall, the observed extensive resorption in acute Charcot patients, in the presence of M-CSF and sRANKL, as compared with the diabetic or healthy control groups, may suggest that the osteoclast precursors circulating in acute Charcot patients are in a higher activated state and as such are more primed to becoming osteoclasts (mediated through RANKL) than those in the control groups. In order to ascertain that RANKL was a major osteoclastic activator in patients with Charcot’s osteoarthropathy, excess concentrations of OPG, the soluble receptor decoy to RANKL, were added to the cultures with M-CSF and RANKL. The rationale for this approach was that if osteoclastogenesis is mediated solely through RANK–RANKL interaction, addition of excess concentrations of OPG (as had been previously determined to be sufficient to block osteoclastogenesis through RANKL [15]) would completely abolish the process of osteoclast differentiation and activation. In the current study, although osteoclast formation and resorption in the diabetic and healthy control groups was completely blocked by the addition of OPG, the latter did not achieve total inhibition of osteoclast formation and resorption in patients with acute Charcot’s osteoarthropathy. These results suggest that although RANKL-dependent pathways do play a significant role in the osteoclastic activity of Charcot’s osteoarthropathy, an alternative pathway (other than RANK/RANKL) may also be involved. Osteoclastogenic mediators other than RANKL that have been reported to stimulate osteoclast differentiation independently of the RANKL pathway include TNF-α [11, 12], IL-6 [13], IL-8 [14] and LIGHT [15]. In acute Charcot’s osteoarthropathy, it is possible that one or a combination of these factors may have initiated the circulating osteoclast precursors to be in a more ‘primed’ condition, a situation which as such could explain the observed resorption in Charcot monocyte cultures supplemented with M-CSF alone, without the exogenous addition of any osteoclastogenic mediators. The osteolysis of Charcot’s osteoarthropathy may be explained by our observation that osteoclast precursors from Charcot patients develop into mature osteoclasts that exhibit increased resorptive activity, especially in response to RANKL, unlike the increased resorption in response to bacterial infection, which is not mediated by RANKL [17]. Increased expression of RANKL has been previously demonstrated in pathological osteolysis associated with the development of various bone diseases [18] and a similar mechanism may contribute to osteolysis of Charcot’s osteoarthropathy [5]. Furthermore, patients with Charcot’s osteoarthropathy have severe neuropathy, which itself can also lead to increased expression of RANKL as a result of the loss of nerve-derived peptides known to antagonise its effect such as calcitonin gene-related peptide [5]. In addition to the RANKL-dependent pathway, our results suggest that a RANKL-independent pathway, mediated by pro-inflammatory cytokines, may also be important. Indeed, Charcot’s foot is characterised by excessive inflammation and proinflammatory cytokines have been implicated in its pathogenesis [19]. In support of this, a recent immunohistochemical analysis of bone samples isolated from Charcot’s osteoarthropathy patients showed excessive osteoclastic activity in a microenvironment enriched with mediators of bone resorption (IL-1, IL-6 and TNF-α) [20]. Thus a RANKL-independent pathway, which is also known to play a role in other osteolytic disorders such as rheumatoid arthritis [21] and aseptic loosening [22], could contribute also to the pathogenesis of the Charcot’s osteoarthropathy. This study has indicated, for the first time that the RANKL-dependent pathway is important in the pathogenesis of Charcot’s osteoarthropathy, thereby raising the possibility of the use of RANKL inhibition in the management of Charcot’s foot. However, our observations also suggest that a RANKL-independent pathway may play a role, but further investigation is required to fully clarify the mechanism involved. If confirmed, specific pharmacological agents that counteract the RANKL-independent pathway, such as anti-TNF strategies, may be useful in the treatment of Charcot’s osteoarthropathy. Whatever the relative importance of either pathway, this in vitro technique of generating human osteoclasts from PBMCs may allow specific characterisation of osteoclastic activity in each patient and could, in the future, lead to individually tailored anti-osteoclastic treatment for the patient with acute Charcot’s osteoarthropathy.

J_Chem_Ecol-3-1-1915630

Jasmonic Acid-Induced Changes in Brassica oleracea Affect Oviposition Preference of Two Specialist Herbivores

Jasmonic acid (JA) is a key hormone involved in plant defense responses. The effect of JA treatment of cabbage plants on their acceptability for oviposition by two species of cabbage white butterflies, Pieris rapae and P. brassicae, was investigated. Both butterfly species laid fewer eggs on leaves of JA-treated plants compared to control plants. We show that this is due to processes in the plant after JA treatment rather than an effect of JA itself. The oviposition preference for control plants is adaptive, as development time from larval hatch until pupation of P. rapae caterpillars was longer on JA-treated plants. Total glucosinolate content in leaf surface extracts was similar for control and treated plants; however, two of the five glucosinolates were present in lower amounts in leaf surface extracts of JA-treated plants. When the butterflies were offered a choice between the purified glucosinolate fraction isolated from leaf surface extracts of JA-treated plants and that from control plants, they did not discriminate. Changes in leaf surface glucosinolate profile, therefore, do not seem to explain the change in oviposition preference of the butterflies after JA treatment, suggesting that as yet unknown infochemicals are involved. Introduction Plants can be attacked by many herbivorous insects and have evolved a variety of defense strategies, including morphological barriers, synthesis of toxic or repellent secondary metabolites, and the release of synomones that attract natural enemies of the herbivores. These defenses can be constitutive, i.e., expressed independent of the presence of an attacker, or inducible, in which case defense compounds accumulate in response to attack (Karban and Baldwin, 1997). Herbivores can detect induced defensive compounds and respond by avoiding these plants, which signal lower suitability as a host plant (Landolt, 1993; De Moraes et al., 2001; Kessler and Baldwin, 2001; Meiners et al., 2005). Induced plant defense can affect herbivorous insects directly through the production of toxic compounds or indirectly through the production of cues that indicate intra- or interspecific competition for the herbivores (Schoonhoven et al., 2005). Moreover, induced plant defense signals can reduce the enemy-free space for the herbivores. For parasitoids and predators, induced infochemicals may indicate the presence of their host or prey on the plant (Turlings et al., 1990; Dicke and Vet, 1999). Phenotypic changes in individual plants may therefore affect insects at different trophic levels, and thus, the composition of the insect community and food web associated with the plant (Price et al., 1980; Van Zandt and Agrawal, 2004; Takabayashi et al., 2006). Already in the 19th century, Kirby and Spence (1863) observed that Pieris brassicae females preferred to lay their eggs on plants devoid of eggs. Later, this was confirmed under more controlled conditions by Rothschild and Schoonhoven (1977) for both P. brassicae and Pieris rapae. This avoidance of infested plants is caused by a physiological response of the plant to oviposition, rather than by compounds excreted by the butterflies themselves (Blaakmeer et al., 1994). The butterflies also avoid egg deposition on leaves with feeding larvae (Rothschild and Schoonhoven, 1977). It was postulated that butterflies avoid laying eggs on herbivore-infested plants because herbivore attack induces defense compounds in plants that can influence the performance of their offspring and to reduce the risk of inter- or intraspecific competition and parasitism (Thompson and Pellmyr, 1991; Shiojiri et al., 2002). Egg-induced chemical changes in Brassica plants are also known to arrest Trichogramma parasitoids that parasitize Pieris eggs (Fatouros et al., 2005). Oviposition-site selection involves an important behavioral decision in the life cycle of an herbivorous insect because hatching larvae have limited dispersal capacity (Renwick and Chew, 1994). P. rapae is a solitary butterfly that lays one egg at a time, whereas P. brassicae is gregarious and lays batches of about 20–100 eggs. P. rapae appears to spread the risk of larval mortality, laying few eggs within any patch. This has the advantage of being able to exploit isolated plants (Davies and Gilbert, 1985). P. brassicae, however, needs patches of plants because one large egg batch will require more than one plant for all caterpillars to develop into adults. Oviposition-site selection is performed in consecutive phases of searching and contact evaluation. P. rapae butterflies use visual, olfactory and tactile cues during these phases (Rothschild and Schoonhoven, 1977; Renwick and Radke, 1988). Acceptance of a site may be determined by the balance of positive and negative factors (Renwick and Radke, 1988). Renwick and Radke (1988) suggest that olfaction does not play a role in attraction to a host plant, but may be involved in avoidance of non-host plants. P. rapae and P. brassicae are crucifer specialists and are known to use glucosinolates, toxic secondary metabolites characteristic for Brassicaceae, as oviposition stimulants. Glucobrassicin and sinigrin are effective oviposition stimulants for P. brassicae and P. rapae (Renwick et al., 1992; Van Loon et al., 1992a). A major signal-transduction pathway involved in induced plant defense is the octadecanoid pathway (Arimura et al., 2005). A central compound in the pathway is jasmonic acid (JA), which has an important role in direct and indirect defense in many plant species. In response to JA or methyl jasmonate (MeJA) treatment, increased concentrations of several compounds have been documented in a range of plant species, e.g., proteinase inhibitors (Moura and Ryan, 2001), polyphenol oxidases (Thaler et al., 1996), nicotine (Baldwin et al., 1996), trypsin inhibitors (Cipollini and Sipe, 2001), glucosinolates (Cipollini and Sipe, 2001; Van Dam et al., 2004; Mewis et al., 2005), and increased volatile emission (Boland et al., 1995; Dicke et al., 1999; Koch et al., 1999). Herbivores are reported to be affected by JA treatment of plants. Several studies have focused on the larval stage of the herbivores and have shown reduced relative growth rates and leaf consumption (Van Dam et al., 2000; Gols et al., 2003; Van Dam et al., 2004). In field experiments, spraying of JA decreased the abundance of caterpillars, flea beetles, aphids, and thrips (Thaler et al., 2001). Other studies addressed the influence of JA application to plants on oviposition-site selection behavior of adult herbivores. These studies showed that JA application can result in induced resistance as well as induced susceptibility (Stanjek et al., 1997; Kessler and Baldwin, 2001; Lu et al., 2004). Here, we studied how JA application affects oviposition of two specialist herbivores on cabbage, Pieris rapae L. and P. brassicae L. (Lepidoptera: Pieridae) that are closely related, yet differ drastically in the amount of eggs they put on one plant. Our study is the first to compare closely related herbivores with a different oviposition strategy, which might affect the consequences of JA-induced responses. JA is known to mediate the induction of chemical defense responses in plants to feeding damage and deposition of eggs (Dicke and Van Poecke, 2002; Hilker and Meiners, 2006; Mumm and Hilker, 2006). By using JA, we were able to examine the effects of induced defense responses in cabbage. Moreover, JA application has the advantage that visually detectable damage and the presence of herbivores or eggs are avoided. Finally, JA allows control over the strength of induction through controlled dosages. We hypothesized that JA treatment would inhibit the oviposition of the butterflies. We made solvent extracts to address the identity of the active plant compounds that influenced butterfly behavior. Rather than testing whole-leaf extracts, however, we extracted the glucosinolates from the surface of both control and JA-treated plants and tested the oviposition preference of the butterflies for these glucosinolate fractions on a neutral substrate. Furthermore, we included a control experiment to exclude a potential direct effect of JA on oviposition behavior. We addressed the following questions: (1) does JA treatment of cabbage plants affect host plant selection of the two Pieris butterfly species; (2) are there differences between solitary and gregarious butterflies; (3) does JA treatment affect glucosinolate levels in leaf surface extracts; and (4) do changes in glucosinolate levels determine the changes in oviposition preference? Methods and Materials Plants and Insects Brussels sprouts plants, Brassica oleracea var. gemmifera L. (Brassicaceae) cultivar Cyrus were grown from seed in a greenhouse in plastic pots (11 × 11 × 11 cm) at 20–28°C, 40–80% RH and a 16-hr light/8-hr dark photoperiod. All experiments were conducted with 6- to 7-wk-old plants. Stock colonies of the large cabbage white P. brassicae and the small cabbage white P. rapae were maintained on Brussels sprouts in a climatized room at 20–22°C, 50–70% RH and a 16-hr light/8-hr dark photoperiod. Chemical Analysis Brussels sprouts plants were sprayed with 0.1 mM JA or control solution. JA ((±) jasmonic acid, purity >97%; Sigma-Aldrich, St Louis, MO, USA) was applied to the surface of the leaves, i.e., plants were sprayed with a JA solution with 0.1% Tween 20 until run-off or just with 0.1% Tween 20 for the control. The next day, glucosinolates (GLS) were extracted from the surface of the intact Brussels sprouts leaves. Each sample consisted of four leaves (between the third to sixth leaf from the base of a plant) that were cut at the base of the petiole. Directly after cutting, the lamina was dipped for 5 sec in 300 ml of dichloromethane, and after a 5-sec interval, it was dipped for 5 sec in 150 ml of methanol (Städler and Roessingh, 1990; Van Loon et al., 1992a; Griffiths et al., 2001). The methanol was evaporated from the crude methanol dip-volume with a rotary evaporator (IKA-Werke GmbH, Staufen, Germany). For each treatment, 11 plants were sampled. The extract was redissolved in methanol, desulphatased on a DEAE-Sephadex A25 column, and separated on a reverse phase C-18 column by using high performance liquid chromatography (HPLC) as described in Van Dam et al. (2004). Glucosinolate detection was performed with a photodiode array (PDA) detector (200–350 nm) with 229 nm as the integration wavelength. Sinigrin (sinigrin monohydrate, ACROS, NJ, USA) was used as an external standard. We used the correction factors at 229 nm from Buchner (1987) and the EC (EC, 1990) to calculate the concentrations of the glucosinolates. Desulfoglucosinolate peaks were identified by comparison of HPLC retention times and UV spectra with standards kindly provided by M. Reichelt, MPI Chemical Ecology and a certified rape seed standard (Community Bureau of Reference, Brussels, code BCR-367R). The surface area was measured directly after dipping, and the dry mass of the leaves was measured after drying at 50°C for 72 hr. The GLS content was calculated in pmol per cm2 leaf material. Herbivore Oviposition Preference Test Pieris adults emerged from pupae in a large oviposition cage (67 × 100 × 75 cm) in a greenhouse compartment at 22–24°C and 50–70% RH. Apart from natural daylight, cages were illuminated by sodium vapor lamps (type SON-T, 500 W, Philips, The Netherlands) from 8:00 a.m. until 2:00 p.m. In this cage, they were provided with a 10% sucrose solution and an oviposition substrate; depending on the experiment, a plant or an artificial leaf made of green cardboard paper sprayed with sinigrin. For the experiments, one male and one female butterfly were introduced per oviposition cage (67 × 50 × 75 cm) in the same greenhouse compartment, on the day before the experiment. In these cages, the butterflies were also provided with sucrose solution. At 8:30 a.m., the treated leaves or papers and respective controls were introduced into the cages, and the butterflies were allowed to oviposit until the beginning of the afternoon. At 2:00 p.m., the leaves were removed, and the number of eggs was counted. The experiments were carried out in several cages per day and 3–4 d per treatment with new pairs of butterflies each day, adding up to a total of 24–36 independent replicates. Surface Application of JA The effect of JA-induced changes in Brussels sprouts plants on butterfly behavior was tested in oviposition experiments with P. brassicae and P. rapae. Three concentrations of JA solution, 0.01, 0.1, and 1 mM, corresponding to approximately 1.25, 12.5, and 125 μg JA/g fresh weight (or 0.25, 2.5, and 25 nmol JA/cm2) respectively, were sprayed on the plants and tested against a control (plants treated with 0.1% Tween 20). The next morning, just before the start of the experiment, the fourth, fifth, and sixth leaves from the base of the plants were cut, and their petioles were placed directly into a vial with tap water and introduced into the cages with butterflies. Systemic Uptake of JA For P. rapae, two application methods were used to assess the effect of JA-induced changes in Brussels sprouts on oviposition preference. For the second application method, the fourth, fifth, and sixth leaves were cut from untreated plants and placed in a 0.1-mM aqueous JA solution 22 hr before the start of the experiment. Total uptake of the solution was on average 6.3 ± 1.5 ml per control leaf and 6.0 ± 1.6 ml for JA-treated leaves (corresponding to approximately 20 μg JA/g fresh weight or 5 nmol JA/cm2, assuming homogeneous distribution over the leaf tissue after uptake). Effect of Pure JA on Oviposition Preference In the next experiment, green cardboard paper sprayed with an oviposition stimulant was used to test the effect of pure JA on the oviposition behavior of P. rapae on an inert substrate. Sinigrin has been shown to be a suitable oviposition stimulant for Pieris butterflies (Van Loon et al., 1992a) and was therefore used to stimulate oviposition on the artificial substrate in this experiment. The paper (8 × 11.5 cm) was treated with 1 ml of a 5-mM sinigrin solution (Janssen Pharmaceutica, Tilburg, The Netherlands) by spraying it with a Desaga chromatographic sprayer (Heidelberg, Germany). Subsequently, after drying, papers were sprayed with either 1 ml of a 1-mM JA solution or water (control substrates) just before the test (210 μg JA/carton or 11 nmol JA/cm2). Bioassays with Purified Glucosinolate (GLS) Fractions GLS were extracted from the leaf surface as described for the chemical analysis. For each treatment, control and 0.1 mM JA, 60 plants were used for the extraction, of which four to five leaves per plant were dipped. Subsequently, the extracts were fractionated following the protocol of Sørensen (Sørensen, 1990). The GLS fractions were stored in the freezer until analysis. The GLS were dissolved in methanol to make two concentrations, one corresponding to the amount of GLS extracted from the material of two plants in 0.8 ml and one concentration corresponding to the amount of GLS from one leaf in 0.8 ml. Hereafter, we will express these concentrations in gram leaf equivalents (gle). One gle corresponds to the amount of GLS extracted from 1 g fresh and intact leaf. With an average weight of 6 g per leaf, the highest concentration corresponds to 48 gle and the lower concentration to 6 gle. With a sprayer, a volume of 0.8 ml of one of the solutions was sprayed on green paper following the same method as described above for the test of pure JA. P. rapae butterflies were offered a two-choice situation, with one paper sprayed with GLS extracted from control plants and one paper with GLS from JA-treated plants. Performance of P. rapae Caterpillars The development of first instar caterpillars to pupae was observed on control and JA-treated plants. Control plants were sprayed with a 0.1% Tween 20 solution and JA-treated plants with a solution of 0.5 mM JA with 0.1% Tween 20. Thirty newly hatched P. rapae caterpillars were evenly distributed over two plants per treatment, 24 hr after treatment, and were placed in cages (67 × 50 × 75 cm) in a greenhouse compartment at 22–24°C and 50–70% RH. The plants were replaced with new plants twice a week, so that the maximum time between induction and larval feeding never exceeded 5 d. The number of days until pupation and pupal weight were recorded. Statistical Analyses Each individual butterfly female was subjected to a two-choice situation, in which most individuals oviposited on both control and JA-treated leaves. As the egg load differed between individuals, the number of eggs on each treatment per individual was treated as a paired sample. The oviposition data for P. rapae were normally distributed; therefore, they were analyzed with a paired t-test. The oviposition data for P. brassicae were not normally distributed, and therefore, analyzed with the nonparametric equivalent of the paired t-test, the Wilcoxon matched-pair signed-ranks test. The data on the developmental time of the caterpillars in the performance test were not normally distributed and were analyzed with a Mann–Whitney U test for differences between the treatments. Pupal weight was normally distributed and analyzed with an analysis of variance (ANOVA). Changes in GLS content were analyzed with a Mann–Whitney U test. Statistical analyses were performed with SPSS 11.0. Results Chemical Analysis Five GLS were detected in B. oleracea leaf surface samples: glucoiberin, sinigrin, 4-hydroxyglucobrassicin, glucobrassicin, and 4-methoxyglucobrassicin (Table 1). No significant difference was detected between JA-treated and control leaves for the total amount of GLS per cm2. The amounts of glucobrassicin, the most abundant glucosinolate in these samples, 4-methoxyglucobrassicin, and sinigrin, did not significantly differ between control and JA-treated leaves. The amounts of glucoiberin and 4-hydroxyglucobrassicin collected in the leaf surface extracts were significantly lower for JA-treated leaves compared to control leaves (Table 1). The same results were obtained when calculated for the GLS content expressed as nmol per mg dry weight (not shown). Table 1Glucosinolate content in surface extracts of Brassica oleracea leaves in pmol/cm2 for control and JA-treated plantsCompoundControl Treatment Median (range)a,bJA Treatment Median (range)a,cZPGlucoiberin10.7 (6.3–21.7)0 (0–6.0)−2.7130.007Sinigrin9.3 (0–23.5)4.2 (0–6.8)−1.7740.0764-Hydroxyglucobrassicin0.7 (0–6.1)0 (0–0)−2.2070.027Glucobrassicin34.9 (18.4–98.2)88.6 (38.6–132.5)−1.4790.1394-Methoxyglucobrassicin0.5 (0–2.5)0 (0–1.3)−1.3700.171Total amount of glucosinolates64.1 (27.0–155.7)90.0 (42.9–138.8)−0.5630.573aInterquartile range from first to third quartilebN = 10cN = 11 Herbivore Oviposition Preference: Surface Application of JA As P. brassicae lays its eggs in batches, both the number of egg batches and the number of eggs per leaf were counted. For the 1 mM JA treatment, the number of batches was significantly lower on JA-treated leaves than on control leaves (N = 36, Z = −2.628, P = 0.009, Wilcoxon matched-pair signed-ranks test), and the total number of eggs was significantly lower as well (N = 36, Z = −2.035, P = 0.042, Wilcoxon matched-pair signed-ranks test). For the 0.1-mM JA treatment, the result was similar (N = 27, batches: Z = −2.223, P = 0.026; eggs: Z = −2.138, P = 0.032, Wilcoxon matched-pair signed-ranks test) (Fig. 1). Experiments with 0.01 mM JA application did not show discrimination by the butterflies between the treated and control leaves (results not shown). Also, P. rapae butterflies significantly preferred to oviposit on control leaves compared to JA-treated leaves (Fig. 2). The leaves treated with the two highest concentrations of JA tested, 1 and 0.1 mM, were avoided in favor of the control leaves (paired t-test, respectively, t = 3.805, df = 23, P = 0.001 and t = 3.681, df = 23, P = 0.001). The lowest concentration of JA tested, i.e., 0.01 mM, did not affect the distribution of eggs over the leaves (t = −0.662, df = 23, P = 0.52, paired t-test). Fig. 1Pieris brassicae oviposition on control and JA-treated plants. Two concentrations of JA were tested against a control. The box represents the interquartile range from first to third quartile, the line across the box indicates the median; asterisks indicate statistical differences between the preference for control and JA-treated plants (*P < 0.05, **P < 0.01, Wilcoxon matched-pair signed-ranks test). a Egg batches per female per leaf. b Eggs per female per leafFig. 2Pieris rapae oviposition preference (measured as the number of eggs per female per leaf) between control and JA-treated B. oleracea. Three concentrations of JA were tested against a control in 24 replicated experiments for each concentration. Mean numbers of eggs per female + SEM are given; asterisks indicate statistical differences between the preference for control and JA-treated plants (n.s. P > 0.05, **P < 0.01, paired t-test) Systemic Uptake of JA In the experiment with P. rapae that employed systemic uptake of JA through the petiole, the same result was obtained: the number of eggs on the JA-treated leaves (10.0 ± 1.57) was lower than on the control (19.12 ± 2.82) leaves (t = 3.976, df = 31, P < 0.001, paired t-test). Effect of Pure JA on Oviposition Preference When paper treated with sinigrin and JA was compared to paper with only sinigrin, there was no difference in the number of eggs the butterflies deposited on the two substrates (t = −0.438, df = 26, P = 0.67, paired t-test) (Fig. 3). These results show that the observed effect of the JA treatment on herbivore oviposition behavior was due to induced changes in leaf tissue rather than to a direct repellent or deterrent effect of JA itself. Fig. 3Oviposition of P. rapae on green paper sprayed with sinigrin plus JA and green paper with sinigrin only. Mean number of eggs per female per paper leaf + SEM (n.s. P > 0.05, paired t-test, N = 27) Bioassays with Purified Glucosinolate Fractions The butterflies did not discriminate between the two GLS fractions (Table 2), the number of eggs on paper with the GLS from control plants, and the number of eggs on paper with GLS from JA-treated plants was not different for either concentration (concentration 6 gle: Z = −1.514, N = 20, P = 0.130, Wilcoxon matched-pair signed-ranks test; 48 gle: t = −0.523, df = 19, P = 0.607, paired t-test) (Fig. 4). Table 2Glucosinolate (GLS) content in fractions from leaf surface extracts from control and JA-treated plants in pmol/cm2, Z and P-values of Mann–Whitney U testGlucosinolateGLS from control plantGLS from JA-treated plantGlucoiberin4.93.9Sinigrin3.15.14-HydroxyglucobrassicinNot detectedNot detectedGlucobrassicin16.081.04-MethoxyglucobrassicinNot detected0.8Fig. 4Oviposition of P. rapae on green paper sprayed with purified GLS-fractions from leaf surface extracts of control and JA-treated plants. Mean number of eggs per female per paper leaf + SEM (n.s. P > 0.05, paired t-test, N = 20) Performance of P. rapae Caterpillars About two-thirds of the caterpillars survived until pupation, and a similar number of caterpillars reached the pupal stage on both treatments, 19 on control and 18 on JA-treated plants (Fig. 5). The caterpillars on JA-treated plants pupated on average after 15 d, while the caterpillars on the control plants pupated significantly sooner, on average after 13 d (Mann–Whitney U, Z = −4.071, P < 0.001). The average pupal weight on the control plants, 165 ± 3.4 mg, was similar to that on the JA-treated plants, 158 ± 3.3 mg (ANOVA, F = 2.665, df = 1, P = 0.112). Fig. 5Development time of P. rapae caterpillars from hatching until pupation on control [open squares (□)] and JA-treated [closed triangles (▴)] plants. Cumulative number of pupae per treatment Discussion Our data show that JA treatment of Brussels sprouts leaves reduces the acceptance of leaves for oviposition by P. rapae and P. brassicae in a similar way. Treatment of leaves with 0.1 or 1 mM JA reduced the proportion of eggs the butterflies laid on these leaves. A concentration of 0.01 mM JA did not change oviposition preference. The former concentrations are comparable to the concentrations of JA or MeJA that were applied to several plant species in other studies and that reduced development of Spodoptera exigua, Trichoplusia ni, Manduca sexta, thrips, and aphids (Thaler et al., 1996; Avdiushko et al., 1997; Van Dam et al., 2000; Omer et al., 2001), and abundance of Manduca quinquemaculata, S. exigua, thrips, and flea beetles in the field (Kessler and Baldwin, 2001; Thaler et al., 2001). For cabbage plants, Lu et al. (2004) found inducible resistance in a susceptible Brassica species (Chinese cabbage, Brassica campestris L.) and induced susceptibility in a resistant Brassica species (common cabbage, B. oleracea) for Plutella xylostella L. To exclude that JA itself caused the above effect, we tested the phytohormone on an inert substrate and studied two different application methods to the leaf material. We considered this an essential control that is lacking in other studies. The results of these experiments show that it was not JA itself that caused the difference in oviposition preference between control and JA-treated leaves, thus providing proof that processes in the plant induced by the JA treatment changed the acceptability of the leaves. It has also been reported previously that development of cabbage looper or tobacco hornworm larvae is not affected when MeJA is added to an artificial diet, but it is retarded when MeJA was applied to cabbage or tobacco plants (Avdiushko et al., 1997). In leaf surface extracts of JA-treated and untreated Brussels sprouts, we found five glucosinolates. After JA application, glucobrassicin, the major glucosinolate in the B. oleracea cultivar we used, occurred at a level twice as high as in control plants, and glucoiberin and 4-hydroxyglucobrassicin concentrations decreased. The total glucosinolate content did not change significantly. However, most other studies on glucosinolate content in Brassicaceous plants after induction by JA- or MeJA-treatment or insect attack have reported an increase in glucosinolates, although there is substantial variation among different species, or even genotypes, and type of induction (Bodnaryk, 1994; Cipollini and Sipe, 2001; Mikkelsen et al., 2003; Mewis et al., 2005). Moreover, glucosinolates may not be evenly distributed throughout the leaf. We measured glucosinolate content in a surface extract after 24 hr, whereas most studies have measured glucosinolate content in whole leaf extracts and after a longer induction time. Recently, Reifenrath et al. (2005) postulated that the wax layer does not contain glucosinolates, and the polar glucosinolates that are found by using the solvent extraction method are washed from the inner leaf to the outside through the stomata. Nevertheless, we chose a surface extraction method because the butterflies retrieve chemosensory information from the leaf surface, as they do not damage the leaf before ovipositing. Surface extracts are thus likely to give a better reflection of the chemosensory information used than whole leaf extracts. Both butterfly species distinguish between induced and non-induced leaves, most likely based on chemical differences, as JA-induced leaves do not display herbivore presence or damage. The different levels of two out of five glucosinolates in the surface extracts may provide a chemosensory basis for the oviposition preference observed, although the isolated GLS from the two treatments yielded no differences in acceptance of the paper for oviposition. While the isolated GLS on paper stimulated oviposition behavior, they appear not to be the main cue to discriminate between the JA-induced and non-induced cabbage plants. We did not quantify other chemicals, stimulants, deterrents, or precursors that might mediate preference behavior, such as isothiocyanates, terpenoids, other glycosides, or amino acids (Huang et al., 1993; Renwick and Chew, 1994; Soldaat et al., 1996; Agrawal and Kurashige, 2003). Both P. rapae and P. brassicae can perceive a broad range of chemicals (Van Loon et al., 1992b; Hern et al., 1996). Electroantennogram responses to a range of plant volatiles are similar for both species (Van Loon et al., 1992b), although host plant selection by Pieris butterflies appears largely based on contact chemoreception rather than olfaction (Renwick and Chew, 1994). Host plant selection is suggested to depend on a balance of stimulants and deterrents and not just on the detection of presence or absence of particular compounds (Huang et al., 1993; Bruce et al., 2005). Therefore, glucosinolates, in combination with other stimulants or deterrents, may determine the acceptance of a host plant by the butterflies. The octadecanoid pathway, in which JA is a key molecule, is involved in induction of synomones in response to oviposition and to herbivore damage (Meiners and Hilker, 2000; Dicke and Van Poecke, 2002; Hilker and Meiners, 2006; Mumm and Hilker, 2006). JA treatment of plants results in emission of synomones that attract natural enemies like predatory mites and parasitoids (Dicke et al., 1999; Hilker and Meiners, 2002; Van Poecke and Dicke, 2002; Hilker and Meiners, 2006). This attraction results in a higher natural enemy density around damaged plants, and therefore, it is advantageous for the herbivores to avoid oviposition on induced plants. Moreover, intact plants lack competitors, either intra- or interspecific. Herbivores may use induced plant cues to detect the presence or absence of other herbivores, especially because plant cues are, although less reliable, often easier to detect than cues from the herbivores themselves (Vet and Dicke, 1992). Furthermore, induced plants may affect herbivores directly by influencing the performance of their offspring. For P. rapae, the development time differed between caterpillars feeding on JA-induced and caterpillars feeding on non-induced plants. Development of the caterpillars to pupae took longer on the induced plants, which exposes them to natural enemies for a longer time and gives them a disadvantage in the competition for resources with other herbivores. These results comply with those of Agrawal and Kurashige (2003), who showed that growth of P. rapae larvae was reduced on herbivore-induced Brassicaceae. In summary: (1) JA treatment of B. oleracea results in avoidance of host plants by the two Pieris butterflies; (2) the related gregarious and solitary butterfly species tested responded in a similar fashion to JA-treated plants; (3) JA treatment reduced the contents of two out of five glucosinolates in leaf surface extracts of Brussels sprouts; and (4) the purified GLS fractions could not explain the observed avoidance behavior. The results indicate that JA-induced infochemicals play an important role in host plant selection behavior of these butterflies; however, the phytochemicals involved still have to be elucidated.

Pediatr_Nephrol-4-1-2335295

Improved renal survival in Japanese children with IgA nephropathy

Since the beginning of the 1990s, Japanese medical practitioners have extensively prescribed angiotensin-converting enzyme (ACE) inhibitors for children with mild IgA nephropathy (IgA-N) and steriods for those with severe IgA-N. We have performed a retrospective cohort study to clarify whether the long-term outcome has improved in Japanese children with IgA-N. Renal survival was defined as the time from onset to end-stage renal disease (ESRD). We divided the study period into two time periods based on the occurrence of the initial renal biopsy:1976–1989 and 1990–2004. Actuarial survivals were calculated by Kaplan–Meier method, and comparisons were made with the logrank test. The Cox proportional hazard model was used for multivariate analysis. Between 1976 and 2004, 500 children were diagnosed as having IgA-N in our hospitals. The actuarial renal survival from the time of apparent disease onset was 96.4% at 10 years, 84.5% at 15 years and 73.9% at 20 years. Renal survival in the 1990–2004 period was significantly better than that in 1976–1989 (p = 0.008), and a marked improvement in renal survival in patients with severe IgA-N was also observed (p = 0.0003). Multivariate analysis indicated that diagnosis year was a significant factor for ESRD-free survival independently of baseline characteristics. The results of this study show that there has been an improvement in terms of renal survival in Japanese children with IgA-N. Introduction IgA nephropathy (IgA-N) is the most common primary glomerulonephritis worldwide. It was initially considered to be a benign disease with a favorable prognosis, but data from long-term follow-up studies subsequently revealed that the disease progressed to renal failure in 20–50% of adult patients [1, 2]. Although there has been a prevailing belief that the prognosis of IgA-N is more benign in children, the results of more recent studies do not support this [3]. At the beginning of the 1990s, a strict policy of treatment for IgA-N was adopted by the majority of physicians [4]; this consisted of the extensive use of angiotensin-converting enzyme inhibitors (ACEIs) as an almost universally used therapy for IgA-N, and other therapies, including corticosteroids, for more severe IgA-N [4]. This approach has also been adopted in Japan since the beginning of the 1990s, with ACEIs generally prescribed for children with mild IgA-N showing focal mesangial proliferation and steroids prescribed for children with severe IgA-N showing diffuse mesangial proliferation. Combined therapy with prednisolone, azathioprine, heparin–warfarin and dipyridamole for 2 years in patients with severe IgA-N showing diffuse mesangial proliferation was started at the Kobe University and Wakayama Medical University hospitals in 1990 [5, 6]. At present, the effect of this therapy on long-term renal survival in children with IgA-N remains unknown as there have been few studies on outcome in sufficiently large cohorts of pediatric patients [4]. In the study reported here, we investigated data from 500 children with IgA-N to determine whether the long-term outcome has improved in Japanese children with IgA-N. Methods Patients The medical history of children under the age of 20 years who underwent routine renal biopsies at Kobe University and Wakayama Medical University hospitals between January 1976 and December 2004 were analyzed retrospectively. Clinical data and follow-up information were obtained from the medical records, and if needed, further complementary information was obtained by telephone contact with the patients, family members and or physicians. A diagnosis of IgA-N was based on the presence of IgA as the sole or predominant immunoglobulin in the glomerular mesangium in the absence of systemic disease [7]. Diffuse or focal mesangial proliferation was defined on the basis of World Health Organization criteria [8]: diffuse mesangial proliferation was defined as more than 80% of glomeruli showing moderate or severe mesangial cell proliferation, i.e. more than three cells per peripheral mesangial area; focal mesangial proliferation was defined as less than 80% of glomeruli showing moderate or severe mesangial cell proliferation. All children with IgA-N were diagnosed by one investigator (N.Y.) using the same criteria during the entire study period. Renal biopsies were performed in children with persistent proteinuria (early morning urine protein/creatinine ratio ≥0.2g/g) with or without hematuria. Renal biopsy criteria were not changed during the entire study period. Renal survival was defined as the period from the time of apparent disease onset to end-stage renal disease (ESRD) requiring renal replacement therapy. Given the change of policy for treating IgA-N patients in the early 1990s, we divided the study period when the initial renal biopsy had been performed into two time periods: 1976–1989 (early period) and 1990–2004 (late period). Treatment Since we started a strict policy of treatment for IgA-N at the beginning of the 1990s, we investigated the initial treatment profile of children with IgA-N. After a pilot study period, we started the first randomized controlled trial (RCT) by the Japanese Pediatric IgA Nephropathy Treatment Study Group (JPIGANTS) for treatment of children with severe IgA-N showing diffuse mesangial proliferation in 1990 [5]. We started the second RCT by the JPIGANTS for treatment of children with severe IgA-N in 1994 [6]. We started the third RCT by the JPIGANTS for treatment of children with severe IgA-N in 2001. The details of each treatment are shown in Table 1. Table 1Treatment in three randomized controlled trials for childhood IgA nephropathy with diffuse mesangial proliferationPeriod1990–1993 (first RCT)1994–1998 (second RCT)2001–present (third RCT)Group 1PrednisolonePrednisolone (as in group 1 in the first period)Prednisolone  2 mg/kg/day, 4 weeks  2 mg/kg/day, 4 weeks  2 mg/kg/2 days, 4 weeks  2 mg/kg/2 days, 4 weeks  1.5 mg/kg/2 days, 4 weeks  1.5 mg/kg/2 days, 4 weeks  1 mg/kg/2 days, 21 months  1 mg/kg/2 days, 9 months  0.5 mg/kg/2 days, 12 monthsAzathioprineAzathioprine (as in group 1 in the first period)No azathioprine  2 mg/kg/day, 24 monthsDipyridamoleDipyridamole (as in group 1 in the first period)Dipyridamole  5 mg/kg/day, 24 months  6–7 mg/kg/day, 24 monthsHeparinNo heparin  APTT 60 s, 28 daysWarfarinWarfarinWarfarin  TT 30–50%, 23 months  TT 30–50%, 24 months  TT 20–50%, 24 monthsMizoribine  4 mg/kg/day, 24 monthsGroup 2Heparin-WarfarinDipyridamole (as in group 1)Prednisolone (as in group 1)Prednisolone (as in group 1)Mizoribine (as in group 1)TT, Thrombotest; APTT, activated partial thromboplastin time; RCT, randomized controlled trial For the treatment of children with mild IgA-N showing focal mesangial proliferation we started a RCT by the JPIGANTS in 1990. During this trial (1990–1993) about half of the children with mild IgA-N in our hospitals received Sairei-to, a Chinese herb, for 24 months, and the other half received no medication. We began to use ACEIs for treatment of mild IgA-N with focal mesangial proliferation as the first choice from the beginning of 2000. Statistical analyses The results were analyzed with SAS ver. 9.1.2 (SAS Institute Japan Ltd., Tokyo, Japan). The associations between categorical variables were examined using the Fisher’s exact test. Continuous characteristics of the groups were compared with the Mann-Whitney U-test. Actuarial survival curves were calculated according to the Kaplan-Meier method [9], and comparisons were made with the logrank test [10]. Univariate and multivariate analyses were performed to assess the difference between the two periods in terms of renal survival. For multivariate analysis, we used the Cox proportional hazard model [11]. A two-tailed p value of less than 0.05 was taken as the level of significance. Results Patients Between 1976 and 2004, 1759 children underwent a first renal biopsy examination at the Kobe University and Wakayama Medical University hospitals. Among these, 500 Japanese children (28.4%; 279 boys and 221 girls) were diagnosed as having IgA-N: 219 in 1976–1989 and 281 in 1990–2004. There was no evident change in the number of patients per year between the early (1976–1989) and late (1990–2004) periods. The median patient age at diagnosis was 10.9 years (range 2.5–19.6 years), and the median follow-up period for the patients overall was 5.9 years (range 1.3–20.5 years). The baseline characteristics of children with IgA-N are shown in Table 2. There were significant differences in some characteristics between the two periods. Age at diagnosis had a tendency to be higher in the late group, while the ratio of asymptomatic proteinuria and hematuria at initial presentation was significantly higher in the late group. The ratio of heavy proteinuria (≥ 1 g/m2 per day) at diagnosis was significantly higher in the early period, but the ratio of patients showing diffuse mesangial proliferation was higher in the late period. Based on these data, we concluded that there was no evident difference in disease severity between the two periods. Table 2Baseline characteristicsCharacteristicNumber of patients (%)pTotal (n = 500)1976–1989 (n = 219)1990–2004 (n = 281)Sex (M/F)279/221132/87147/1340.08Age at diagnosis, year, median [range]10.9 [2.5–19.6]10.1 [3.4–16.8]11.6 [2.5–19.6]< 0.001Initial presentationAsymptomatic proteinuria and hematuria384 (76.8%)150 (68.5%)234 (83.3%)< 0.001Macroscopic hematuria93 (18.6%)60 (27.4%)33 (11.7%)< 0.001Edema23 (4.6%)9 (4.1%)14 (5.0%)0.67Proteinuria at diagnosis (g/m2/day)  <1361 (72.2%)146 (66.7%)215 (76.5%)0.008  ≥1139 (27.8%)73 (33.3%)66 (23 .5%)Estimate creatinine clearance at diagnosis (ml/min per 1.73 m2)  <6013 (2.6%)3 (1.4%)10 (3.6%)0.16  ≥60487 (97.4%)216 (98.6%)271 (96.4%)Renal biopsy at diagnosis   Diffuse mesangial proliferation171 (34.2%)63 (28.8%)108 (38.4%)0.03   Focal mesangial proliferation329 (65.8%)156 (71.2%)173 (61.6%) Treatment The initial treatment for IgA-N showing focal mesangial proliferation or diffuse mesangial proliferation is given in detail in Table 3. There was a clear change of treatment for IgA-N between the early and late periods. This is particularly true in terms of ACEI use for focal mesangial proliferation and combined therapies for diffuse mesangial proliferation, both of which increased dramatically during the late period. Table 3Change of initial treatment for IgA nephropathy in Japanese childrenTreatmentFocal mesangial proliferationDiffuse mesangial proliferation1976–1989 (n = 156)1990–2004 (n = 173)1976–1989 (n = 63)1990–2004 (n = 108)No treatment96 (61.6%)23 (13.2%)19 (30.2%)1 (0.9%)Antiplatelet and/or anticoagulant35 (22.4%)2 (1.2%)14 (22.2%)7 (6.5%)Prednisolone, (± antiplatelet and/or anticoagulant)4 (2.6%)6 (3.5%)7 (11.1%)25 (23.1%)Prednisolone + immunosuppressant, (± antiplatelet and/or anticoagulant)7 (4.5%)8 (4.6%)19 (30.2%)74 (68.5%)Chinese herb (Sairei-to)14 (9.0%)46 (26.5%)4 (6.3%)0 (0.0%)ACEI and/or ARB0 (0.0%)88 (50.9%)0 (0.0%)1 (0.9%)ACEI, Angiotensin-converting enzyme inhibitor; ARB, angiotensin II receptor blocker Renal survival Among a total of 500 patients who had IgA-N therapy, the actuarial renal survival from the time of apparent disease onset was 96.4% at 10 years, 84.5% at 15 years and 73.9% at 20 years (Table 4). As shown in Fig. 1 and Table 4, in children diagnosed as having IgA-N in 1976–1989, the actuarial renal survival from the time of apparent disease onset was 94.0% at 10 years, 80.1% at 15 years and 70.1% at 20 years. In children diagnosed as having IgA-N in 1990–2004, the actuarial renal survival from the time of apparent disease onset was 98.8% at 10 years, 98.8% at 15 years (p = 0.008; Fig. 1, Table 4). Table 4Actuarial renal survival analysis of 500 children with IgA nephropathy Initial renal biopsy yearNumber of patients10-year renal survival15-year renal survival20-year renal survivalpaTotal1976–200450096.4%84.5%73.9%Total1976–198921994.0%80.1%70.1%0.0081990–200428198.8%98.8%–Diffuse mesangial proliferation1976–19896378.5%68.6%b–0.00031990–200410897.8%97.8%b–Focal mesangial proliferation1976–1989156100.0%97.7%–0.51990–2004173100.0%100.0%–ap values on the logrank testbThe 13-year survivalFig. 1Kaplan–Meier plot of renal survival stratified by the initial biopsy year for children with IgA nephropathy. 95% CI 95% Confidence interval Figures 2 and 3 show the long-term outcome for children in the two different mesangial proliferation groups. For children with severe IgA-N showing diffuse mesangial proliferation, both the 10- and 13-year renal survivals were 97.8% when the diagnosis was made in the period 1990–2004; when the diagnosis was made in the period 1976–1989, renal survivals were 78.5% and 68.6%, respectively (p = 0.0003; Fig. 2, Table 4). For children with mild IgA-N showing focal mesangial proliferation, both the 10- and 15-year renal survivals were 100.0% when diagnosis was made in 1990–2004, compared with 100.0% and 97.7%, respectively, in 1976–1989 (p = 0.5; Fig. 3, Table 4). Although we observed better renal survival in patients diagnosed in the period 1990–2004 than in the period 1976–1989 for children with IgA-N showing focal mesangial proliferation, the difference did not reach statistical significance. The children with IgA-N showing diffuse mesangial proliferation in 1990–2004 had excellent long-term renal survival. Fig. 2Kaplan-Meier plot of renal survival stratified by the initial biopsy year for children with severe IgA nephropathy showing diffuse mesangial proliferationFig. 3Kaplan-Meier plot of renal survival stratified by the initial biopsy year for children with mild IgA nephropathy showing focal mesangial proliferation Table 5 shows the results of the univariate analysis using the logrank test and the multivariate analysis using the Cox proportional hazard model of prognostic factors for ESRD-free survival. Mesangial proliferation degree (focal or diffuse), proteinuria at diagnosis (<1 or ≥1 g/m2 per day), estimate of creatinine clearance at diagnosis (≥60 or <60 ml/min per 1.73 m2) and initial renal biopsy (diagnosis) year (1976–1989 or 1990–2004) were included as factors for analyses. Mesangial proliferation degree and initial renal biopsy year were significant in both the univariate and the multivariate analysis. For children with IgA-N, the most influential prognostic variable was mesangial proliferation degree. Proteinuria at diagnosis was significant in the univariate but not in the multivariate analysis. These results from the multivariate analysis showed that initial renal biopsy year was a significant factor for ESRD-free survival independently of mesangial proliferation degree, proteinuria at diagnosis and estimate of creatinine clearance at diagnosis (hazard ratio = 0.08, 95% CI 0.004–0.43; Table 5). Table 5Univariate and multivariate analysis of the prognostic value of factors for end-stage renal disease-free survivalFactorUnivariateMultivariateHR95% CIpHR95% CIpMesangial proliferation focal; diffuse10.922.80–72.46<0.00110.272.42–70.750.001Proteinuria at diagnosis <1; ≥1 (g/m2/day)5.271.65–19.770.012.140.57–8.780.26CCl at diagnosis ≥ 60; < 60 (ml/min per m2)14.122.11–57.410.015.580.74–30.220.09Initial renal biopsy year 1976–1989; 1990–20040.140.01–0.740.020.080.004–0.430.002CI, Confidence interval; HR, hazard ratio; CCl, creatinine clearance Discussion Although this was a retrospective study, the data seem to provide unique and valuable information about IgA-N in children for several reasons. First, the study subjects were a complete non-selected cohort of patients with IgA-N from among all children who underwent first renal biopsy examinations at Kobe University and Wakayama Medical University hospitals between 1976 and 2004. Five hundred children with IgA-N participated in our study, which to our knowledge is the largest series used to investigate the outcome of IgA-N in children to date. Furthermore, these patients were diagnosed histologically by a single investigator (N.Y.) based on the same criteria during the whole study period. Renal biopsy criteria were also unchanged during the whole study period. Therefore, the cohort in the present study is considered to have been rigidly homogeneous in terms of the analysis of the disease outcome. One of the conditions that facilitated our study was the school screening program, which was started by the Japanese government in 1974. In Japan, all children between the ages of 6 and 18 years are screened annually, and those found to have urinary abnormalities are referred for further investigation. Thus, in general, treatment for IgA-N is started early in the course of disease because the duration of disease before treatment is short as a result of this school screening program [12]. Therefore, in this study we were able to observe the disease courses of patients during the initial renal biopsy period, 1976–2004, who underwent initial treatments for IgA-N at a relatively homogeneous stage of the disease. Accumulated experience indicates that long-term corticosteroid and/or immunosuppressive treatment during the severely progressive stage of the disease does not confer any benefit in adult patients [2]. In contrast, it is known that centers in countries with active urine screening programs are more likely to diagnose mild disease with a good prognosis, thus favorably influencing the overall outcome of the cohort [13]. Another favorable parameter of this study was the enforcement of consecutive systemic nationwide clinical trials of IgA-N. After a pilot study period, the first RCT for treatment of children with severe IgA-N showing diffuse mesangial proliferation was started by the JPIGANTS in 1990 [5]. To date, two RCTs of treatment for children with severe IgA-N showing diffuse mesangial proliferation by the JPIGANTS have been completed [5, 6], and the third RCT of treatment for children with severe IgA-N showing diffuse mesangial proliferation is currently being conducted by the same group. For the treatment of children with mild IgA-N and focal mesangial proliferation, a RCT was also started by the JPIGANTS in 1990. After completion of the RCT, we have conducted prospective clinical trials of treatment for children with mild IgA-N and focal mesangial proliferation. These prospective trials have enabled for us to clarify the optimum treatment for IgA-N and to collect precise information on patients with IgA-N. Because of the variable rate of progression to renal failure and the probable multifactorial pathogenesis of IgA-N, it is desirable to evaluate the effectiveness of any treatment by a prospective controlled trial. However, although the ultimate endpoint in any clinical trial of progressive IgA-N is the development of chronic renal insufficiency, most pediatric patients do not develop it during the study period [14]. Therefore, data from a properly analyzed long-term retrospective study may also be important to evaluate disease outcome and treatment effectiveness [4, 9–11]. Retrospective studies with accurate statistical evaluation using life-time analysis and multivariate survivorship analysis according to the Cox regression model are thought to work effectively together with well-designed RCTs. The primary purpose of this study was to clarify whether long-term outcome was improved in Japanese children with IgA-N retrospectively using adequate statistical methods. Although it seemed that there were no clinically significant differences in the baseline characteristics of the groups, there were some statistically significant differences in the baseline characteristics of the groups between the two periods when initial renal biopsies were performed. These different baseline characteristics may influence the final outcomes reported in this study. To compensate for this limitation, we used a multivariate survivorship analysis according to the Cox regression model. Since we have to adjust for differences of baseline characteristics, which were thought to be important prognostic factors [15], mesangial proliferation degree, proteinuria at diagnosis, estimate of creatinine clearance at diagnosis and the initial renal biopsy (diagnosis) year were included as factors for analyses. This multivariate analysis indicated that the initial renal biopsy year was a significant factor for ESRD-free survival after adjustment of these baseline characteristics. These data support an improvement of renal survival in Japanese children with IgA-N. There was a notable lack of the usual dominance of males over females in our study and, in addition, there was a difference between the 1976–1989 and the 1990–2004 periods, with more females being recruited in the second period. Although this difference was not statistically significant (p = 0.08), we decided that it be worthwhile discussing possible reasons for the differences in gender representation between the two periods. Therefore, we analyzed males and females separately. However, there was no gender-based factor in the results (data not shown). At the beginning of the 1990s, a strict policy of treatment was adopted by the majority of physicians in Japan. It is likely that therapeutic intervention early after the onset of clinically apparent disease will provide the best opportunity for improving the outcome of patients with IgA-N, thus reducing the number of patients who develop ESRD [5, 6]. This study showed a clear change in the treatment of IgA-N between the early and late groups (Table 3). In particular, the use of ACEIs for focal mesangial proliferation and combination therapies for diffuse mesangial proliferation increased dramatically in the late group. Corticosteroids have been widely used to treat moderate to severe IgA-N, particularly in pediatric patients. Some evidence has been obtained recently pertaining to the role of corticosteroids in the treatment of IgA-N [2, 13, 14, 16–18]. With regard to children and based on the results of two multicenter RCTs, we reported previously that the treatment of childhood IgA-N with diffuse mesangial proliferation using prednisolone, azathioprine, warfarin and dipyridamole for 2 years early in the course of disease reduced the severity of immunologic renal injury and prevented any increase in the percentage of sclerosed glomeruli [5, 6]. In contrast, in the first RCT, heparin–warfarin and dipyridamole treatments for 2 years did not reduce urinary protein excretion, serum IgA concentration and mesangial IgA deposition, and they did not prevent any increase of sclerosed glomeruli [5]. The results of the second RCT showed that the treatment of children with severe IgA-N using prednisolone alone for 2 years reduced the severity of immunologic renal injury, but did not prevent any further increase of glomerular sclerosis [6]. Therefore, treatment with the combination therapy using prednisolone and immunosuppressant, such as azathioprine, may be better than the prednisolone monotherapy for patients with severe IgA-N. We found an improvement of long-term renal survival in Japanese children with IgA-N (Table 4, Figs. 1 and 2). Renal survival in 1990–2004 was significantly better than in 1976–1989 (Fig. 1), and a marked improvement of renal survival in IgA-N showing diffuse mesangial proliferation was observed over time (Fig. 2). This improvement may be related to the 2-year therapy, including corticosteroids, for all patients with IgA-N showing diffuse mesangial proliferation as a treatment policy, although in principle the effectiveness of any treatment can only be evaluated properly by a controlled trial. It is conceivable that the outcome in the period 1976–1989 may have reflected the natural course of disease, whereas that in 1990–2004 may have reflected modification of the disease course by the treatments. The median follow-up period of the patients overall was 5.9 years (range 1.3–20.5 years). Although this is a considerably long period, it may not be sufficiently long for the follow-up of IgA-N, which has a slow progressive course. Since the elapsed time from apparent disease onset to ESRD was used to evaluate the disease outcome, chronic renal insufficiency before ESRD was not considered in our study. Thus, we do not know whether the observed improvement of outcome means a complete cure of the disease or merely a delay of the disease course. Further follow-up of the cohort for a long period is therefore important. One aspect which we should consider in our study is the possibility that at least some of the differences in outcome between the early and late groups may be due to general changes in care that have occurred over time. Such changes are frequently not recognized, such as a gradually greater awareness of managing marginally raised blood pressure, among others. In conclusion, this study has demonstrated a substantial improvement of long-term renal survival in Japanese children with IgA-N. On the basis of the results of this study we were unable to provide the reason for this improved outcome directly; however, it is likely that adequate management using a strict policy of treatment may have been responsible.

Bioinformation-2-2-2174421

DoD2007: 1082 molecular biology databases

Molecular biology databases are an integral part of biological research. To date, many databases were established with varied options to access associated biological data. Depending on the data being annotated, some are architecturally similar while others are specialized. In order to provide a partial solution to data integration, we report Database of Databases (DoD2007), constructed using html and javascript. The database has a web-based user interface with simple global search, specific database search, keyword help as well as links to abstracts, full-text and database home pages. Majority of data were derived form Nucleic Acids Research database issue and other published resources. The current release includes 15 categories with updated descriptions and links to 1082 databases, of which, 209 are new entries. New databases included in this issue are represented with ‘+’ sign before the name and a ‘*’ symbol provided for those that remained silent. Background Molecular Biology databases have become an integral part of scientific research. They are widely used to understand the underlying mechanism of genomes, expression patterns, bimolecular interactions, metabolism, understanding evolutionary relationships etc. as well as providing knowledge that helps to examine specific state of a disease or condition and assist in drug discovery and development. This sort of biological knowledge is disseminated to a variety of scientific researchers through specialized databases made possible through internet technologies, software's and tools. As more and more genomes are being sequenced and annotated, huge amount of data are accumulating. [1] Biological databases designed would cater to meet the needs of the scientific community. But depending on the data being annotated, some are architecturally similar while the others are specialized. The challenges to develop an integrated system are due to several factors such as variety and amount of data available and data heterogeneity in different sources. [2] Therefore, data integration has proved problematic and to provide a partial solution, we report an update on collection of molecular biology databases with a search interface, Database of Databases (DoD2007). The latest edition, DoD2007 was compiled with similar capabilities [3,4] but with additional features such as a global search for all databases and links to database entries from category list, added patents database as a new category and one sub-category each in nucleotide, Genomics database and two in Protein Sequence database, respectively. Methodology DoD2007 was updated from the reported database issue of Nucleic Acids Research [5] and various journal sources. The 14 categories reported in our previous update [4] remain unchanged except a new category; ‘patents db’ has been incorporated in DoD2007 update, keeping in view the importance of patent rights in scientific discipline. Currently, this category includes two databases, patome [6] and DNA patents database. [7] DoD2007 was constructed using html and javascript. The modes of access of databases listed in DoD2007 vary depending on the content, format and access methods. Majority of the databases are provided with direct access to search data from DoD2007. Some databases require multiple search items while some others require sequence or annotation to be pasted in the search box and therefore they are provided with a direct link to the database home page. Over time, DoD2007 has come to enhance the listed molecular biology databases as its use by researchers, educators and students in diverse disciplines has expanded. The recent release DoD2007 contains three distinct newly developed functionalities, which are outlined below. First, we have developed a user interface to search databases globally. In other words, a global search can be utilized to search the categories reported in DoD2007, database names and other pages located in the database. Keywords used to search the database can be general or specific and depends on the terms indexed for search. Second, an organized category list was created in a separate page. The availability of these databases as a list is intended to promote better understanding of database segregation used to generate categories and to quickly recognize the new ones with those that are inaccessible. These databases have links to the database entries because a link back to the main page makes it easy for both new and proficient users to conduct a variety of searches. However, some users may not realize the keywords that need to be used to search a database and even some users may show interest in the published resource. Hence, a keyword help and links to abstract or full-text articles were created in respective main pages for intuitive and efficient presentation. Finally, a number of sub-categories are created such as ‘Operons’ and ‘Comparative genomics’ in ‘nucleotide db’ and ‘Genomics db’ respectively along with two sub-categories viz. ‘Protein Interactions’ and ‘Amino Acid Repeats’ in ‘Protein Sequence db’. Features of DoD2007 DoD2007 is a unique, categorized and easy-to-use interface for all molecular biology databases. New databases in various categories were appended and the number of databases in DOD2007 reached up to 1082 that are 209 databases ahead from the previous issue. Major additions, 49 new entries, were reported in Genomics database. However, the list is not complete. A brief description of databases under all categories was given as a pop-up window with a provision to view full-text of the published article and a direct link to the database home page. The search option under ‘link to database’ leads to the respective home pages of the database. Boolean search items are not allowed in the DoD2007 but spaces between search strings enables the search. During the search, results are displayed in a separate window as this helps in further database scan without losing the home page. DoD2007 facilitates integrated database search to scan a number of relevant databases from respective categories. A few databases incorporated in DoD2007 require user log-in or restricted access and a few are inaccessible or discontinued. To distinguish such databases, new entries in DoD2007 are provided with a ‘+’ sign and inaccessible ones are recognized by a ‘*’ sign before the name. The current list of database categories and the number of entries in each category are given in table 1 (supplementary material). An image of the database is given in figure 1. Since its first appearance in the year 2005, DoD constituted 719 molecular biology databases and the year 2005 resulted in an addition of 154 entries making up to 873 databases (DoD2006) and the year 2006 has 209 new entries being appended in DoD2007. On the other hand, nearly 4 percent of databases are inaccessible such as HLA Ligand/Motif database [8] of Immunological database or discontinued while few others remained silent such as European rRNA Database [9] and Small RNA Database [10] of RNA sequence database etc. On an average, there is a continuous increase in the number of databases which specifies the importance of such databases to biological community. Therefore, in order to account for the rise in number of databases, a graph as shown in figure 2 was plotted for each category since year 2004. The displayed graph showed a substantial increase in Nucleotide, Protein and Genomics databases respectively. Utility DoD2007 provides the updated descriptions and links to existing and new databases that serves as an interface and user-friendly access to molecular biology scientific community. Number of inaccessible databases reported in the years 2005 and 2006 remain unchanged in DoD2007 so as to enable the users to know the type of database that once existed. Conclusion DoD2007, a freely available web-based database resource enabling ease of access, serves as a general reference source both for community of researchers working in molecular biology and educators who deal with particular databases of their interest. Brief description about the databases and keyword help makes the users familiar with the contents of the database, respective keywords and database links. The database shall be updated on a yearly basis. Supplementary material Data 1

Eur_J_Epidemiol-3-1-2071967

The Rotterdam Study: objectives and design update

The Rotterdam Study is a prospective cohort study ongoing since 1990 in the city of Rotterdam in the Netherlands. The study targets cardiovascular, neurological, ophthalmological and endocrine diseases. As of 2008 about 15,000 subjects aged 45 years or over comprise the Rotterdam Study cohort. The findings of the Rotterdam Study have been presented in some 600 research articles and reports (see http://www.epib.nl/rotterdamstudy). This article gives the reasons for the study and its design. It also presents a summary of the major findings and an update of the objectives and methods. Introduction The Rotterdam Study was designed in the mid-1980s as a response to the demographic changes that were leading to an increase of the proportion of elderly people in most populations [1]. It was clear that this would produce a strong rise in elderly people living with diseases, as most diseases cluster at the end of life, and that to discover the causes of diseases in the elderly one would have to study risk factors of those diseases [2]. A major approach to finding causes is the prospective follow-up study, which has proven quite effective in finding causes of heart disease and cancer. Remarkably, in the 1980s there were hardly any follow-up studies focussing on the elderly. The design of the Rotterdam Study The basic design of the study was straight-forward: a prospective cohort study among 7,983 persons living in the well-defined Ommoord district in the city of Rotterdam in the Netherlands (78% of 10,215 invitees). They were all 55 years of age or over and the oldest participant at the start was 106 years [3]. The study started with a pilot phase in the second half of 1989. From January 1990 onwards participants were recruited for the Rotterdam Study. Figure 1 gives a diagram of the various cycles of the study. Fig. 1Diagram of examination cycles of the Rotterdam Study (RS). RS1 refers to the baseline examination of the original cohort (pilot phase 07/1989-12/1989; cohort recruitment 01/1990-09/1993). RS2, RS3 and RS4 refer to re-examination of the original cohort members. RSPlus1 refers to the extension of the cohort with persons in the study district that became 55 years since the start of the study or those of 55 years or over that migrated into the study district. RSPlus2 refers to the re-examination of the extension cohort. RSYoung1 refers to the baseline examination of all persons aged 45 and over living in the study district that had not been examined (i.e. mainly comprising those aged 45–55 years) In 1999, 3,011 participants (out of 4,472 invitees) who had become 55 years of age or moved into the study district since the start of the study were added to the cohort. In 2006, a further extension of the cohort was initiated in which about 6,000 subjects aged 45–54 years, living in the Ommoord district, were invited (expected number of participants about 4,000). By the end of 2008, the Rotterdam Study is therefore expected to comprise about 15,000 subjects aged 45 years or over. The participants were all examined in some detail at baseline. They were interviewed at home (2 h) and then had an extensive set of examinations (a total of 5 h) in a specially built research facility in the centre of their district. These examinations focussed on possible causes of invalidating diseases in the elderly in a clinically state-of-the-art manner, as far as the circumstances allowed. The emphasis was put on imaging (of heart, blood vessels, eyes, skeleton and later brain) and on collecting bodily fluids that enabled further in-depth molecular and genetic analyses. These examinations were repeated every 3–4 years in characteristics that could change over time. And so we had examination cycles from 1990 to 1993, from 1993 to 1995, from 1997 to 1999, from 2000 to 2001, from 2002 to 2004, from 2004 to 2005 and from 2006 to 2008 (Fig. 1). The participants in the Rotterdam Study are followed for a variety of diseases that are frequent in the elderly (and many are also in the not so elderly): coronary heart disease, heart failure and stroke, Parkinson disease, Alzheimer disease and other dementias, depression and anxiety disorders, macular degeneration and glaucoma, diabetes mellitus and osteoporosis. The Rotterdam Study has been approved by the institutional review board (Medical Ethics Committee) of the Erasmus Medical Center and by the review board of the Netherlands Ministry of Health, Welfare and Sports. The approval has been renewed every 5 years. Separate approval has been obtained for the introduction of major new elements in the study (e.g. MRI investigations). In the remainder of this article the objectives and major findings will be presented with an update of the methods for cardiovascular diseases, neurological diseases, ophthalmologic diseases, psychiatric diseases, endocrine diseases, as well as for genetic and for pharmaco-epidemiologic studies. Cardiovascular diseases Objectives Research on the epidemiology of cardiovascular diseases focuses on three primary areas of interest: studies on risk factors for atherosclerosis and coronary heart disease, studies on the detection of subjects at high risk of coronary heart disease, and studies on cardiovascular conditions in older age. Two groups of putative risk factors for atherosclerosis and coronary heart disease are included. The first are endocrine factors, including estrogens and androgens, insulin and insulin-like growth hormone I, and thyroid gland and adrenal gland hormones. The second group contains factors involved in haemostasis, inflammation and endothelial function. Research also focuses on genetic factors in these areas in relation to risk of coronary heart disease. The ability of classical cardiovascular risk factors to identify subjects at high risk of coronary heart disease is limited. Risk stratification may be improved when based on the presence of atherosclerosis. To this end, repeated measurements of non-coronary atherosclerosis and measurements of coronary, carotid and aortic arch calcification have been included in the study. Another line of research focuses on cardiovascular diseases in the elderly that are in large part the consequence of ischemic heart disease, like heart failure and atrial fibrillation. An important topic in this area is the early diagnosis of heart failure using echocardiographic assessment of asymptomatic systolic and diastolic dysfunction of the left ventricle. Atrial fibrillation is another major chronic condition frequent at older age. Examination of the determinants and prognosis of atrial fibrillation is part of this research line. Major findings Findings on determinants of atherosclerosis and coronary heart disease include the relation between hypothyroidism and myocardial infarction [4] and the association of genetic variation in the estrogen receptor alpha gene with risk of myocardial infarction in women [5]. The study demonstrated that high levels of CRP [6] and lipoprotein-associated phospholipase A2 activity [7] were associated with risk of coronary heart disease and stroke. Genetic findings in this area include a positive association of coronary heart disease with genetic variation in the complement factor H gene [8] but not with variation in the CRP gene [6]. The study showed that carotid intima-media thickness and plaques measured by ultrasound, the ankle-arm index, and aortic calcification detected by X-ray are independent predictors of risk of coronary heart disease [9]. The study also showed that vessel wall stiffness, as measured by aortic pulse wave velocity (PWV), improved prediction of cardiovascular disease beyond the classical risk factors [10]. The Rotterdam Study was the first to publish results from population-based research on the predictive value for coronary heart disease of coronary calcification assessed by electron-beam CT [11]. The study enabled accurate assessment of the incidence and lifetime risk of heart failure and atrial fibrillation in an elderly population [12, 13]. It was shown that atherosclerosis and inflammation are also involved in these conditions [14, 15]. Methods update Repeated measures of non-coronary atherosclerosis included carotid intima-media thickness and plaques by ultrasound, the ankle-arm index and aortic calcification by X-ray [9]. Electron-beam CT and multi-detector CT were used to accurately quantify calcification in the coronary, aortic arch and carotid arteries [11]. In the second additional cohort of the Rotterdam Study (RS Young1, see Fig. 1), the measurement of plaque vulnerability with high-resolution MRI of the carotid arteries will be added. Other outcome measures include electrocardiography, echocardiography and measures of vessel wall stiffness (PWV and carotid distensibility) [10]. Determinants were assessed by physical examinations, collection of blood samples, and by questionnaires and interview. The role of genetic factors is studied using the candidate gene approach and more recently the genome wide association study. Clinical cardiovascular outcomes are collected during our continuous follow-up and include non-fatal myocardial infarction and cardiac death, revascularizations, heart failure and atrial fibrillation. The methods of diagnosis of these outcomes have been described in previous papers [6, 12, 13]. Neurological diseases Objectives Neuroepidemiologic research in the Rotterdam Study focuses on the frequency, etiology and early recognition of the most frequent neurologic diseases in the elderly, including dementia (in particular Alzheimer disease), Parkinson disease and stroke. In neurodegenerative and cerebrovascular disorders clinical symptoms typically become manifest late in the disease course, the occurrence of clinical disease does not reflect the underlying spectrum of disease-related pathology, and most of the clinical syndromes are etiologically heterogeneous. Therefore, an additional research focus is on the causes and consequences of pre-symptomatic brain pathology that can be assessed with non-invasive imaging modalities. Major findings Neurodegenerative and cerebrovascular diseases are highly frequent in the elderly. The prevalence increases from age 55 to 65 years to age 90 years and above from less than 1% to over 40% for dementia [16], from less than 0.5% to more than 4% for Parkinson disease [17], and from approximately 1% to nearly 10% for stroke. The incidence figures follow this pattern of a strong increase with age over the entire age range, with the age-specific incidence of dementia being identical for men and women at least until the age of 85 [18] but with men having a higher age-specific incidence of both stroke and Parkinson disease than women throughout the age range [19, 20]. Vascular pathology and vascular risk factors are associated with worse cognitive performance [21], which also translates in people with vascular pathology or risk factors for vascular disease having an increased risk of dementia, including Alzheimer disease [22]. Moreover, several life style factors are associated with the risk of dementia and Alzheimer disease [23–25], suggesting that onset of dementia may at least partly be delayed or prevented. Commonly used drugs may have a role in this [26]. The classical risk factors for stroke also predict risk of stroke in the Rotterdam Study [27]. More recently identified risk factors, including inflammatory markers, may be etiologically relevant but thus far add little to the identification of people at risk [28]. Possibly underlying this is that a large amount of stroke goes clinically undetected [29]. Nearly 20% of elderly people have at least one silent brain infarct, and thereby a nearly fourfold increased risk of clinical stroke, a more than doubled risk of dementia including Alzheimer disease, and an increased risk of depression [29]. Neuroimaging reveals that brain pathology is widespread [30] and can go clinically undetected for a long time. In addition to the silent infarcts, many apparently healthy elderly have ischemic changes in their cerebral white matter that are associated with an increased risk of dementia, stroke and depression [31]. Also brain atrophy, especially of the hippocampus, is already present years before onset of even the earliest sign of cognitive impairment or subjective complaints [32]. This emphasizes the need to shift the attention in etiologic research of neurodegenerative and cerebrovascular disease to the causes of pre-symptomatic and underlying brain changes. Methods update Assessment of dementia and Alzheimer disease In the baseline and follow-up examinations participants undergo an initial screen for dementia with the Mini Mental State Examination (MMSE) and the Geriatric Mental Schedule (GMS), followed by an examination and informant interview with the Cambridge Examination for Mental Disorders of the Elderly (CAMDEX) in screenpositives (MMSE <26 or GMS >0), and subsequent neurological, neuropsychological and neuroimaging examinations [16, 18]. Of subjects who cannot be reexamined in person, information is obtained from the GPs and the regional institute for outpatient mental health care. A consensus panel makes the final diagnoses in accordance with standard criteria (DSM-III-R criteria; NINCDS-ADRDA; NINDS-AIREN). Assessment of parkinsonism and Parkinson disease Participants are screened in the baseline and follow-up examinations for cardinal signs of parkinsonism (resting tremor, rigidity, bradykinesia or impaired postural reflexes). Persons with at least one sign present are examined with the Unified Parkinson Disease Rating Scale and a further neurologic exam. Parkinson disease is diagnosed if two or more cardinal signs are present in a subject not taking antiparkinsonian drugs, or if at least one sign has improved through medication, and when all causes of secondary parkinsonism (dementia, use of neuroleptics, cerebrovascular disease, multiple system atrophy or progressive supranuclear palsy) can be excluded [17, 19]. Assessment of stroke and stroke subtypes History of stroke at baseline was assessed through interview and verified in medical records. Putative incident strokes get identified through the linkage of the study database with files from general practitioners, the municipality, and nursing home physicians’ files, after which additional information (including brain imaging) is collected from hospital records. A panel discusses all potential strokes and subclassifies strokes into ischemic, hemorrhagic or unspecified [20, 28]. Assessment of cognitive function Global cognitive function is measured through the Mini Mental State Examination (MMSE) in all surveys. From the third survey onwards we added a 30 min test battery that was designed to assess executive function and memory function, and which includes a Stroop test, a Letter Digit Substitution Task, a Word Fluency Test, and a 15 words Word List Learning test. Rotterdam Scan Study: brain imaging within the Rotterdam Study In 1991, a random sample of 111 participants underwent axial T2-weighted magnetic resonance (MR) imaging to assess presence and severity of white matter lesions [33]. In 1995, a random sample of 563 non-demented participants underwent brain MR imaging in the context of the Rotterdam Scan Study. The scanning protocol included series of axial proton-density, T2-weighted and T1-weighted images, as well as a high-resolution 3D-HASTE sequence [31]. From August 2005 onwards, a dedicated 1.5 Tesla scanner is operational in the research centre of the Rotterdam Study, and brain imaging is performed in all study participants without contra-indications. The scanning protocol includes 4 high-resolution axial sequences (3D T1-weighted; 2D PD-weighted; 2D FLAIR; and 3D T2* GRE), 2D phase-contrast imaging, and diffusion tensor imaging (DTI). Psychiatric diseases Objectives The aim of the psychiatric research in the Rotterdam Study is to investigate the determinants, correlates and consequences of common psychiatric problems in the elderly. The focus has been on depressive disorders but anxiety disorders, sleep disturbances and complicated grief are also being studied. Major findings The one-month prevalence of both major and minor depression at baseline was 1.5%, the prevalence of anxiety disorder was 8.2%. We found evidence for the vascular depression hypothesis. More severe coronary and extra-coronary atherosclerosis were associated with a higher prevalence of depression, as were cerebral haemodynamic changes [34, 35]. However, we could not rule out that earlier depressive episodes may have contributed to the development of atherosclerosis. Moreover, our data did not support a specific symptom profile of vascular depression as previously defined [36]. Most of our studies of nutritional and metabolic factors provided no support for an etiological role in depression. Folate deficiency, homocysteine, fatty acid ratio (omega-6/omega-3) and vitamin E had little or no effect on the risk of depression after control for health and health related behaviour [37–39]. In contrast, low levels of vitamin B12 and high levels of inflammatory protein IL-6 convey a strong risk for depression, although the latter effect may be confined to high-risk groups [40]. Several SNPs have been investigated, mostly with negative results. However, we found that an ER-alpha polymorphism had a substantial effect on anxiety but not on depression in women [41]. Methods update In the first years of the Rotterdam Study psychiatric data collection was very limited. However, from the third examination onwards, which began in 1997 (see Fig. 1), depressive symptoms and disorders are have been ascertained in all participants. The 4,603 participants with valid depression scores in this examination thus constitute the baseline population for longitudinal studies of depression. Because assessment of anxiety disorders, sleeping disturbances and complicated grief were added in the fourth examination, they can currently be studied only cross-sectionally. Assessment of determinants Psychiatric research in the Rotterdam Study focuses on biological risk factors. The vascular depression hypothesis was tested with different measures of atherosclerosis, arterial stiffness and cerebral blood flow [34, 35]. We also examined whether blood levels of vitamins and fatty acids, immune parameters, and markers of folate metabolism increased the likelihood of depression [37–40]. In one ongoing project, diurnal patterns of cortisol secretion are related to psychiatric and other disorders. Studies of genetic polymorphisms and brain morphology are underway. Current data collection includes a dexamethasone suppression test to measure hypothalamic-pituitary-adrenal axis activity in all participants, which is unique in a population-based study. Assessment of outcomes Depression is assessed using a two step procedure. First, participants completed the Center for Epidemiologic Studies Depression scale (CES-D) during the home interview to measure self-reported depressive symptoms [42]. As a second step, screen positive subjects, who scored above a well-established CES-D cut-off of 20 points, have a psychiatric interview using the Present State Examination (PSE/SCAN) [43]. This semi-structured interview is conducted by a clinician and yields DSM-IV diagnoses including major and minor depression and dysthymia. The following anxiety disorders are assessed with a slightly adapted Munich version of the Composite International Diagnostic Interview: generalized anxiety disorder, specific and social phobia, agoraphobia without panic disorder, and panic disorder [44]. Sleep quality and disturbance is measured with the Pittsburgh Sleep Quality Index. In addition, sleep duration and fragmentation are assessed with actigraphy, a method that infers wakefulness and sleep from the presence or absence of limb movement [45]. The Inventory of Complicated Grief is used to identify traumatic grief [46]. This is a condition distinct from normal grief and bereavement-related depression, characterized by symptoms like disbelief about the death and searching for the deceased. Ophthalmologic diseases Objectives The primary objectives of the ophthalmological part of the Rotterdam Study are to study frequency and risk factors of common and invalidating eye diseases with emphasis on age-related macular degeneration (AMD) and primary open angle glaucoma (POAG). Another line of research focuses on the associations between retinal vessel abnormalities and cardiovascular and neurologic diseases. Major findings AMD was the main cause of blindness in the high age-group [47]. Major risk factors that were found for AMD are smoking, atherosclerosis, hyperopia [48–50]. The APO-E ε-4 allele showed an inverse association with AMD [51]. First-degree relatives of patients with late ARM developed ARM at an increased rate at a relatively young age [52]. The heterogeneity of genetic risk among AMD families is considerable, and the proportion of high-risk families is relatively small [53]. The anticipated protective effect of cholesterol-lowering drugs on AMD could not be substantiated [54]. A protective effect on AMD was demonstrated in participants using a diet rich of antioxidants and zinc intake in our study population. Support for a multifactorial origin of AMD was found in the cumulative effects of aberrant CFH and CRP genes, inflammations and smoking as genetic and environmental stimulators of the complement cascade [55]. The prevalence estimates of POAG in Rotterdam was highly dependent on the applied criteria [56]. That was a major reason for proposing a classification system without final, subjective adjudication [57]. Systemic blood pressure and hypertension were associated with elevated intraocular pressure but not with prevalent POAG [58]. Relatives of patients with POAG have a strongly increased risk of glaucoma. Enlarged cup-disc ratio was the earliest and most prominent feature of familial aggregation [59]. The incidence of OAG rises significantly with age. Most of the patients with incident OAG were unaware of having OAG [60]. The incidence of visual field loss rises fivefold between 55 and 80 years to 20/1,000 person years in the general population [61]. Glaucoma is the main cause for visual field loss, followed by stroke [62]. Larger retinal venular diameters are associated with generalized atherosclerosis, inflammation and cholesterol levels. Retinal venular diameters are variable and may play their own independent role in predicting cardiovascular disorders [62]. Dilated retinal venules at baseline were predictive for stroke, cerebral infarction, dementia, white brain matter lesions, impaired glucose tolerance, diabetes mellitus and mortality [63, 64]. In systemic hypertension, the arteriolar diameters are narrow and arteriolar and venular narrowing precedes the onset of hypertension [65]. Methods update Repeated ophthalmic measurements include best corrected ETDRS visual acuity, refractive error, Goldmann applanation tonometry, keratometry, slitlamp examination of the anterior segment and visual field testing. In pharmacological mydriasis we make colour photographs of the lens, 35 degree photography of the macular area, and simultaneous stereoscopic imaging of the optic disc and macular area. Digital photography of the macular area and optic disc are added since the fourth follow-up examination. Scanning laser ophthalmoscopy (HRT II) measurements of the optic disc, macular pigment density measurements and optical coherence tomography of the macular area and optic disc were added during follow-up. For the assessment of AMD we use stereoscopic 35° macular images centred on the fovea. The images are graded according to the International Classification and Grading System for AMD [66]. The procedure for assessing POAG includes Goldmann applanation tonometry, visual field screening, ophthalmoscopy and stereoscopic fundus photography [57]. Assessment of retinal vascular diameters is done by analyses of digitized fundus transparencies with a semi automated system [58, 67]. Genetic studies Objectives The first objective of the laboratory team is to collect, store and manage the biological tissues sampled in the Rotterdam Study. The second objective of the group concerns genotyping and assessment of biomarkers. Major findings Both from biomarker studies as well as molecular genetic studies important findings have emerged. For example, among the biomarker analyses the study documenting the relationship between homocysteine and osteoporosis has been novel [68] and since widely replicated. Several candidate gene studies have also yielded major new insights coming from both exploratory studies as well as from collaborative replication efforts across all disease-oriented research lines in the Rotterdam Study. A unique feature of the Rotterdam Study is exploited by studying the relationship between pleiotropic gene variants and not one but multiple diseases and disease-related endpoints. For example, the studies on the promoter region of the IGF-1 gene revealed a series of consistent associations ranging from birth weight to diabetes [69], while other consistent associations involve the estrogen receptor alpha (ESR1) gene in relation to osteoporosis [70], osteoarthritis, height, myocardial infarction [5], age-at-menopause, and depression. The Rotterdam Study is involved as a major collaborative centre for studies for novel genes coming from genome-wide analysis (GWA) studies, including CFH in age-related macula degeneration [71], NOS1AP in QT interval [72], and several SNPs involved in height, type 2 diabetes, and breast cancer. Finally, Rotterdam Study investigators play a leading role in the emerging large global consortia focussing on the contribution of complex disease gene variants by prospective meta-analyses across many epidemiological cohorts, such as for osteoporosis in the GENOMOS Study and EUROSPAN. Methods update Data collection, storage and management At each examination, blood, serum, plasma (citrate, heparine and EDTA based), sputum, and urine are collected. Fasting blood samples are collected along with challenged samples as part of a glucose tolerance test. Sputum is collected before and after a dexamethasone-suppression test. Sputum is frozen at −196°C before and after the challenge and stored at −80°C. To obtain serum and plasma, tubes are centrifuged according to a protocol standardising time and conditions from the drawing of blood to centrifugation. All samples are snap frozen at −196°C using liquid nitrogen and stored at −80°C. RNA is isolated from blood within 5 h after sampling and stored at −20°C. DNA is isolated from blood and extraction has been recently automated using a Hamilton STAR pipetting platform and AGOWA magnetic bead technology. DNA sample storage is in Matrix 2D-barcode tubes in 96-well format. Overnight urine samples are collected, frozen at −196°C and stored at −80°C. For data management, an in-house customized laboratory management system has been developed. Sample retrieval will be automated with an in-house customized laboratory track and trace system. Blood assessments For all participants, serum cholesterol, HDL, LDL, triglycerides, glucose and glucose levels are assessed. In urine, micro albumin and creatinine are determined in all participants. There have been a large number of specific blood/serum/plasma-based biomarker assessments including steroids (e.g. estrogens, androgens, vitamin D, cortisol), interleukins, CRP, IGF1, insulin, iron-parameters (iron, ferritin and transferrin saturation), fibrinogen, homocysteine, folic acid, riboflavine, pyridoxine, SAM/SAH ratio, cobalamine, Lp-PLA2, Fas/Fas-L, vitamins, a-beta42/40 and thyroid hormones (TSH). Genotyping facilities Affiliated laboratory facilities include a medium/high-throughput platform for candidate gene studies and genome wide association (GWA) analyses. The facilities use high-end automated machinery including a Caliper/Zymark ALH 3000 pipetting robot (including a TwisterII, and integrated plate sealer, plate reader (OD 260/280), a Tecan EVO 150 Freedom pipetting robot, a Deerac Equator NS808 nanoliter liquid dispenser, 15 electronic PCR machines (ABI 9700, 2 × 384), an ABI7900HT Taqman machine (running 1 ng gDNA in 2 μl reactions), a WAVE 3500HT dHPLC, and two ABI3100 sequencing machines. DNA sample handling is centred on 384-well plates. Candidate gene studies are done mostly using Taqman genotyping with throughputs at 30,000 genotypes per day. Continuous efforts are focussed on reducing the required amount of genomic DNA, which is now down to 1 ng per genotype. Genome-wide genotyping studies are based on 500/1,000 K Affymetrix arrays and 317, 550 and 1,000 K Illumina arrays with throughputs at 100–300 arrays per week. The genotyping facility has been partly sponsored by NWO investment grants (911-03-012; 175.010.2005.011), is part of the ErasmusMC Biomics core facility, and serves as knowledge centre for polymorphism analysis attracting national and international interested parties, both academic and industrial. Candidate gene studies We have genotyped >200 polymorphisms across the complete cohort and conducted a large number of candidate gene studies in the Rotterdam Study. These mostly concern individual potentially functional single nucleotide polymorphisms (SNPs) per gene, but sometimes also haplotype tagging SNPs (e.g. ESR1, ESR2, HSD11B1, fibrinogen), and also high-density SNP screening (e.g. the vitamin D receptor gene [73]). The candidate genes studied include the apolipoprotein E gene (APOE), the angiotensin-converting enzyme (ACE), the gene encoding angiotensinogen (AGT), angiotensin II type 1 receptor (AT1R) gene, G protein beta3 (GNB3), adducine gene, Cholesteryl Ester Transfer Protein (CETP), Hepatic Lipase, Phosphodiesterase 4D (PDE4D), ALOX5AP encoding 5-lipoxygenase activating protein, a polymorphism in the regulatory region of the Insulin-like Growth Factor 1 (IGF-1) gene, the hemochromatosis (HFE) gene, Complement factor H gene (CFH), and several polymorphisms in genes from the estrogen-, thyroid-, cortisol-, vitamin D-, IGF-, and Wnt-signalling pathways, the homocysteine pathway, and several matrix molecules. Genome wide association (GWA) studies A recently developed approach to identify robust genetic factors for complex disease is the Genome Wide Association (GWA) analysis. This is based on genotyping epidemiological cohorts with ultra-high density SNP arrays with up to 1 million SNPs. The method has already been shown to successfully identify genetic factors for several traits and diseases including age-related macula degeneration, inflammatory bowel disease, body mass index, type 2 diabetes and breast cancer. Through a large grant from the Dutch research organization NWO one of the world’s largest GWA studies is conducted involving 10,000 DNA samples from the Rotterdam Study. This GWA is based on the Illumina 550 K arrays and finished by the end of 2007. The GWA data will be useful for all research lines within the Rotterdam Study, and will also serve as a control GWA dataset for other research centres in and outside the Netherlands for both SNP frequencies as well as copy number variations (CNVs). In addition to this, a pilot study has been performed on 500 women from the Rotterdam Study using the Affymetrix 500 K platform allowing direct platform comparisons and merging of data to further increase genome coverage. In addition, our group has been active in developing new software for GWA analyses [74]. Pharmaco-epidemiologic studies Objectives A major objective of the pharmaco-epidemiologic studies is to investigate the role of drugs as determinants of disease in the Rotterdam Study. This includes studying efficacy and effectiveness of drugs, as well as adverse reactions to drugs. Major findings Important findings have been published on pharmaco-epidemiological topics concerning the main outcomes in the Rotterdam Study. Studies about the association between dementia and antihypertensive drugs [75] and NSAIDs [26] have strongly suggested a protective effect of both groups of drugs. Several studies have been performed on cardiovascular topics [76–78]. In one of these studies, NSAIDs were associated with an increased risk of heart failure [79]. In line with the suspicion that QTc-prolonging drugs may cause sudden cardiac death, it was demonstrated in the Rotterdam Study that a prolonged QTc is indeed an important risk factor [77]. Furthermore, in one study it was demonstrated that high-dose corticosteroids increase the risk of atrial fibrillation [78]. In the important area of locomotor diseases, studies have demonstrated that thiazide diuretics protect against hip fracture [79] and that statins reduce the risk of vertebral fracture [80]. On the other hand, the risk that long-term use of certain NSAIDs may aggravate signs of osteoarthritis has been emphasized [81]. In the area of ophthalmological diseases, a protective effect of cholesterol-lowering agents on macular degeneration has been suggested [82]. In other areas, such as pharmacogenetics and other causes of interactions between drugs, several important findings have been published [83–91]. Methods update For several reasons, a drug is a highly attractive determinant in clinical epidemiology. First, drugs are probably the most important therapeutic intervention in health care. Despite rigorous clinical research before registration, many important effects of drugs are discovered after marketing. Second, all marketed drugs have proven biological activity, meaning that it concerns a determinant which really matters. Third, and as a consequence of the availability of complete medication histories in Dutch health care, the role of drug exposure can be assessed in a detailed way. In the Rotterdam Study, there is an almost complete coverage of the population as of 1 January, 1991 thanks to the fact that all pharmacies which serve the Ommoord region are on one computer network. To date, almost 3 million prescriptions have been delivered to the population of the Rotterdam Study and of each prescription, details are available about the product name and contents, ATC-code, dosage and duration of drug therapy. Drugs are a group of determinants, which can be studied in association with a large variety of diseases. In the Rotterdam Study, there is a strong interest in the association between drugs and the cardiovascular, neurological, endocrinological and ophthalmological diseases, which have been the main topics since starting. However, there is also important information about the association with psychiatric diseases, cancer, and chronic obstructive pulmonary disease. Moreover, important information about secondary outcomes, such as drug blood levels, other laboratory information, and information about hospital discharge diagnoses, is gathered on a continuous basis to facilitate pharmaco-epidemiological studies. Management of the Rotterdam Study The Rotterdam Study is directed by a Management Team comprising Jan Heeringa, MD, study coordinator, Eric Neeleman, head IT, Frank van Rooij, head data-management, and the scientific principal investigators Albert Hofman (PI Rotterdam Study, chairman), Monique Breteler (PI Neurological diseases), Cornelia van Duijn (PI Genetic studies), Gabriel Krestin (PI Radiology), Huibert Pols (PI Endocrinology), Bruno Stricker (PI Pharmaco-epidemiology), Henning Tiemeier (PI Psychiatric diseases), André Uitterlinden (PI Genome wide analysis), Johannes Vingerling (PI Ophthalmologic diseases) and Jacqueline Witteman (PI Cardiovascular diseases).

Intensive_Care_Med-2-2-1315315

SAPS 3—From evaluation of the patient to evaluation of the intensive care unit. Part 2: Development of a prognostic model for hospital mortality at ICU admission

Objective To develop a model to assess severity of illness and predict vital status at hospital discharge based on ICU admission data. Introduction One of the crucial steps in the evaluation of risk-adjusted outcomes is the choice of the reference database for estimating adequate reference lines for the analyzed variables. For the SAPS 3 to reflect the standard of practices and outcome in intensive care at the beginning of the 21st century, we decided to collect data from a large sample of intensive care units (ICUs) worldwide. Other models have restricted data collection to large ICUs in Europe or North America—SAPS II [1], MPM II [2], APACHE II [3] and APACHE III [4], a strategy that minimizes the heterogeneity of the sample but restricts the generalization of the results. At the statistical level, there is also a need for change, in order to take into account the hierarchic nature of our data [5, 6]. Current general outcome prediction models do not consider the existence of clinical and nonclinical factors, aggregated at the ICU level, that can have an important impact on prognosis. Instead, they assume that these factors are either not important or are randomly distributed throughout large samples and that the variation between ICUs is small. This assumption is not likely to be borne out at the ICU level for either nonclinical factors (e.g. organization and management, organizational culture) or clinical factors (e.g. clinical management, diagnostic and therapeutic strategies). If the variation between ICUs is not negligible, it will compromise the stability of the equations used to compute predicted mortality. Furthermore, the published models consider the relation between performance and severity of illness to be constant, and that may not be the case, since performance can vary within ICUs according to the severity of illness of the patients [7, 8]. To overcome this problem, we chose to adopt a new strategy for the development of the SAPS 3 score and to apply statistical modelling techniques that control for the clustering of patients within ICUs instead of assuming the independence of observations. Conceptually, the SAPS 3 admission core comprises the following parts: First, the SAPS 3 ADMISSION SCORE, represented by the arithmetic sum of three subscores, or boxes:Box I: What we know about the patient characteristics before ICU admission: age, previous health status, co-morbidities, location before ICU admission, length of stay in the hospital before ICU admission, and use of major therapeutic options before ICU admission.Box II: What we know about the circumstances of ICU admission: reason(s) for ICU admission, anatomic site of surgery (if applicable), planned or unplanned ICU admission, surgical status and infection at ICU admission.Box III: What we know about the presence and degree of physiologic derangement at ICU admission (within 1 h before or after admission). Second, the SAPS 3 PROBABILITY OF DEATH during a certain period of time (in the case of the main model, the probability of death at hospital discharge). Given our objective of evaluating not only individual patient outcome but also the effectiveness of ICU practices, we focused the model on data available at ICU admission or shortly thereafter. This model will be completely open and available free of any direct or indirect charges to the scientific community. Methods and statistical analysis Primary variable selection Based on the SAPS 3 Hospital Outcome Cohort as described in Part 1 of this report, continuous predictive variables were categorized in mutually exclusive categories based on smoothed curves such as LOWESS [9], showing the univariate dependence of hospital mortality on the predictive variables. Classes of categorical variables were also collapsed according to their univariate hospital mortality levels using multidimensional tables and clinical judgment as appropriate, depending on the nature of the data. Additively, regression trees (MART) [10] were applied to check the cutoffs. Missing values were coded as the reference or “normal” category for each variable. When dual data collection was used—maximum and minimum values recorded during a certain time period—missing maximum values of a variable were replaced by the minimum, if documented, and vice versa. Some regression imputations were performed if noticeable correlations to available values could be exploited. For a detailed description of data collection and handling, see Part 1 of this report. Selection of variables was done according to their association with hospital mortality, together with expert knowledge and definitions used in other severity of illness scoring systems. The objective of using this combination of techniques rather than regression-based criteria alone was to reach a compromise between over-sophistication of the model and knowledge from sources beyond the sample with its specific case mix and ICU characteristics. Cross validation For being able to cross-validate the model, we randomly extracted five roughly equal-sized parts based on number of patients from the database, as suggested previously [11]. In a second approach, partitioning was based on ICUs and not on patients. It was thus possible to run the model-building procedure five times in each of the two approaches, each time taking four parts of the sample as a development set and the remaining one as the validation set. This allowed to estimate the variability of prediction resulting from the construction process of the prognostic score. A further check of the stability of the predictions was made by partitioning the sample according to major patient characteristics, such as surgical status and infection status. The quality of predictions in the validation sets was assessed by looking at the goodness-of-fit in terms of the p values for the Hosmer-Lemeshow tests Ĉ and Ĥ [13] and the discriminative capability of the models by the use of the area under the receiver operating characteristic (aROC) curve [14, 15]. Another criterion to judge the appropriateness of the model was the fit in certain subsamples, defined according to major patient typologies [16]. Reducing model complexity To reduce the complexity of the model classes, we concentrated on logistic regression. In the first step a stepwise logistic regression was used to identify the significant predictors in each of the five subsamples. A threshold of 0.01 for the p value was generally applied for inclusion in the model to separate irrelevant predictors [12]. At this stage we also evaluated if interactions among these predictors would influence results. Interactions, however, did not make a valuable contribution for the prediction. Significant predictors (n=70) were in a second step entered into a logistic regression model. The criterion for a predictor to enter the model was homogeneity across the five model-building processes: in principle, predictors should enter the model in all five development sets, but depending on the frequency of the predictor in the samples, the magnitude of the effect, and medical reasoning, some predictors were included if they appeared in the model in at least three subsamples. An example is the presence of Acquired Immunodeficiency Syndrome (AIDS): it was selected as a comorbidity in only 81 patients (0.48%), but the mortality—without controlling for other variables—in these patients was 42%. By taking all the above steps to identify the set of predictors, although deliberately not using any formal numeric criterion, we reduced the complexity of the model to minimize the amount of overfitting: This process resulted in 61 item classes (representing 20 variables) remaining in the final model. Using the parameter estimates from the logistic regression as starting values, a multilevel model was applied in the next step, using patient characteristics as fixed effects and ICUs as a random effect. Estimates were again calculated for the five development sets (for both, patient and ICU -based development subsamples). At this stage it was checked if rounding of coefficients (which allows for an easier manual computation of the score) would influence results, which was found not to be the case. Consequently, this was the approach chosen for the final construction of the SAPS 3 admission score sheet. The stability of the processes of variable selection and reducing complexity was further checked by bootstraping with replacement the total sample 100 times, both at patient level and at ICU level. Predicting hospital mortality After this step was completed, a shrinking power transformation was applied. This procedure uses log-transformation of the score to reduce the influence of extreme score values (outliers) on the mortality prediction. For this purpose, the SAPS 3 score and the transformed log (SAPS 3 + g) scores were used to predict hospital mortality. Conventional logistic regression was used in the evaluation of this step because of convergence problems for the corresponding multilevel model in a few subsamples. The best shrinkage model then was selected (excluding the trivial model with the SAPS 3 score as the single predictor) by checking which of the terms in the model contributed best to the prediction and was moreover stable over the respective validation sets and specific subsamples. This procedure was applied on both, patient and ICU -based subsamples. After finishing these steps of cross-validation, the final estimates for the selected predictors of the SAPS 3 score as well as the selected shrinkage procedure were then calculated from the total sample of patients. To arrive at the customised models for each major geographic region, specific customised equations were calculated, relating, by logistic regression, the transformed log (SAPS 3 + g) admission scores computed as described above to the vital status at hospital discharge. This process allows both the intercept and the slope of the curve relating the SAPS 3 admission score to change across different regions. The goodness-of-fit of these equations was evaluated by means of the same methodology used for the global sample. SAS for Windows, version 8.02 (SAS Institute Inc., Cary, NC, USA) and MLwiN version 1.10.0007 (Centre for Multilevel Modelling, Institute of Education, London, UK) and the R Software Package (http://www.r-project.org) were used for the development of the model. Results Based on the methodology described, 20 variables were selected for the SAPS 3 admission score (Tables 1 and 2):Five variables for evaluating Box I: age, co-morbidities, use of vasoactive drugs before ICU admission, intrahospital location before ICU admission, and length of stay in the hospital before ICU admission;Five variables for evaluating Box II: reason(s) for ICU admission, planned/unplanned ICU admission, surgical status at ICU admission, anatomic site of surgery, and presence of infection at ICU admission and place acquired;Ten variables for evaluating Box III: lowest estimated Glasgow coma scale, highest heart rate, lowest systolic blood pressure, highest bilirubine, highest body temperature, highest creatinine, highest leukocytes, lowest platelets, lowest hydrogen ion concentration (pH), and ventilatory support and oxygenation.Table 1 SAPS 3 admission scoresheet—Part 1Box I035678911131518Age, years<40>=40<60>60<70>=70<75>=75<80>=80Co-MorbiditiesCancer therapy 2)Chron, HF (NYHA IV),Haematological cancer 3),4)Cirrhosis, AIDS 3)Cancer 5)Length of stay before ICU admission, days 1)<14>=14<28>=28Intra-hospital location before ICU admissionEmergency roomOther ICUOther 6)Use of major therapeutic options before ICU admissionVasoactive drugsBox II034569ICU admission: Planned or UnplannedUnplannedReason(s) for ICU admissionplease see Part 2 of the scoresheetSurgical status at ICU admissionScheduled surgeryNo surgery 7)Emergency surgeryAnatomical site of surgeryplease see Part 2 of the scoresheetAcute infection at ICU admissionNosocomial 8)Respiratory 9)Box III1513111087532024578Estimated Glasgow Coma Scale (lowest), points3–4567–12>=13Total bilirubine (highest), mg/dL<2>=2<6>=6Total bilirubine (highest), µmol/L<34.2>=34.2<102.6>=102.6Body temperature (highest), Degrees Celsius<35>=35Creatinine (highest), mg/dL<1.2>=1.2<2>=2<3.5>=3.5Creatinine (highest), µmol/L3–456<106.1>=106.1<176.8>=176.8<309.4>=309.4Heart rate (highest), beats/minute<120>=120<160>=160Leukocytes (highest), G/L<15>=15Hydrogen ion concentration (lowest), pH<=7.25>7.25Plateletes (lowest), G/L<20>=20<50>=50<100>=100Systolic blood pressure (lowest), mm Hg<40>=40<70>=70<120>=120Oxygenation 10), 11)PaO2/FiO2<100 and MVPaO2/FiO2>=100 and MVPaO2<60 and no MVPaO2>=60 and no MVThe definition for all variables can be found in detail in Appendix C of the ESM. For names and abbreviations which are differing from those in the ESM, explanations are given below. Generally, it should be noted that no mutually exclusive conditions exist for the following fields: Comorbidities, Reasons for ICU admission, and Acute infection at ICU admission. Thus, if a patient has more than one condition listed for a specific variable, points are assigned for all applicable combinations.1 This variable is calculated from the two data fields: ICU Admission date and time—Hospital admission date and time (see Appendix C of the ESM) 2 Cancer Therapy refers to the data definitions in Appendix C of the ESM: Co-Morbidities: Chemotherapy, Immunosupression other, Radiotherapy, Steroid treatment3 If a patient has both conditions he/she gets double points.4 Chronic HF (NYHA IV)/Haematological cancer refer both to the data definitions in Appendix C of the ESM: Co-Morbidities: Chronic heart failure class IV NYHA, Haematological cancer. 5 Cancer refers to the data definitions in Appendix C of the ESM: Co-Morbidities: Metastatic cancer.6 Other refers to the data definitions in Appendix C of the ESM: Intra-hospital location before ICU admission: Ward, Other.7 No surgery refers to the data definitions in Appendix C of the ESM: Surgical Status at ICU Admission: Patient not submitted to surgery.8 Nosocomial refers to the data definitions in Appendix C of the ESM: Acute infection at ICU admission—Acquisition: Hospital-acquired.9 Respiratory refers to the data definition in Appendix C of the ESM: Acute infection at ICU admission—Site: Lower respiratory tract: Pneumonia, Lung asbcess, other. 10 PaO2, FIO2 refer to the data definitions in Appendix C of the ESM: Arterial oxygen partial pressure (lowest), Inspiratory oxygen concentration.11 MV refers to the data definition in Appendix C of the ESM: Ventilatory support and mechanical ventilation.Table 2SAPS 3 admission scoresheet – Part 2Box II – continuedICU admission 12)16Reason(s) for ICU admission   Cardiovascular: Rhythm disturbances 13)–5   Neurologic: Seizures 13)–4   Cardiovascular: Hypovolemic hemorrhagic shock, Hypovolemic non hemorrhagic shock. / Digestive: Acute abdomen, Other 3)3   Neurologic: Coma, Stupor, Obtuned patient, Vigilance disturbances, Confusion, Agitation, Delirium4   Cardiovascular: Septic shock. / Cardiovascular: Anaphylactic shock, mixed and undefined shock 3)5   Hepatic: Liver failure6   Neurologic: Focal neurologic deficit7   Digestive: Severe pancreatitis9   Neurologic: Intracranial mass effect10   All others0Anatomical site of surgery   Transplantation surgery: Liver, Kidney, Pancreas, Kidney and pancreas, Transplantation other–11   Trauma – Other, isolated: (includes Thorax, Abdomen, limb); Trauma – Multiple–8   Cardiac surgery: CABG without valvular repair–6   Neurosurgery: Cerebrovascular accident5   All others012)Every patient gets an offset of 16 points for being admitted (to avoid negative SAPS 3 Scores).13) If both reasons for admission are present, only the worse valve (–4) is scored. An estimation of the variability of the coefficients in the overall sample and in the five disjoint subsamples is given in Table E8 of the Electronic Supplementary Material (ESM), together with their respective coefficients (unrounded and rounded) and p values. The SAPS 3 admission score can thus, in theory, vary from a minimum of 0 points to a maximum of 217 points. The distribution of the SAPS 3 admission score in our sample is presented in Fig. 1. The minimum value observed was 5, and the maximum value was 124, with a mean of 49.9±16.6 (mean ± SD) and a median of 48 (38–60). The highest explanatory power came from Box I, with Box II and Box III being less important for the outcome; the three boxes represent 50%, 22.5% and 27.5%, respectively, of the total Nagelkerke’s R-Square. The relationship between the SAPS 3 and vital status at hospital discharge is given by the equation: Logit = −32.6659 +ln(SAPS 3 score +20.5958) ×7.3068 and the probability of mortality by the equation: Probability of death = elogit/(1+elogit).Fig. 1 Distribution of the SAPS 3 admission score in the SAPS 3 database The relationship between the SAPS 3 admission score and the respective probability of death in the hospital is described in Fig. 2. Overall, no combined discrepancy between observed and expected outcomes across all of the strata was outside sampling variability as demonstrated a Hosmer-Lemeshow goodness-of-fit test Ĥ of 10.56 (p=0.39) and a Hosmer-Lemeshow goodness-of-fit test Ĉ of 14.29 (p=0.16) (Figs. 3, 4 and Table E9, ESM). The overall discriminatory capability of the model, as measured by aROC curve, was 0.848. The goodness-of-fit according to major patient typologies (surgical status, trauma, and infection) can be found in Table 3. Calibration and discrimination presented differences across different geographic areas: the best predictive results were achieved in patients from Northern Europe (observed-to-expected [O/E] mortality ratio 0.96 [0.83–1.09]) and the worst predictive results were obtained in patients from Central and South America (O/E mortality ratio, 1.30 [1.23–1.37]); see also Table 4 and Fig. 5 and Appendix B in the ESM.Fig. 2 Relationship between the SAPS 3 admission score and the respective probabilities of hospital mortalityFig. 3 Hosmer-Lemeshow goodness-of-fit test Ĉ in the overall sample. Predicted risk of hospital death, observed hospital mortality rate, and the corresponding number of patients per decile are shown. Columns: Number of patients; squares: mean SAPS 3-predicted mortality per decile; circles: mean observed mortality per decileFig. 4Hosmer-Lemeshow goodness-of-fit test Ĥ in the overall sample. Predicted risk of hospital death, observed hospital mortaliy rate, and the corresponding number of patients per decile are shown. Columns: Number of patients; squares: mean SAPS 3-predicted mortality per decile; circles: mean observed mortality per decileTable 3 Performance of the model across major patient typologiesPatient characteristics GOF test ĤpGOF test ĈpO/E ratio95% CIaROCTrauma patients19.920.039.030.531.030.93–1.120.854Non-operative admissionsa14.860.1417.80.061.010.98–1.040.825Scheduled surgerya11.50.3227.39<0.010.970.90–1.030.825Emergency surgerya4.970.8912.880.231.000.95–1.050.809No infectionb8.570.5714.770.141.000.97–1.020.846Community-acquired infectionc8.40.5911.760.31.000.96–1.050.786Hospital-acquired infectiond15.210.127.110.721.020.97–1.070.77GOF: Hosmer-Lemeshow goodness-of-fit; O/E: observed-to-expected mortality; CI: 95% confidence interval; aROC: area under receiver operating characteristic (curve)aNon-operative admissions, scheduled surgery emergency surgery: see data definitions appendix C, ESMbNo infection: Patients not infected at ICU admissioncCommunity-acquired infection: Patients with community-acquired infection at ICU admissiondHospital-acquired infection: Patients with hospital-acquired infection at ICU admissionTable 4 Performance of the model in the global sample and in different geographic areasRegionsGOF test ĤpGOF test ĈpO/E ratio95% CIaROCAustralasia15.250.128.090.620.920.85–0.990.839Central, South America78.01<0.0180.82<0.011.301.23–1.370.855Central, Western Europe56.45<0.0147.89<0.010.840.79–0.900.861Eastern Europe19.450.0318.690.041.091.00–1.190.903North Europe2.440.992.340.990.960.83–1.090.814Southern Europe, Mediterranean countries14.180.1620.780.021.020.98–1.050.834North America10.570.399.630.470.910.78–1.040.812Global database10.560.3914.290.1610.98–1.020.848GOF: Hosmer-Lemeshow goodness-of-fit; O/E: observed-to-expected mortality; CI: 95% confidence interval; aROC: area under the receiver operating characteristic (curve).Fig. 5 Observed-to-expected (O/E) mortality ratios by region. Observed-to-expected (O/E) mortality ratios are shown by region. Bars indicate 95% confidence intervals For a more precise estimation of the probability of death in the hospital across the different geographic regions, specific customised equations were calculated (Table 5). This customised approach allows each ICU to choose its own reference line for the prediction of hospital mortality: either the overall SAPS 3 hospital mortality sample or its own regional subsample. This approach can be supplemented in the future by customised equations at the country level if data are available and if a more precise estimation of outcome in a specific setting is needed. The overall goodness-of-fit of these customised equations for each region is presented in Table 5. A complete list of the number of patients and the respective O/E mortality ratios by country, according to the global equation and the regional equations, are presented in Tables E10 and E11 of the ESM, with point estimates varying at the global level from 0.68 (0.56–0.80) to 2.05 (1.27–2.82). Most O/E ratios are close to the identity line, as expected for a stable model.Table 5 Customized SAPS 3 admission equations for the different geographic areasAreaEquationGOF ĤpGOF ĈpO/ECIAustralasiaLogit=−22.5717 + ln (SAPS 3 score + 1) ×5.316310.430.402.200.991.000.93–1.07Central, South AmericaLogit=−64.5990 + ln (SAPS 3 score + 71.0599) ×13.23228.940.547.030.721.000.94–1.06Central, Western EuropeLogit=−36.0877 + ln (SAPS 3 score + 22.2655) ×7.986715.130.1312.150.271.000.94–1.06Eastern EuropeLogit=−60.1771 + ln (SAPS 3 score + 51.4043) ×12.684710.130.437.120.711.000.92–1.08North EuropeLogit=−26.9065 + ln (SAPS 3 score + 5.5077) ×6.27463.450.972.220.991.000.86–1.14Southern Europe, Mediterranean countriesLogit=−23.8501 + ln (SAPS 3 score + 5.5708) ×5.57095.280.8713.120.221.000.97–1.03North AmericaLogit=−18.8839 + ln (SAPS 3 score + 1) ×4.39794.220.934.470.921.000.86–1.14GOF Ĥ: Hosmer-Lemeshow goodness-of-fit Ĥ test; GOF Ĉ: Hosmer-Lemeshow goodness-of-fit Ĉ test; p: respective p-values; O/E: observed-to-expected mortality ratio; CI: 95% confidence interval Discussion We have presented the results of a large multicentric, multinational study aimed at updating the SAPS II model. This study was necessary for several reasons. First, the reference line used by SAPS II was derived from a database collected in the early 1990s; since that time, there have been changes in the prevalence of major diseases and in the availability and use of major diagnostic and therapeutic methods that are associated with a shift toward poor calibration of older models such as SAPS II and APACHE III [17, 18]. Second, SAPS II was developed from a database built exclusively from patients in Europe and North America. This sample may not be representative of the case mix and medical practices that constitute the reality of intensive care medicine in the rest of the world (e.g. Australasia or South America), where variability in structures and organization is probably related to outcome [19]. Third, since computation of predicted mortality is based on a reference database, the user should be able to choose between them, i.e., a global database, which provides a broader comparison at the potential cost of less relevance to local conditions, and a regional database, which provides a better comparison with ICUs in geographic proximity but at the cost of losing comparability with ICUs in other parts of the world. A third possibility could be added—a country-representative database—but such a database would raise the problem of whether the ICUs selected were representative of a certain country. Fourth, the development of computers in recent years has created easy access to strong computational power. One of the implications of this is that it is now possible to develop a new outcome prediction model, based on digital data acquisition and analysis, with minimal differences in definitions and application criteria. These advances were coupled with extensive automatic logical and error-checking capabilities and the availability of data collection manuals online. Moreover, developers of the SAPS 3 model could take advantage of computer-intensive methods of data selection and analysis, such as the use of additive partition trees and logistic regression with random effects. Several new statistical techniques have been used in recent years to allow a more stable prediction of outcome, such as genetic algorithms and artificial neural networks [20, 21], dynamic microsimulation techniques [22], and first- and second-level customization strategies [23–25]. However, the value of these techniques is for the moment limited, usually because they are based on regional databases [24–26] that prevent extrapolation to other settings; moreover, their superiority in even the regional setting still needs to be established. Finally, the SAPS 3 conceptually dissociates evaluation of the individual patient from evaluation of the ICU. Thus, for individual patient assessment, the system separates the relative contributions to prognosis of (i) chronic health status and previous therapy, (ii) the circumstances related to ICU admission, and (iii) the presence and degree of physiologic dysfunction. It is interesting to note that one half of the predictive power of the model is achieved with Box I, i.e., with the information that is available before ICU admission. The prognostic capabilities of the model can be further improved by 22.5% by using data related to the circumstances of the ICU admission (Box II), and by another 27.5% by the incorporation of physiologic data (Box III). These numbers are different from those published by Knaus et al. [4] but are based on what we have learned in the last years about prognostic determinants in the critically ill patient. For performance evaluation, several reference lines should be used, with risk-adjusted mortality in different patient typologies and not only O/E mortality ratios at hospital discharge in the overall ICU population [27]. The results of the SAPS 3 study showing that different O/E ratios were observed in different regions of the world should be explored further, since, apart from regional differences in case mix (not taken into account by the model), they can also be related to regional variations in structures and organization of acute medical care, to different lifestyles (e.g., prevalence of obesity, or alcohol and tobacco use) and/or—though less likely—to genetic differences among populations. We would like to re-emphasize that the model presented here is based exclusively on data (including physiologic data) available within 1 h of ICU admission and calibrated for manual data acquisition; consequently, it should be expected to overestimate mortality when an automatic patient data management system with a high sampling rate is used [28, 29]. Limiting acquisition of physiologic data to the hour of ICU admission should minimise the impact of this factor when compared with models based on the most deranged data from the first 24 h after ICU admission, probably at the expense of a small decrease in the ROC curve, a greater sensitivity to the exact time point at which admission to ICU occurs, and therefore more reliant on the assumption that measured physiology alone (as opposed to changes in physiology) predict outcome. It also allows the prediction of mortality to be done before ICU interventions take place. This gives the SAPS 3 admission model a major advantage over existing systems, such as the SAPS II or the APACHE II and III, since all these systems can be affected by the so-called Boyd and Grounds effect: the occurrence of more abnormal physiologic values during the first 24 h in the ICU, leading to an increase in computed severity of illness and a corresponding increase in predicted mortality. These increases may, however, be due not to a greater intrinsic severity of illness of the patient but to the provision of suboptimal care in the first 24 h of ICU admission, when a stable patient may be allowed to deteriorate [30]. Further studies should be done of factors occurring after ICU admission that influence risk-adjusted mortality. We should keep however in mind that this approach comes with one potential pitfall: a possible decrease in the amount of data available for the computation of the model; also, the shorter time period for data collection can eventually increase the likelihood of missing physiological data and the reliance on the assumption that missing physiological data are normal. This effect should be small, considering the widespread availability of monitoring and point-of-case analysers. Having demonstrated the internal validity of the SAPS 3 admission model by the extensive use of cross-validation techniques, we should stress that external validation is also necessary. The fact that the overall database was not collected to be representative of the global case-mix (and especially the case-mix of specific regional areas or patient typologies such as specific diseases) should be empirically tested. Furthermore, the rate of deterioration of our estimates over time should be followed by the appropriate use of temporal validation, especially to avoid what Popovich called grade inflation [18]. The SAPS 3 system was developed to be used free of charge by the scientific community; no proprietary information regarding the scientific content is retained. All the coefficients needed for the computation of outcome probabilities are available in the published material. The SAPS 3 can even be computed manually, using a simple scoresheet, although it was designed to be integrated into computerised data acquisition and storage systems that allow the automatic check of the quality of the registered data. In conclusion, we can say that at the end of this stage of the project, we have been able to overcome some of the problems inherent in current risk-adjustment systems. We have minimized user-dependent problems through the publication of careful, detailed definitions and criteria for data collection [31]. We have also addressed the patient-dependent problems by expanding the reference database and making it more representative of reality, in order to include the maximum possible range of variations for patient-centred variables and resulting patient-centred outcomes. This approach was complemented by the development of specific customised equations for major areas of the world, allowing ICUs to choose a reference line for outcome prediction—the global database or the regional database for their own area. Users of these models should keep in mind that benchmarking is a process of comparing an ICU with a reference population. The appropriate choice of reference population is difficult, and we cannot simply change it because the observed-to-predicted mortality rate is not the one we want. For this reason, the choice should depend on the objective of the benchmark: more precise estimation will need local or regional equations, developed from a more homogeneous case mix. A generalisable estimation will, on the other hand, need more global equations developed from a more representative case mix. Last but not least, we have successfully addressed some of the problems of prognostic model development, especially those related to the underlying statistical assumptions for the use of specific methods for selection and weighting of variables and the conceptual development of outcome prediction models. In the future, multi-level modelling with varying slopes (and not just random intercepts) might be able to give a better answer to researchers but for the moment they would make the models to complex to be managed outside a research environment. Electronic Supplementary Material (PDF 794 KB)

Matern_Child_Health_J-2-2-1592155

Preconception Care in International Settings

Objectives: This literature review briefly describes international programs, policies, and activities related to preconception care and resulting pregnancy outcomes. Methods: Electronic databases were searched and findings supplemented with secondary references cited in the original articles as well as textbook chapters, declarations, reports, and recommendations. Results: Forty-two articles, book chapters, declarations, and other published materials were reviewed. Policies, programs, and recommendations related to preconceptional health promotion exist worldwide and comprise a readily identifiable component of historic and modern initiatives pertaining to women's health, reproductive freedom, and child survival. Conclusions: The integration of preconception care services within a larger maternal and child health continuum of care is well aligned with a prevention-based approach to enhancing global health. Introduction Preconception health is widely recognized as a critical component of domestic and international maternal and child health promotion. Broadly defined as the provision of biomedical and behavioral interventions prior to conception in order to optimize women's wellness and subsequent pregnancy outcomes [1], the notion of preconception care can be found in various global policy and practice recommendations concerning women's health and child survival. Although programs and guidelines may vary in response to local needs, the overarching concept of preconception care is present among developing and industrialized countries, within socialized or market-based health care systems, and independent of political, cultural, or religious beliefs. Furthermore, the incorporation of various preconception care strategies and ideologies within international maternal and child health programs and policies spans nearly 30 years, indicating a long-term recognition of the relative importance of such interventions as a means of optimizing pregnancy outcomes. This article describes international efforts to reduce adverse maternal and infant outcomes through programs, policies, and activities related to preconception care. Because widespread support for preconception health promotion is comparatively recent, explicit mention of preconception care programs and policies per se is often difficult to identify within the international literature. Furthermore, because many developing countries adopt strategies guided by the declarations of international agencies or other coalitions [2], the development of preconception health initiatives must be traced through the history of various international health movements, starting with the promotion of global primary health care, followed by Safe Motherhood initiatives, and culminating with the women's rights and health movements. Thus, whereas the activities described herein may not be immediately identifiable as preconception care, they represent essential pieces of the preconception health puzzle. Materials and methods A MEDLINE search with English language and human subject restriction was conducted for the years 1980 through May 9, 2005 using the reference terms “preconception (pre-conception) care,” “preconception counseling,” “preconceptional care,” “periconception care,” and “periconceptional care.” A total of 756 citations were identified using the initial search parameters. References for which preconception care programs or activities outside of the United States comprised the primary topic of interest were included. Articles detailing clinical or scientific evidence of specific periconceptional interventions were excluded. Also reviewed were other published articles as well as textbook chapters, declarations, reports, and recommendations not retrievable from these databases. The search terms were also entered in various internet search engines and relevant web-based information was used to supplement the findings. Pertinent abstracts from the National Summit on Preconception Care (CDC, June 2005) were also reviewed. Results A total of 42 articles, book chapters, declarations, and other published materials were reviewed. The findings were categorized into three main areas in order to present the material in a coherent fashion: international conferences and accords, professional organizations, and international preconception care programs. International conferences and accords As preconception health is closely linked to women's health, language in support of preconception care is found in declarations and agreements derived at international conventions on the topic of women's wellness and reproductive health. For example, although the primary goal of the 1978 International Conference on Primary Health Care was to promote the health of all by outlining essential primary health care strategies to be implemented throughout the world, one important provision of the declaration outlined the importance of maternal and child health care and family planning as integral components of primary health care [2–4]. The Safe Motherhood Initiative, initially drafted in 1987 at the first international Safe Motherhood Conference in Nairobi, Kenya, focused on reducing morbidity and mortality associated with pregnancy and childbirth among developing countries [3]. The Safe Motherhood Inter-Agency Group (IAG) cosponsored the event and is now comprised of various international and national agencies whose mission is to enhance maternal and neonatal survival through the development and promotion of cost-effective interventions. In a global technical consultation held 10 years after the launch of this initiative, the IAG assessed relevant health indicators and developed a package of services to be offered to all women as a means of ensuring safe motherhood. Various preconception care services were outlined, including family planning, adolescent reproductive health education services, and community-based education pertaining to sexual and reproductive health as well as safe motherhood [5]. Covenants related to preconception care were adopted by the 1988 World Congress of Obstetrics and Gynaecology and its predecessor, the General Assembly of the International Federation of Gynaecology and Obstetrics. The 1988 World Congress heralded the promotion of women's health with the first step being prevention as a social responsibility; the General Assembly of the International Federation advocated governments undertake concrete measures to improve women's health and social status [6]. The second International Congress for Maternal and Neonatal Health, held in Monastir, Tunisia, in November 1984, included in its recommendations the need to promote the expansion of health services including family planning services and other preconception interventions such as smoking cessation, increased birth spacing, and tetanus vaccination [7]. The 1993 Report of the World Conference on Human Rights reaffirmed the need for equal access to health care for women, including a wide range of family planning services and called for reductions in infant and maternal mortality rates and overall improvements in women's health and nutrition [5, 8]. Similarly, the International Day of Action for Women's Health, first held on May 28, 1990, through the collaborative efforts of the Women's Global Network for Reproductive Rights and the Latin American and Caribbean Women's Health Network/ISIS International, reiterated the need for significant reductions in maternal mortality, particularly in Latin American countries [3]. International conferences held in subsequent years reflected an important paradigm shift in which pregnancy outcomes and maternal health were considered within a broader context encompassing reproductive health and women's health as well as socio-cultural factors [3]. Held in Cairo in 1994, the third decennial International Conference on Population and Development (ICPD) emphasized the importance of reproductive freedom and developed a definition of reproductive health that included “access to appropriate health care services that enable women to go safely through pregnancy and childbirth and provide couples with the best chance of having a healthy infant” [9]. In 1995, two additional international meetings took place—the World Summit for Social Development and the Fourth World Conference on Women—in which the tenets set forth by the ICPD with regard to maternal mortality and morbidity were reaffirmed and new goals for improving access to maternal health services were also derived [3, 5, 9, 10]. More recent international activities related to preconception health include the 1998 World Health Day and the Millennium Declaration. World Health Day 1998 focused on safe motherhood with the assertion that every pregnancy faces risks that may be ameliorated by enhancing access to quality maternal health services, delayed childbearing, and preventing unwanted pregnancies [11]. The Millennium Declaration, adopted by all member states of the United Nations in September 2000, included eight millennium goals to be achieved on a global level. The goals were designed as a measure of development against which countries may compare health indicators, programs, and policies. Maternal health and child mortality are explicitly detailed in two main goals and are closely related to other key areas including poverty, gender equality, nutrition, education, and infectious diseases [3, 12]. References to preconception health can also be found in the declarations of conferences addressing the issues of child health and survival. For example, the 1989 Convention on the Rights of the Child recognized the right of children to attain the highest possible standard of health and contained provisions for reductions in infant mortality, prenatal and postnatal care for mothers, family planning education, and services and health education for children and parents regarding nutrition, clean drinking water, and environmental contaminants [13]. In addition, the World Declaration on the Survival, Protection, and Development of Children, developed at the 1990 World Summit for Children, included tasks pertaining to the reduction of maternal mortality with a specific emphasis on family planning and child spacing as well as efforts to reduce infant and child mortality [14]. Professional organizations and associations International health care professionals also have long recognized the utility of preconception care, as evidenced by directives and recommendations supported by various professional organizations. In 1952, after significant changes to the system of reporting maternal deaths sparked concern within the medical community about maternal mortality rates in the United Kingdom, a series of inquiries were initiated that ultimately resulted in the ongoing Confidential Enquiries into Maternal and Child Health [3, 15]. In its current manifestation, the program seeks to improve the health of mothers, infants, and children through careful investigation of all maternal and perinatal deaths as well as issues associated with diabetes during pregnancy and child health. In 1982, the International Federation of Gynaecologists and Obstetricians (FIGO) joined the World Health Organization (WHO) in creating a task force to assess safe motherhood within regional and international communities. Since then, FIGO has conducted needs assessments and funded demonstration projects to ascertain cost-effective methods of identifying and managing pregnancy complications [3, 16, 17]. One such project, the Averting Maternal Death and Disability Program (AMDD), was established at the Columbia University Mailman School of Public Health and has partnered with numerous international agencies and communities to improve access to quality emergency obstetric care [17]. Nurses and midwives also have played an important role in the promotion of preconception care among the medical community. Themes associated with International Nurses Day have incorporated various components of the preconception care paradigm, including safe motherhood, school health, healthy families, women's health, community health, domestic violence, and the AIDS stigma. The International Confederation of Midwives, a member of the FIGO working group on Safe Motherhood and Newborn Health, has hosted a number of workshops related to safe motherhood and midwifery. Topics addressed at the workshops have included HIV/AIDS, gender-based violence, and most recently, maternal and newborn health during birth and the postnatal period [3, 18]. International preconception care guidelines and programs The structure of preconception care programs in the international community differs considerably according to the type of health system employed and the level of economic support. Wealthier countries tend to have better organized health systems [19] and therefore use broader policies and guidelines to direct the provision of preconception services within the health care sector. Thus, countries such Canada [1], the United Kingdom [20, 21], Spain [22], Australia, Hungary [23], and the Netherlands [24] are more likely to adopt various recommendations related to preconception health, whereas less-developed countries in the regions of Latin America [25], Africa, India, and the Middle East tend to use more targeted interventions [26]. For example, many African countries provide continuing education to community health care workers to help them deal with issues in preconception counseling and reproductive choices for patients with HIV. Within the developing countries of Latin America, preconception care has afforded such nations as Ecuador, Honduras, Nicaragua, Paraguay, and Peru an opportunity to address inequitable (as compared with more developed countries) maternal and perinatal morbidity and mortality rates [25]. Maternal mortality in these countries is cited as one of the primary causes of death among women 15 to 39 years of age. Using a $250,000 grant provided by WHO, these countries have undertaken a demonstration project to deliver preconception health education via community programs. The specific aims of the project are to develop and deliver a comprehensive preconception care package that includes (1) risk assessment (identify individual, family, and social risks and barriers to prenatal care); (2) health promotion (ensure proper nutrition; avoid substance, tobacco, and alcohol use; provide family planning; perform PAP smear screening, and provide ongoing care); and (3) treatment delivery (treat medical conditions and infections such as malaria and sexually transmitted diseases, update immunizations, provide nutritional supplementation such as folic acid, and conduct home visits). In China before 2003, a premarital health check was required of all couples planning to marry and consisted of medical examinations and testing as well as health education. The requirement has since been abolished and this change has been followed by concomitant declines in the rates of premarital examinations [27]. As a result, pilot studies have been designed to test the feasibility and content of a social marketing campaign for preconception care in China. The findings indicate that although women are interested in information about preconception health, numerous barriers to implementing a national program exist, including vertical health systems, a lack of coordinated efforts among governmental organizations providing family planning and primary health care, and an abundance of potentially confusing media health messages [27]. Preconception care services are sometimes provided in private clinics such as the Pre-Pregnancy Preparation Service (PPPS) in Hong Kong which provides pre-pregnancy counseling, medical testing, health assessments, and educational services to approximately 4,000 couples each year [28]. To reach women and children residing in remote regions of the Philippines, a mobile health services program has been developed whereby a multidisciplinary team of physicians and nurse midwives travel to specified areas and provide services for 1 to 3 days per month. The team screens pregnant women for various risk factors and nonpregnant women of reproductive age for chronic medical conditions. Family planning services are also provided [29]. The Russian Association of Gynecologists and Obstetricians has developed and implemented the People's Health Movement with a primary objective of promoting health and preventing disease throughout the stages of a woman's reproductive life. Based on policy recommendations arising from an alert that the state of maternal and child health was one of crisis proportion, preconception care services inclusive of family planning are now being initiated [30]. The Office de la Naissance et de l’Enfance in Belgium is currently establishing a national preconception care campaign to educate all women of reproductive age and all health professionals engaged in the care of women and children regarding the importance of preconception health care. The marketing tools and plans for disseminating the information are now being developed [31]. In France, preconception care has focused on more targeted recommendations, such as good glycemic control prior to conception. It had previously been established that preterm delivery among diabetic mothers was much higher than among the general population and that fetal loss and congenital malformations were assoicated with poor glycemic control [32]. Using one of the goals of the 1989 International Diabetes Federation meeting in St. Vincent’s, Italy, tertiary perinatal centers began offering preconception care in an attempt to reduce the level of adverse pregnancy outcomes of diabetic mothers [33]. Services included risk assessment of potential diabetic complications, education on nutritional and glycemic self-monitoring, and optimization of insulin treatment regimens as directed by treatment guidelines. Data from a cross-sectional study of 12 perinatal centers in France indicate that nearly half of all women with Type 1 diabetes received preconception care during 2000–2001 as compared to 24% of women with Type 2 diabetes. Because rates of adverse outcomes among infants of diabetic mothers were similar to those during 1986–1988, it was concluded that more effort was needed to achieve the targets set forth in the St. Vincent's declaration [33]. Canada's National Guidelines on Family-Centred Maternity and Newborn Care devotes an entire chapter to preconception care and describes the multitude of intrinsic and extrinsic factors that influence preconception health. Accordingly, various settings appropriate for the administration of preconception care interventions are discussed as well as various social and medical issues, including stress, social support, abuse and violence, healthy lifestyle practices, and nutrition [1]. Although information about the development of international preconception care clinics is scant, such programs have been noted in Hungary, the United Kingdom, and the Netherlands. In 1989, Hungary established the Optimal Family Planning Service (OFPS) under the direction of the WHO Collaborating Centre for the Community Control of Hereditary Diseases. The OFPS was comprised of 32 regional health care centers providing periconceptional care free of charge. The aim of this program was to reduce adverse perinatal outcomes and protect maternal health among all women. In 1996, the Hungarian government increased the number and scope of the centers to incorporate these services within primary health care [23]. Assessments of various indicators in Hungary 10 years after the creation of the OFPS indicate that the rates of major congenital anomalies decreased and that the use of protective factors such as folic acid supplementation, rubella vaccination, and infection screening increased. Barriers to the use of preconception care services included unnecessary medical examinations, an inability to effect change among unintended pregnancies, and additional costs [24]. Despite the existence of numerous clinical genetic centers in the Netherlands, only two preconception clinics have been described in the literature. The first, an outpatient preconception care clinic established by the University Medical Center of Nijmegen, provides thorough evaluations of potential pregnancy risk and preconceptional interventions for women referred by health care practitioners [34]. Treatment for hyperhomocysteinemia and changes in drug therapies for women taking potentially teratogenic medications were the most common preconception interventions at the clinic. A pilot preconception clinic in Maastricht was developed as a forum in which women planning a pregnancy could address questions and concerns. Couples attending the clinic responded favorably to the program and indicated that they would not have asked their general practitioner such questions [35]. These findings are supported by the results of recent surveys of Dutch women in which more than 70% of the respondents reported an interest in preconception counseling when offered by their own general practitioner [36]. However, although the majority of Dutch health care providers are in favor of establishing preconception care clinics [37, 38], lack of specific knowledge and time constraints have been cited as barriers to the provision of adequate preconception counseling [38]. Preconception clinics have also been described in London and Glasgow. The London clinic opened in 1978 and consisted of one obstetrician who advised women with previous pregnancy complications about the management of subsequent pregnancies [39, 40]. During the first 18 months of the clinic's operation, 56 women were seen. Opened in 1982, the Glasgow clinic also served women affected by prior adverse pregnancy outcomes. The clinic remained operational for 9 years and relied on a research nurse for initial evaluation and screening followed by a physician consultation 4 weeks later [40, 41]. The most common reasons for referral to the clinic were previous miscarriage, previous fetal abnormality, and chronic maternal disease. Improvements in pregnancy outcome were noted only among the women with chronic conditions [41]. Conclusions Global recognition of the potential benefits associated with preconception health promotion is not new. International policy directives and practice recommendations related to women's health, reproductive freedom, and child survival almost always include provisions for the enhancement of women's wellness and social status as a means of reducing adverse pregnancy outcomes. However, the relative success of preconception care programs in both developed and developing countries is directly related to the availability and accessibility of health care for women. Accordingly, WHO's World Health Report 2005: Make Every Mother and Child Count details the inherent interrelation of the needs of the mother and child, indicates that reproductive health comprises an essential element of the continuum of maternal and child health, and calls for a reformulation of interventions from vertical programs to those offering a wider range of services [42]. A similar longitudinal approach to women's wellness and reproductive health has been promoted in the U.S. in light of increasing rates of low birth weight and persistent racial disparities in maternal and infant outcomes [43–45]. The present description of international programs and perspectives on preconception care is limited in scope by numerous factors. The literature search was as comprehensive as possible but not exhaustive because of language restrictions, difficulty in retrieving documentation of international preconception care programs and policies, the relative scarcity of published information about international preconception care practices, and the myriad of maternal and child health outcomes that fall under the umbrella of preconception health and for which both population-based and targeted interventions exist. Finally, data pertaining to maternal and infant outcomes after the introduction of preconception care activities is often unavailable. Thus, this review provides only a broad overview of international activities related to preconception health. However, the findings are instructive as a preliminary step in the compilation and dissemination of information related to international preconception care recommendations and practices. Future research should include a regionalized inspection and comparison of preconception care programs, policies, and recommendations within and among various countries. The impact of these programs on community-specific indicators of maternal and perinatal health should also be evaluated. Irrespective of setting, the wider application of the core components of preconception care (i.e., risk assessment, health promotion, and intervention) has the potential to enhance the health and overall well-being of women, infants, and children around the globe. In light of the numerous factors influencing pregnancy outcome, an integrated approach to preconception health promotion has been proposed as perhaps the most effective and efficient means of implementing preconception health care [43–47]. The benefits of service integration are apparent in even the most resource-poor settings, as evidenced by the introduction and expansion of the WHO's strategy, the Integrated Management of Childhood Illness, in various countries [48]. Indeed, the perspective that preconception care is part of a larger continuum of care is well aligned with a prevention-focused public health paradigm. As such, the widespread promotion of preconception health may be the most timely and logical step toward ensuring global health.

Clin_Exp_Metastasis-4-1-2413104

Serum proteome analysis for profiling protein markers associated with carcinogenesis and lymph node metastasis in nasopharyngeal carcinoma

Nasopharyngeal carcinoma (NPC), one of the most common cancers in population with Chinese or Asian progeny, poses a serious health problem for southern China. It is unfortunate that most NPC victims have had lymph node metastasis (LNM) when first diagnosed. We believe that the 2D based serum proteome analysis can be useful in discovering new biomarkers that may aid in the diagnosis and therapy of NPC patients. To filter the tumor specific antigen markers of NPC, sera from 42 healthy volunteers, 27 non-LNM NPC patients and 37 LNM NPC patients were selected for screening study using 2D combined with MS. Pretreatment strategy, including sonication, albumin and immunoglobulin G (IgG) depletion, was adopted for screening differentially expressed proteins of low abundance in serum. By 2D image analysis and MALDI-TOF-MS identification, twenty-three protein spots were differentially expressed. Three of them were further validated in the sera using enzyme-linked immunosorbent assay (ELISA). Our research demonstrates that HSP70, sICAM-1 and SAA, confirmed with ELISA at sera and immunohistochemistry, are potential NPC metastasis-specific serum biomarkers which may be of great underlying significance in clinical detection and management of NPC. Introduction Nasopharyngeal carcinoma (NPC) is a rare malignancy in most parts of the world, though it is highly prevalent in Southern Asia, where the incidence is about a 100-fold higher than in other populations. It is one of the most confusing, commonly misdiagnosed and poorly understood diseases. Previous studies show that the cancer is an Epstein-Barr virus-associated malignancy with a remarkable racial and geographical distribution. The etiology of NPC is thought to be associated with a complex interaction of genetic, viral, environmental and dietary factors. Thanks to the advancements in genomics, proteomics and bioinformatics in recent decades, the etiology, carcinogenesis and progression of the disease is better understood. Research into these components may unravel the pathways in NPC development and potentially decipher the molecular characteristics of the malignancy [1, 2]. NPC is an insidious tumor and is usually at the stage of metastasis involving lymph nodes or other organs before it can be found. Thus early and accurate diagnosis is very important for therapy and prognosis of NPC patients. Unfortunately, no effective method of accurately diagnosing new-onset NPC is available now. We presume if specific serum biomarkers associated with NPC metastasis can be identified on the basis of advancements in genomics, proteomics and bioinformatics, an approach to the early detection and monitoring of NPC may be found. 2D comparative proteome analysis is a new technology for the separation and identification of disease-specific proteins, and it has been applied successfully to screen potential biomarkers for NPC in cell lines and tumor tissues [3–8]. In our previous studies, we employed proteomic techniques to study protein changes of CNE2, a poorly differentiated squamous carcinoma cell line of human NPC cells, induced by 12-O-Tetradecanoyl-phorbol-13-acetate (TPA). It is likely that TPA promotes NPC in necessary cooperation with EBV, or that it may function as an antiproliferative or differentiative revulsant in noninitiated cells [4]. However, if a marker can be detected only in surgical specimens, its clinical significance is limited, especially for early screening or diagnosis. Recently, Saeid R. et al. reported their pioneering work on NPC serum analysis by 2D without any pretreatment, revealing the enhanced expression of relatively abundant proteins as ceruloplasmin (CPL) [9]. In order to find relatively low-abundant serum proteins which may be more valuable in predicting NPC progression, we pretreated serum with sonication, albumin and IgG depletion before 2D analysis. By comparing 2D image analyses for healthy volunteers, non-LNM NPC and LNM NPC patients, we may identify the differentially expressed proteins so that specific serum biomarkers associated with NPC metastasis can be found. Materials and methods Subjects Serum samples from 42 healthy volunteers and 64 NPC patients were available for comparative proteomic analysis. Samples from NPC patients were obtained before any clinical treatment. All the patients had a nasopharyngeal biopsy and NPC was confirmed by two independent pathologists unaware of the patient’s condition. Careful medical examinations and CT scan determined non-LNM cases and LNM cases in the NPC patients. Clinical data of all the subjects for 2D and ELISA are shown in Table 1. In the screening study, serum samples from the same group were pooled to smooth intrinsic individual differences and enhance common characteristic traits only related to disease status [10], aliquoted, and stored at −80°C before analysis. In the validation study, 26 of the patients received a mean of 17.8 months of follow-up (range, 6–22 months) after radiotherapy and/or chemotherapy. Six of the followed-up patients died of NPC, and the other survived the follow-up period. Table 1Serum samples in the studySamples for 2-DESamples for ELISAHealthyNPCHealthyNPCLymph node status/NegativePositive/NegativePositiveNumber422737302730Gender (Male/Female)22/2021/623/1415/1521/620/10Age (±SD)42 ± 7.943 ± 10.950.5 ± 11.742.5 ± 8.943 ± 10.947.6 ± 10.2 Depletion of high-abundance proteins Sigma-Aldrich ProteoPrep® Immunoaffinity Albumin and IgG Depletion Kit was used for depletion of proteins, albumin and IgGs, allowing visualization of co-migrating proteins with albumin and IgG on a 2D gel and also higher sample loads for improved visualization of lower abundance proteins. A total of 30 μl of serum was diluted with 100 μl of equilibration buffer from the depletion kit (Sigma) and sonicated using four ultrasonic bursts of 15 s each. Between sonications, the samples were chilled on ice. Then serum albumin and IgG was depleted according to the manufacturer’s instructions. After collection of the bound proteins, the majority (>95%) of the unbound proteins were left in the depleted serum sample. Modified Bradford assay was used to measure the protein concentrations of all serum samples [11]. 2D analysis The Immobiline Dry strip (pH 4–7, length 17 cm, Bio-Rad) was rehydrated with 600 mg protein in 350 ml rehydration buffer containing 7 M urea, 2 M thiourea, 4% CHAPS, 65 mM DTT, 5 mM tributylphosphine, 1% IPG buffer and 0.002% bromophenol blue for 14 h at room temperature. IEF was performed using Protean IEF cell (Bio-Rad, Hercules, CA, USA) with a total of 70 kVh. Then the strip was subjected to two-step equilibration in equilibration buffer containing 6 M urea, 30% glycerol, 2% SDS and 50 mM Tris–HCl (pH 6.8) with 1% w/v DTT for the first step, and 2.5% w/v iodoacetamide for the second step. The second-dimensional SDS-PAGE gel (13% T, 2.7% C, and 20 × 20 × 1 mm3) was carried out using a Protean II xi 2D cell (Bio-Rad) according to the following procedure: 30 min at a constant current of 12 mA followed by 24 mA per gel until the bromophenol blue front reached the bottom of the gel. Subsequently the gels were stained with MS accommodated silver staining [12]. 2D of each sample was run in triplicate to minimize run-to-run variation. The silver-stained 2D gel were scanned using a Power-Look 1100 imaging scanner (Umax, Dallas, TX, USA) and analyzed using PDQuest 7.1 software package (Bio-Rad). In-gel enzymatic digestion Protein spots were excised from gel with an operating knife blade, destained twice with 30 mM potassium ferricyanide and 100 mM sodium thiosulfate (1:1 v/v) and then equilibrated in 50 mM NH4HCO3 to pH 8.0. After dehydrating with ACN and drying in N2 at 37°C for 20 min, the gel pieces were rehydrated in 15 ml trypsin solution (10 μg/ml in 25 mM NH4HCO3) at 4°C for 30 min and incubated at 37°C overnight. Peptides were extracted twice using 0.1% TFA in 50% CAN and dried with N2. MALDI-TOF-MS identification The peptide mixtures were solubilized with 0.5% TFA, with saturated α-cyano-4-hydroxy-trans-cinnamic (CHCA) solution in 0.1% TFA/50% acetonitrile as the matrix and analyzed using M@LDI R (Micromass, Manchester, UK). Mass spectra were externally calibrated with lock mass 2,465.199 Da and internally calibrated with autodigested peaks of trypsin (MH+: 2,211.105 Da). Protein identification and database searching Protein identification using peptide mass fingerprinting (PMF) was performed by the MASCOT search engine (http://www.matrixscience.com/, MatrixSicence Ltd., London, UK) against the MSDB protein database. The errors in peptide masses were in the range of 25 ppm. One missed tryptic cleavage site per peptide was allowed during the search. Proteins matching more than four peptides and with a MASCOT score higher than 64 were considered significant (P < 0.05). Carboamidomethylation of cysteine was selected as the static modification and oxidation of methionine as the differential modification. Protein identification results were filtered with peakErazor software. Enzyme-linked immunosorbent assay (ELISA) confirmation ELISAs were conducted to confirm the protein identification and differential expression of sICAM-1, SAA and HSP70 in sera from the three groups. Measurements were done by commercially available ELISA kits (StressXpress Hsp70 ELISA Kit: EKS-700B, Stressgen Biotechnologies, Victoria; High sensitivity sICAM-1 ELISA Kit: BMS241, Bender Medsystems Company, Australian; Human SAA ELISA Kit: EL10015, YES Biotech Laboratories Ltd, Canada), in accordance with the manufacturer’s instructions. All sera were stored at −80°C before they were measured. Both standards and samples were run in duplicate. The main protocols were as follows: monoclonal coating antibody was adsorbed onto microwells; target protein present in the sample or standard bound to antibodies was adsorbed to the microwells; second antibody was added and bound to target protein captured by the first antibody; following incubation unbound enzyme was removed during a wash step; a color reaction was formed and absorbance was measured at 450 nm. The standard curve was used to determine the concentration of target protein in an unknown sample. Immunohistochemistry (IHC) assay The previous studies showed that sICAM-1 in serum could be released by membrane-associated ICAM-1 of cancer cell [13, 14]. However, SAA is synthesized and secreted into serum by hepatocytes in the liver during the acute phase [15, 16]. In the present study, therefore, expressions of only HSP70 and ICAM-1 were detected in tissue by IHC assay. Three-micrometer sections of paraffin-embedded tissues were cut for a prior antigen retrieval step in boiling 10 mM citrate buffer pH 6 for 20 min. The fixed tissues were subjected to immunostaining using an ultrasensitive S-P technique (MaiXin, China). Briefly, the slides were incubated overnight at 4°C with goat anti-ICAM and anti-HSP70 polyclonal antibody (1:200) (Wuhan Boster Biological Technology Co., Ltd., China). The concentrations of the primary antibodies used have been optimized previously. After washing with PBS three times, they were treated with biotin-conjugated second antibody before adding streptavidin-peroxidase. For color reaction, diaminobenzidine (DAB) was used. All series included positive controls. For negative controls, the antibody was replaced by PBS. To evaluate the expression, the cellular localization of immunoreactivity was determined and scored for both intensity (negative, weak, moderate and strong) and proportion (5–25, 26–50, 51–75 and 76–100%) of stained cells. To allow comparison of immunostaining with pathological stage and metastasis for each cancer, integer values were assigned to the scores of intensity (0–3) and proportion of tumor cells stained (0–4). If the evaluations did not agree, the specimens were reevaluated and then classified according to the assessments given, most frequently, by the observers. The proportion and intensity scores were then added to obtain a total score, which was labeled as one of the four following categories: (a) high positive (+++), total score of 6–7; (b) moderate positive (++), total score of 4–5; (c) low positive (+), total score of 2–3; (d) negative (−), total score of 0–2. Statistical analysis All the analyses performed were descriptive. The serum levels of protein expressions were compared using one-way ANOVA test. Associations between immunohistochemical scores and clinicopathological variables of tissue specimens were evaluated by Fisher’s exact test or χ2 test. PDQuest software was used to establish a master gel representing each group [17]. For differential expression of matched gels, protein spots whose intensities were either increased or decreased two-fold or greater were marked and then confirmed by manual inspection of all relevant 2D gel, not only those included in the matchsets, to ensure consistency. Quantitative analysis was performed using the Student’s t-test among three groups of gels. All statistical analysis was performed with SPSS 10.0 software and P value <0.05 were considered statistically significant. Results Quantitative comparison and identification of protein spots on 2D gels Among the three groups, we found twenty-three spots significantly altered (P < 0.05) or absent/emergent, 13 of which were identified successfully (Fig. 1). The corresponding detailed differential expression patterns are shown in Fig. 2a. Quantitative comparison and MS identification of the 13 differentially expressed proteins are shown in Fig. 2b and summarized in Table 2, respectively. Overall, transferrin (TRF) and transthyretin (TTR) were down-regulated and 12-lipoxygenase (12-LOX), serum amyloid A1 protein precursor (SAA), soluble intercellular adhesion molecule-1 (sICAM-1) and lysine-specific histone demethylase1 (LSD1) were up-regulated in two NPC groups compared with the healthy group. Furthermore, increased expression levels of 12-LOX, sICAM-1 and SAA were observed in LNM NPC as compared with non-LNM NPC. However, heat shock protein 70 (HSP70) and cytochrome P450 (CYP450) were expressed significantly and constantly only in LNM NPC patients. The MALDI-TOF-MS mass spectra are shown in Fig. 3. Fig. 1Representative 2D patterns of albumin and IgG depleted serum from healthy (a), non-LNM NPC (b) and LNM NPC groups (c). In the 2D images, spots ‘1–10’ indicate up-regulation and ‘11–13’ down-regulation in NPC group. Black arrows indicate protein spots that were identified successfully by MSFig. 2Detailed 2D patterns of the thirteen differentially expressed proteins among healthy, non-LNM and LNM NPC groups (a). Corresponding gel positions are shown in Fig. 1. Magnified 2D maps demonstrating the different expressions of selected proteins are listed in Table 2. The densitometric analysis of each protein was calculated from nine different gels using PDQuest 7.1 software package (b). Each bar represents the mean ± S.D. of intensity, with significant differences between groups found by two-tailed student’s-test. Note that this panel has a different vertical (y) axis and a break in the axis to accommodate the poor performance of the subject spot No.6. The break region is from 5000 to 6000. *P < 0.05, compared with healthy group; **P < 0.05, compared with non-LNM NPC groupTable 2Identification results of differentially expressed proteins among healthy volunteers, non-LNM NPC and LNM NPC patientsNo.aProtein descriptionMSDB IDProtein levelbMASCOT scorecSequence coveraged (%)Theoretical Mr/pIe1Ovochymase precursorQ7RTY7↑11652124,947/8.672Arachidonate 12-lipoxygenaseAAA51587↑1936875,694/5.823sICAM-1Q99930↑112544270/6.404Cytochrome P450Q53EX9↑2075755,635/6.835Serum amyloid A1 protein precursorYLHUS↑1708513,524/6.286Hemoglobin beta subunits are s-nitrosylated1BUWB↑2075515,565/6.767E1A 10S proteinQ9YLA2↑1436928,493/4.098Lysine-specific histone demethylase 1O60341↑1725893,358/6.119KIAA1622 proteinQ9HCF0↑7861102,351/8.4210Heat shock 70 kDa protein (HSP70)Q6G1F9↑1575368,161/4.8811TransferrinQ9NQB8↓1676777,050/6.8112UPF0366 protein C11orf67AAD40378↓1045812,798/8.5813TransthyretinAAA61181↓876912,836/5.35aThe numbers indicate the spot positions in 2D gel as shown in Figs. 1 and 2b↑, up-regulation in NPC patients; ↓, down-regulation in NPC patientscBy MALDI-TOF MS analysisdCalculated by amino acid counteCalculated from the database entry without any processingFig. 3PMF of protein spots No. 3, 5 and 10, respectively, representing sICAM-1 (panel A), SAA (panel B), and HSP70 (panel C), which were expressed differentially between the groups of non-LNM and LNM NPC (MASCOT scores are 65, 85 and 78, respectively, P < 0.05) ELISA confirmation Figures 4 and 5 show the ELISA results of sICAM-1, SAA and HSP70 performed in the three groups. There were statistically significant differences between health and NPC groups (P < 0.01) as well as between non-LNM and LNM NPC groups (P < 0.05) (Fig. 4). And with the TNM stage advanced, the serum levels of sICAM-1 and HSP70, were detected significantly elevated, but that of SAA not (Fig. 5). Furthermore, bivariate correlation analysis showed a significant relationship between the serum levels of the three proteins (P < 0.01). The follow-up showed that the serum levels of sICAM-1 and HSP70, but not of SAA, were strongly associated with NPC mortality (Student’s t-test, P = 0.001, 0.001 and 0.132, respectively). Our ELISA findings were consistent with results of 2D and silver-staining. Fig. 4Quantitative comparison of sICAM-1, HSP70 and SAA in sera among healthy, non-LNM NPC and LNM NPC groups. Each bar represents the mean ± S.D. of the concentration, with significant differences between groups found by one-way ANOVA test. Note that this panel has a different vertical (y) axis and a break in the axis to accommodate the poor performance of the subject SAA. The break region is from 500 to 4500 μg/l. *P < 0.05, compared with healthy group; **P < 0.05, compared with non-LNM NPC groupFig. 5Quantitative comparison of sICAM-1, HSP70 and SAA in sera among NPC patients at different TNM stage. Each bar represents the mean ± S.D. of the concentration, with significant differences between groups found by one-way ANOVA test. Note that this panel has a different vertical (y) axis and a break in the axis to accommodate the poor performance of the subject SAA. The break region is from 600 to 2000 μg/l. *P < 0.05, compared with NPC patients at TNM stage I; **P < 0.05, compared with NPC patients at TNM stage II; ***P < 0.05, compared with NPC patients at TNM stage III IHC assay In NPC patients, HSP70 and ICAM-1 were observed in tumor cells but not in stromal, endothelial or inflammatory cells. The immunoreactivity was located in both membrane and cytoplasm for ICAM-1 and only in cytoplasm for HSP70 (Fig. 6). The immunoreactivities of HSP70 and ICAM-1 were observed in 37 (64.9%) and 43 (75.4 %) of the 57 cases of NPC, respectively. The correlation of positive rates showing HSP70 and ICAM-1 immunoreactivity with clinicpathological characteristics of the cases is summarized in Table 3. Statistically, positive expression rates of HSP70 and ICAM-1 were significantly correlated with lymph node metastasis (LNM) of NPC. Furthermore, we found that positive expression rates of ICAM-1 (6/6, 100%) and HSP70 (5/6, 83.3%) were high in the cases that died of NPC during the follow-up period (P > 0.05). The results manifested that ICAM-1 and HSP70 expressions were strongly associated with NPC metastasis and mortality, indicating poor prognosis. Fig. 6Immunohistochemical detection of HSP70 and ICAM-1 in NPC (Magnification ×200). The immunoreactivity was located in both membrane and cytoplasmic stainings for ICAM-1 but only in cytoplasmic staining for HSP70. The black arrows indicate the target proteins stained in brown-yellow in the NPC nest. The red arrow in E shows negative immunoreactivity of normal nasopharyngeal epithelia. a, negative expression of HSP70; b and c, positive expressions of HSP70; d, negative expression of ICAM-1; e and f, positive expressions of ICAM-1; a, b, d and e, non-LNM NPC; C and F, LNM NPCTable 3Characteristics of NPC patients dependent on ICAM-1 and HSP70 expressionsCharacteristicsNumberICAM-1 expressionHSP70 expressionNegativePositive (%)P valueNegativePositive (%)P valueGender    Male41932 (78.0) 0.4641526 (63.4) 0.704    Female16511 (68.8) 511 (68.8) AgeMean (45.4)     <45 years26719 (73.1) 0.704917 (65.4) 0.945    ≥45 years31724 (77.4) 1120 (64.5) Lymph node status    Positive30327 (90.0) 0.012525 (83.3) 0.002    Negative271116 (59.3) 1512 (44.4) TNM stage    I + II241113 (54.2) 0.0021212 (50.0) 0.044    III + IV33330 (90.9) 825 (75.8) Survival rate    Survival20614 (70)0.280614 (70)1.000    Mortality606 (100)15 (83.3)Note: Values in parentheses are percentages Discussion Serum is a complex body fluid, containing a large diversity of proteins. More than 10,000 different proteins are present in the human serum and many of them are secreted or shed by cells during different physiology or pathology processes [18]. Consequently, proteomics has raised great expectations for the discovery of biomarkers to improve diagnosis or classification of a wide range of diseases, including cancers [19]. Serum is expected to be an excellent source of protein biomarkers because it circulates through, or comes in contact with, all tissues. During this contact it is likely to pick up proteins secreted or shed by tissues, which has recently been tested and confirmed [20]. However, serum has been termed as the most complex human proteome [21] with considerable differences in the concentrations of individual proteins, ranging from several milligrams to less than one pictogram per milliliter [22]. Another analytical challenge for biomarker discovery arises from the high variability in the concentration and state of modification of some human plasma proteins between different individuals [23]. Despite these limitations, human serum holds immense diagnostic potential. In the last decade, several large-scale projects have been initiated, aimed at characterizing the human plasma/serum proteome. In the mean time, several serum proteomic studies on NPC have been reported. Cho et al. performed protein chip profiling analysis with surface-enhanced laser desorption ionization time-of-flight mass spectrometry (SELDI-TOF-MS) technology on sera from NPC patients and demonstrated that SAA may be a potentially usefully biomarker for NPC [24]. However, the technology for discovery of new cancer biomarkers has recently been questioned for its flaws, such as its qualitative nature, high identification error rate, poor reproducibility and nonspecific absorption matrices [25–27]. 2D-based comparative proteome analysis, though a new technology for the separation and identification of disease-specific proteins, has been applied successfully to screen potential biomarkers for NPC in cell lines and tumor tissues [3–8]. Recently, Saeid et al. reported their pioneering work on NPC serum analysis by 2D, revealing the enhanced expression of such relatively abundant proteins as ceruloplasmin (CPL) [9]. However, in 2D study, abundant proteins, such as albumin and IgG, that account for approximately 60–97% of the total serum proteins [28], mask other proteins that migrate to the surrounding areas and limit the loading amount of serum. As albumin and IgG are known to function as carriers and transporters of important proteins such as hormones, cytokines, and lipoproteins within the blood [29–31], the depletion of these two highly abundant proteins may result in the loss of potentially important proteins bound to them at the same time. In order to release those adsorbed or bound proteins, we sonicated the diluted sera before the depletion and desalting steps as suggested by Quero et al. [32, 33]. The improvements we made in sample preparation enabled us to find some valuable low abundant proteins. Of the 13 successfully identified protein spots, we focused on 10 up-regulated proteins in NPC for further validation. We also accumulated certain knowledge about three proteins by literature profiling. Finally, we decided to further investigate sICAM-1, HSP70 and SAA, which seemed more associated with our research interest, using both ELISA and IHC to validate their differential expressions. Intriguingly, most of these identified proteins have been reported to be associated with carcinogenesis and tumor metastasis. Our research is herein chiefly concerned with the functional implications of the three proteins to NPC at the serum level. Although the source of sICAM-1 has not been fully elucidated, researches show it can be released by cancer cells [13, 14] as well as by peripheral blood mononuclear, endothelial, and fibroblastic cells [34]. Proteolytic cleavage of membrane-bound ICAM-1 may be the most likely mechanism for the generation of sICAM-1 [35]. In patients with certain malignancies, the serum sICAM-1 titers have been found elevated in association with tumor growth and distant metastasis of malignant melanoma [36], lung [37], breast [38], gastric [39], hepatocellular [40], and colorectal cancers [41]. Poor survival of cancer patients correlated with a high level of serum sICAM-1 has also been demonstrated [36, 38, 39]. In our study, the positive expression of tissue ICAM-1 and levels of serum sICAM-1 were significantly correlated to the presence, progression, metastasis and mortality of NPC. sICAM-1 possesses most of the necessary extracellular structures to retain the functional activities of ICAM-1 [13, 35]. It has been reported that ICAM-1 on the surface of cancer cells or antigen presenting cells (i.e., macrophages) is a costimulatory factor that stabilizes T-cell receptor-mediated binding between these cells and T lymphocytes [42]. sICAM-1 would work as an immunosuppressive agent by blocking LFA-1 on T lymphocytes, thus rendering it less available for binding with ICAM-1 on the surface of cancer cells [43]. In this manner, the shedding of sICAM-1 may speed up the metastatic process by escaping host immune surveillance. This, probably, presents an additional potential mechanism accounting for high serum levels of sICAM-1 in NPC patients who have metastasized via hematogenous and lymphatic routes. As serum sICAM-1 may be useful for monitoring hematogenous metastasis, measuring the serum sICAM-1 level might be potentially significant in clinic. SAA is an acute-phase protein with various isoforms in a molecular mass range of 11–14. In normal individuals, SAA is produced by hepatocytes in the liver [16]. After secreted into serum, it rapidly binds to high-density lipoprotein, with 90% of the protein particles bound [15]. A review of the literature shows that only a low level of SAA can be found in the sera of healthy individuals, despite the ubiquitous nature of SAA [44]. This is in sharp contrast to the patients with neoplastic diseases, such as those with renal [45] and colorectal [46] cancers, who showed dramatic elevation of serum SAA. The prognostic significance of SAA for other cancers has also been found by conventional radioimmunoassay, in line with the findings of this study [47]. Cho et al. demonstrated that SAA may be a potentially usefully biomarker for NPC [24]. The report fully confirmed relatively huge concentrations of SAA in serum (0.2–2 g/l) that are thousands of times higher than classical cancer biomarkers (such as CEA, PSA, CA125, etc.) originating from tumor cells. In a similar way, we further confirmed that serum SAA was much elevated in NPC patients, particularly at the process of lymph node metastases when compared with non-LNM patients. However, such biomarkers (acute-phase reactants) are not commonly considered as cancer-specific ones and expected to be elevated in other malignant diseases or inflammatory diseases as well [26, 27]. In previous reports, SAA was found elevated in different malignancies, such as cancers of kidney, colon and prostate, as well as in leukemias and lymphomas [24, 48]. Therefore, SAA may represent a cancer epiphenomenon, unlikely to be of much clinical use in diagnosing and monitoring cancer [49], but it will be interesting to explore whether the rapid production of SAA in the liver or the epithelia of different organs [50] may be related to the stimulation by cytokines abundantly present in the NPC cells. It is also meaningful to investigate how SAA is produced at different stages of clinical manifestation in NPC patients. HSP70 is the main protein produced during cellular response to varied stresses, such as heat shock, ischemia/reperfusion, and oxidative changes [51, 52]. Because HSP70 confers cell protection against different stresses, it has been hypothesized that it plays a protective role in tumor growth in vivo [53, 54]. Indeed, HSP70 is overexpressed in human tumors of varied origins [55], such as colorectal [56], breast [57, 58], prostate [59], liver [60] cancers and melanoma [61]. LNM and poor survival of cancer patients correlated with HSP70 overexpression has also been documented [57, 58, 61–63]. It is suggested that HSP70 is needed for in vivo tumor progression. In addition, HSP70 may be released from tumor cells involving involves both active secretion and passive release from necrotic cells [64]. Physiological mechanisms include co-secretion in exosomes of HSC70 with TRF and the active HSP70 secretion by the nonclassical pathway employed by cytokines [65]. Therefore, HSP70 is detectable in serum so that it could potentially be used as a biomarker for diagnosis or disease classification. Abe et al. reported that HSP70 is a marker of prostate cancer, and may be used in conjunction with PSA to identify patients with early-stage prostate cancer [59]. However, to the best of our knowledge, the association between serum level of HSP70 and NPC status has not been reported. Since our findings demonstrated that there was a modest association between serum HSP70 level and NPC staging, the serum HSP70 level should not be considered as an independent prognostic factor in NPC patients, although it might be a prognostic predictor by univariate analysis. In short, serum HSP70 may have an adjunctive clinical value in monitoring tumor progression and evaluating prognosis in NPC patients, though its clinical application as a major tumor marker is limited. In conclusion, our serum proteomic analysis, using a comprehensive pretreatment strategy, provides a practical and exemplary tool of screening progression-associated serum proteins in NPC research. After comparing 2D image analyses for healthy volunteers, non-LNM NPC and LNM NPC patients, we successfully identified 13 differentially expressed protein spots. We further explored the differential expressions of three of the proteins, namely, sICAM-1, HSP70 and SAA, by ELISA at serum and IHC at tissue, though more work should be completed to pinpoint the direct correlation between serum level of sICAM and HSP70 and their expression level in the tissues. We suggest that the three proteins may be potential serum biomarkers which can serve as effective target points for early diagnosis and therapy of NPC patients, though further clinical research should be done before the potential come true.

Arch_Orthop_Trauma_Surg-3-1-2111040

Trochanteric fractures in the elderly: the influence of primary hip arthroplasty on 1-year mortality

Introduction The aim of the study was to compare the mortality risk and complication rate after operative treatment of pertrochanteric fractures with primary arthroplasty, dynamic hip screw (DHS) or proximal femoral nail (PFN). Introduction While relative consensus exists about the treatment of femoral neck fractures for elderly patients, the optimal treatment for per- and intertrochanteric fractures is still under debate [1, 7]. In a previous study of 1,173 patients with proximal femoral fractures it appeared that trochanteric fractures are still a challenge, as 1-year mortality risk and complication rate were considerably higher compared with femoral neck fractures [7]. As restoration of the preoperative ambulatory level correlated with survival rate after 1 year [7] and elderly patients are often unable to cooperate with partial weight bearing, the primary stability of the device is crucial to allow early mobilization to prevent cardio-pulmonal complications and thrombosis. Excessive collapse of the fracture site and varus displacement is a common problem of sliding hip screws combined with plates or femoral nails especially in elderly women who often suffer from osteoporosis and poor bone quality [16]. Therefore, primary arthroplasty was proposed by some authors [4, 10, 28] but their patient collectives ranging from 37 to 54 were to small to allow general recommendations. Because we used hip arthroplasty even for trochanteric fractures for a long time as a standard procedure [7, 26], we are now able to report on our experiences in a large number of patients. This study compares the mortality risks and 1-year postoperative complications between arthroplasty and osteosynthetic fixation (DHS or PFN) for intertrochanteric fractures. A 1-year mortality was chosen as a main indicator as it depends on the surgical trauma as well as the rapid return to preinjury activity and further complications. A 1-year period was chosen as the mortality risk approaches that of an age-matched reference population after this interval [32, 35]. Patients and methods A total of 308 patients who were treated for trochanteric femoral fractures between 1992 and 2005 with a minimum age of 60 years were enrolled in this study. Seventeen patients with pathologic fractures due to metastases were excluded. Information about survival after 1 year could be achieved from 283 of the 291 patients. Two hundred and seventeen patients (76.7%) were female. The mean age at the time of surgery was 80.5 (±9) years ranging from 60 to 98 years. Women (82 ± 8 years) of our cohort were significantly older than men (75.5 ± 11 years; P < 0.01). The fracture type was classified according to the System of the Orthopedic Trauma Association (AO/OTA) [24]. Pertrochanteric two part fractures are classified as A1-type, multi-part fractures as A2-type and reversed oblique intertrochanteric fractures as A3-type fractures. While all A1-type fractures are considered to be stable, most of the A2 and all A3 type fractures are unstable [19]. The distribution of the fracture types within the treatment groups is displayed in Fig. 1. Fig. 1Distribution of the surgical treatment depending on the fracture type. The dynamic hip screw (DHS) was mostly used in more stable fractures, while the proximal femoral nail (PFN) was mainly used in A3-type fractures (X² = 22.4, P < 0.001) A total of 109 patients (38.5%) were treated with a “dynamic hip screw” (DHS by Synthes®). Within this group there were mainly stable fractures type A1 (52%) or A2 (44%; Table 1); 132 patients (46.6%) were treated with primary arthroplasty, 117 with a total hip replacement (THR) and 15 with a bipolar hemiarthroplasty (HA). Since 2000 PFN (Synthes®) was used in 42 of the 100 cases, mainly for unstable fractures in patients without advanced osteoarthritis. As demonstrated in Table 1, these patients were younger and more frequently male (P < 0.05). Table 1Description of the baseline variables within the treatment groupsArthroplastyDHSPFNAge (years)83 ± 779 ± 975 ± 12Number of patients13210942Gender Female 115 (87%)80 (73%)26 (62%) Male17 (13%)29 (27%)16 (38%)≥4 comorbidities63 (48%)56 (51%)17 (41%)Fracture type A137 (28%)57 (52%)7 (17%) A282 (63%)48 (44%)29 (69%) A312 (9%)5 (4%)6 (14%)Year of injury 1992–1999111 (84%)72 (66%)0 (0%) 2000–200521 (16%)37 (34%)42 (100%) Outcome variables Primary outcome variable was the 1-year mortality risk (Table 2). Furthermore, the occurrence of complications which led to further interventions was analyzed in multiple regression analysis. These were dislocation of the hip, non-union of the fracture, cutting out of the hip screw, infections, thrombosis and hematoma. The Merle d'Aubigne test (0 to 18 points) was used to assess the patients [10]. We further questioned whether the patients returned to their preoperative status (e.g. living at home).Table 2The distribution of fracture type and treatment and its specific mortality1992–19992000–20051992–2005nDeathMortality (%)nDeathMortality (%)nDeathMortality (%)A1711926.829413.81002323.0 THR34926.50034926.5 HA003133.33133.3 DHS371027.019315.8561323.2 PFN000.0700.0700.0A2952829.5661319.71614125.5 THR652436.96233.3712636.6 HA001218.31218.3 DHS30413.318211.148612.5 PFN0030826.730826.7A317635.35120.022731.8 THR12541.60012541.6 HA000000 DHS5120.0005120.0 PFN005120.05120.0All fractures1835329.01001818.02837125.1 THR1113834.26233.31174034.2 HA0015213.315213.3 DHS721520.837513.51092018.4 PFN0042921.442921.4THR total hip replacement, HA hemiarthroplasty, PFN proximal femoral nail, DHS dynamic hip screw Surgical technique The proximal femoral nail (PFN by Synthes®-Switzerland) or the dynamic hip screw (DHS by Synthes®-Switzerland) was used according to the manufacturer’s instructions. For closed reduction the patients were positioned supine on a fracture table. Only 135° four hole-DHS plates without additional trochanteric stabilizing plates or tension band wires were used in this series. For hemiarthroplasty a standard cemented stem (Weller, Aesculap, Germany) and a bipolar head (DePuy, Leeds, England) were used. All femoral stems were cemented using the modern 3D generation cementing technique [2]. In cases of total hip replacement, a polyethylene cup (Aesculap®, Germany) was cemented in the same manner (Fig. 3). Statistics First we performed univariate analysis of independent variables (age, sex, treatment, preoperative risk factors, etc.) and mortality using Fischer’s exact test for qualitative variables and ANOVA for quantitative variables. The homogeneity of variance was checked by Levene test. All tests were two-sided and a P ≤ 0.05 was considered to be significant. The influence of treatment on mortality was checked by a multiple logistic regression model adjusted for the presence of preoperative comorbidities (four or more vs. up to three), fracture classification, age, gender, complications requiring revision surgery and year of surgery. We used RR additional to OR because the more frequent the outcomes becomes, the more the odds ratio will overestimate the risk ratio when it is more than 1 or underestimate the risk ratio when it is less than 1 [34]. Odds ratios (OR) were transferred to relative risks (RR) according to the method of Zhang et al. [34] with the formula: RR = OR/[(1 − P0) + (P0 × OR)], where P0 is the incidence of the outcome in the nonexposed group (reference category), and OR the odds ratio given by the multiple analysis. A RR < 1 refers to a factor category that is mitigating (decreasing) the risk of the outcome (e.g. dying) when present, an RR > 1 refers to a factor category that is increasing that risk when present compared with the reference category. All analyses were conducted with the SPSS statistic software for Windows 12.0 (SPSS, Chicago, IL, USA). Results Surgery Eighty-five percent of our patients were operated within 1 day after admission. In the other cases, operation was delayed to improve the preoperative condition of the patient, e.g. by transfusion or change of medication. This was independent of the treatment groups (P = 0.307). Mean blood loss and operating time differed significantly depending on the type of operation. The mean blood loss for arthroplasty (1,050 ± 700 ml) was significantly higher than for fixation with a DHS (409 ± 360 ml) or PFN (332 ± 277 ml; P < 0.001). Implantation of a hip replacement took a mean of 115 ± 36 min, that of a DHS 73 ± 38 min and of a PFN 84 ± 32 min (P < 0.001) including time for closed reduction on the extension rag. Fracture classification significantly influenced operating time if an arthroplasty was performed. It was 100 ± 24 min for A.1, 121 ± 39 min for A.2 and 127 ± 26 min for A.3-fractures (P = 0.012). The blood loss was similarly influenced (800 ± 460 ml for A.1, 1,200 ± 800 ml for A.2, 900 ± 500 ml for A.3 fractures; P = 0.019) Mortality Six patients (2.1%) died during hospitalization, 25 (8.9%) within 90 days. These were patients with considerable preoperative risk factors; 71 patients (25.2%) died within 1 year (Table 2). Influence of gender and age As demonstrated in Fig. 2, survival was correlated to age and gender; 34.8% of the male and 24.9% of our female patients dies within the first year (relative risk 0.34, P = 0.004; Table 4). Not considering the different treatment groups patients over the age of 80 years had an increased mortality risk (P = 0.004). Fig. 2Influence of age and gender. Independent of the treatment and fracture type the mortality risk was significantly influenced by age for both genders (P = 0.018 for men, P = 0.002 for women). After multivariate analysis, the risk to die was significantly lower for woman (relative risk 0.34, P = 0.004) and higher for patients over the age of 90 years (relative risk 3.61; P = 0.009) Influence of comorbidities The comorbidities and their prevalence are shown in Table 3. Of the single comorbidities only cardiac arrhythmia and cerebral diseases (P = 0.04) showed a significant influence on mortality. Logistic regression analysis demonstrated that four or more comorbidities increased the risk to die by 78% (relative risk 1.78, P = 0.009; Table 4). Forty-eight percent of our patients had four or more comorbidities and showed a mortality risk of 33.3% compared with 17.4% if they had less than four (P = 0.032 in χ2-test). Table 3The prevalence and relative risks of the most common comorbiditiesComorbiditynPoP (Fisher’s exact)RRORCI 95%P (log. reg.)Heart attack1240.3400.431.121.190.56–2.140.69Cardiac arrhythmia470.3540.01**1.101.161.02–2.430.04*Cardiac insufficiency1110.3040.821.441.790.67–3.180.55Hypertonus1540.2681.000.980.970.55–1.730.92Pulmonary diseases660.2880.731.071.10.57–2.160.77Anemia470.3190.351.261.430.69–2.960.33Renal insufficiency470.2611.000.850.810.38–1.750.60Cerebral diseases1270.3360.301.421.811.02–3.220.04*Diabetes mellitus920.3260.281.241.410.79–2.520.25Metabolic diseases300.2330.820.750.70.27–1.800.46Alcohol/nicotine/other drugs290.3210.351.511.980.79–4.990.14Gastrointestinal disease420.2440.310.760.710.32–1.640.43Others570.3040.851.011.010.51–2.030.96Relative risks (RR) are computed from odds ratios (OR) by the following formula: RR = OR/[(1 – P0) + (P0 × OR)]P0 is the proportion of patients with the respective disease dying within 1-year after proximal femoral fractureRR < 1.0 represents a minor risk of mortality; RR > 1.0 represents a major risk of mortality compared with patients without this criteria; * P ≤ 0.05; ** P ≤ 0.01Table 4Multiple logistic regressions analysis of a 1-year mortality after proximal femoral fractures of elderly patientsDescription of the analysisInfluence of interaction of kind of treatment and fracture classification after adjusting age, gender, comorbidities, revision surgery and date of surgery Odds ratio95% CIPRelative riskInfluence: interaction of kind of treatment and fracture classificatione A1*PFN0.0000.0000.9990.00 A1*DHS0.6380.228–1.7870.3920.71 A2*TEP1.3910.566–3.4200.4721.25 A2*PFN1.1240.272–4.6470.8721.08 A2*DHS0.3720.112–1.2370.1070.46 A3*TEP2.1200.502–8.9540.3071.59 A3*PFN1.0170.087–11.8330.9891.01 A3*DHS1.0930.096–12.3870.9431.06Influence: age categoryc 70–79 years1.2310.334–4.5320.7551.19 80–89 years2.7980.785–9.9700.1122.23 >90 years6.3811.603–25.4030.0093.61Influence: genderd Female gender0.3380.161–0.7080.0040.44Influence: comorbidities complications and year of surgery  ≥4 comorbidities2.2381.219–4.1090.0091.78 Revision surgery1.1950.448–3.1840.7221.13 Surgery after 20000.7750.343–1.7520.5410.83Relative risks (RR) are computed from OR (odds ratio) by the following formula:RR = OR/[(1 − P0) + (P0 × OR)]P0 is the proportion of patients in the respective reference category dying within 1-year after proximal femoral fractureRR < 1.0 represents a minor risk of mortality; RR > 1.0 represents a major risk of mortality compared with patients without this criteriaNumber of cases in all models: 282aReference: TEPbReference: A1cReference: <70 yearsdReference: male gendereReference: A1*TEP Influence of fracture-classification and treatment Only pertrochanteric fractures were enrolled in this study and sorted according to the AO/OTA classification (Fig. 1). While the mortality did not differ between A1- (23.0%) and A2-type fractures (25.5%), it was higher in A3-type fractures (31.8%; Table 2). As this fracture type affected only 7% of our patients, it did not significantly influence the mortality risk (P = 0.42). About 18.4% of the patients who received a dynamic hip screw, 21.4% of those who received a proximal femoral nail and 33% in the arthroplasty groups (THR and HA) died within 1 year (P = 0.011). As the indication for surgery was not randomized but decided on fracture classification and clinical status we attempted to reduce these influences in a multiple logistic regression analysis (Table 4). All potential influencing factors as gender, age, fracture type and amount of comorbidities were taken into account. If categorical variables were used to evaluate the interaction between fracture classification and treatment, we could not find a statistical influence of surgical treatment on mortality (Table 4). The relative risk to die was decreased for A.1- and A.2-type fractures, treated with a DHS, but this did not reach significance levels. As only seven patients with a A.1-type fracture were treated with a PFN (and all of them survived), no sensible statistical analysis can be done for this category. As all devices showed good results for this kind of fracture, the decision should be based on other parameters for A.1-type fractures. Because our treatment regime was changed in the beginning of 2000 (introduction of the PFN, new bipolar hemiarthroplasty instead of total hip arthroplasty; Table 2), we implicated the time of surgery in our analysis. Since then the 1-year mortality after arthroplasty decreased from 34.2 to 19% (P = 0.26, Fisher’s exact test). Although over all mortality decreased from 29 to 18%, no influence of time of surgery on the mortality risk could be found using multivariate regression analysis (P = 0.54; Table 4). Influence of complications The only postoperative complication which influenced the survival rate was postoperative pneumonia. It occurred in 13 cases independent of the treatment groups (P = 0.26). Six of the patients died within 1 year (P = 0.048). The DHS had a revision rate of 8.1%, while the PFN needed reoperations in nine cases (22.5%). The main reason was cutting out of the screw in the osteoporotic bone and postoperative hematoma. Due to cutting out and fixation failure secondary hip arthroplasty was performed after failed internal fixation ten times. It happened six times (5.4%) after a DHS and four times (9.8%) after a PFN. The main complication after arthroplasty was dislocation, which occurred in 16 patients (12%). It did not occur with bipolar hemiarthroplasty. While dislocation was only a minor problem in A1-type fractures (7%), it occurred in ten A2-type (12%) and two of the nine A3-type fractures (22%). Patients were usually treated by closed reduction and physiotherapy and it did not influence mortality (P = 0.78). Rehabilitation The mean hospital stay was 18 days and did not differ significantly between the three treatment groups (P = 0.7). It prolonged to a mean of 30 days if a reoperation became necessary (P = 0.003). Patients who were ambulatory before the injury could walk with a walker or sticks at time of discharge; 80% of the surviving patients who were independent before the injury regained their preoperative status after 1 year. Since the year 2000 patients were encountered to a prospective registry. The mean Merle d’Aubigné score (0–18 points) for the surviving 78 patients was 13.8 ± 2.5. No significant differences could be found between the three treatment groups (P = 0.122). Discussion While primary arthroplasty is a standard procedure for femoral neck fractures, little experience exists for trochanteric fractures [4, 5, 9, 28]. They were primarily used as a salvage procedure after failed internal fixation [12, 15, 21, 23, 31]. Most authors implanted special calcar replacement devices, which are expensive and require the removal of large portions of the proximal femur [3, 8, 13, 28]. We used a standard cemented stem and either a standard cemented cup or a bipolar head since the 1970s in our hospital. As it proved superior to former methods of internal fixation even for pertrochanteric fractures it was the treatment of choice since 1992 [26]. From the year 2000 on hemiarthroplasty was routinely used instead of total hip arthroplasty and the proximal femoral nail was introduced for patients without advanced osteoarthritis. The ideal treatment for intertrochanteric fractures is still under debate as none of the existing osteosynthetic devices could prove its superiority in former studies [16, 25, 29, 30]. The most frequent problem in elderly patients who usually are not able to walk without weight bearing was a cutting out of the hip screw which occurred in 4 to 20% of the reported cases [4, 6]. In our series this complication was only seen in comminuted fractures (A2 and A3 type). Thus, 10 times (6.2% of the internal fixation devices) a secondary hemiarthroplasty was performed. Immediate full weight bearing regardless of bone quality is an advantage of primary cemented arthroplasty. In our hospital, all patients are encouraged to stand up the day after surgery. Dislocation was the major complication in patients treated with a total hip replacement after comminuted pertrochanteric fractures. This high dislocation rate is rarely seen after primary hip arthroplasty but well known even for femoral neck fractures [22] or revision surgery [33] and can be attributed to insufficient muscular stability. Fractures of the greater or lesser trochanter might exaggerate this problem. The use of bipolar arthroplasty instead of total hip replacements can reduce this complication to an acceptable rate [11, 28]. Since we use total hip arthroplasty only in patients of good physical shape with severe osteoarthritis we did not see any dislocations. Ambulatory levels and clinical scores are considerable affected by the preoperative status. Since clinical examination usually cannot be obtained for all patients, reliable comparison is difficult. Some authors use telephone interviews [4] or report about less than 50% of their patients for a 1-year follow-up [10]. In both cases it can be assumed that patients with bad functional results are not regularly included. We could also examine only 78% of the last 100 patients clinically and therefore do not want to overemphasize these findings. Instead we chose the 1-year mortality risk as the primary outcome parameter as it could be obtained for all patients and is related to the postoperative ambulatory status as well as to the operative trauma [14]. In a previous study with more than 500 fractures, we could demonstrate that patients who regained their independency and were able to attend a rehabilitation institution had a significantly lower mortality risk [7], which was further supported in this study. Mortality was significantly influenced by patient related factors such as gender, age and comorbidities but not by the fracture type. In accordance with our former studies and other authors [7, 17, 20, 26] few single comorbidity influenced the mortality risk whereas the sum of four or more comorbidities increased the risk to die by approximately 78% (Tables 3 and 4). Patients who received a PFN or DHS were significantly younger and healthier than those in the arthroplasty group. These influences were reduced by multivariate regression analysis (Table 4). Furthermore, the interaction of fracture type and treatment was taken into account. Considering these influences, no significant influence of one of the three treatment groups on the mortality risk could be found. In studies that only use internal fixation devices mortality risks comparable with our over all mortality were found [27], thus—in accordance with other authors [3, 13]—we think that cemented hemiarthroplasty is an alternative to internal fixation in elder patients especially as our revision rate was significantly lower for arthroplasty compared with internal fixation. Only Chan et al. [4] published a study about the use of a standard hemiarthoplasty for the treatment of displaced intertrochanteric fractures. In their small group of 54 patients they found a mortality rate of 31.5% within 1 year which did not differ from comparable age-matched studies. The walking ability of their patients was favorable compared with other studies with internal fixation. We agree that immediate full weight bearing, which can be achieved after cemented arthroplasty, is crucial for these patients. Haentjens et al. compared 37 patients who received a calcar replacement device with 42 patients of a retrospective group, who were treated by internal fixation and found an advantage in functional outcome [10]. Although the mortality rate could not be decreased, early walking with full weight bearing reduced the incidence of pressure sores, pulmonary infection and atelectasis. Kim et al. [18] who compared a calcar replacement prosthesis with intramedullary fixation in a prospective study in two small groups of 29 patients could not find a significant difference concerning the functional outcomes, but the cut-out rate of the hip screw was 7% in their patients. Up to now a clear indication of arthroplasty in the treatment of pertrochanteric fractures does not exist. Although operation time and blood loss are higher no significant difference concerning the mortality rate between primary arthroplasty and internal fixation could be found, but the revision rate in the arthroplasty group was significantly smaller especially since we use bipolar hemiarthroplasty instead of a total hip replacement. If the cemented arthroplasty has been performed accurately, there is very little concern about weight bearing. Furthermore, primary arthroplasty eliminates the possibility for malunion, cut-out of the hip screw and avascular necrosis of the femoral head. But primary arthroplasty is a technically challenging procedure. All lose fragments including the greater and the lesser trochanter have to be attached with cerclage wires before cementing the stem to prevent extrusion (Fig. 3) and especially in utterly comminuted fractures orientation of the rotation and leg length can be demanding. Fig. 3Examples of treatment of A2-type fractures with a hemiendoprosthesis and tension band wires. Depending on the fracture type different ways of fixating the trochanteric region with wires were used. In the case of a calcar fragment reaching distal to the lesser trochanter a temporary reduction with a cerclage wire or reduction bone clamp was used before reaming. Final fixation of the greater trochanter was performed after reduction of the hip with cerclages or a tension band The DHS is not suitable for reverse oblique and comminuted fractures [19, 30] but for stable fractures it seems to be advantageous to short femoral nails regarding reoperation rate and fracture fixation failure in the literature [16, 29]. We further support the recommendation that the DHS should be used for stable A1- and A2.1-type fractures while an intramedullary device is advantageous for unstable fractures. Primary cemented arthroplasty is a viable option for the treatment of trochanteric fractures in a selected group of previously independent mobile patients especially if osteoporosis would prevent full weight bearing or osteoarthritis would make further operations necessary. The routine use of total hip arthroplasty was abandoned in our hospital as the luxation rate was higher than after hemiarthroplasty. It is still used as salvage procedure after failed internal fixation in elder patients with bad bone quality.

Breast_Cancer_Res_Treat-3-1-2045691

Differential effects of tamoxifen and anastrozole on optic cup size in breast cancer survivors

Introduction The main purpose of this study was to determine whether the optic cups of tamoxifen users and anastrozole users differ in size, with the cups of the tamoxifen users being smaller. Introduction Selective estrogen receptor modulators (SERMS) and aromatase inhibitors are the two major classes of medications used in the United States as adjuvant endocrine therapy for early-stage, hormone-receptor-positive breast cancer. Of the SERMS that are FDA-approved for this purpose [1], tamoxifen has been the most widely used, and for several decades it has been prescribed extensively for women of all ages [2]. The first aromatase inhibitor to be FDA-approved for early-stage breast cancer was anastrozole (Arimidex®), in 2002 [2]. However, because aromatase inhibitors do not block estrogen production adequately for pre-menopausal women [3], they are indicated for use by post-menopausal women only [4], for whom estrogen production is entirely non-ovarian [5]. Initial reports indicate that anastrozole is more effective than tamoxifen at preventing cancer recurrence [6, 7]. Thus, anastrozole now is being recommended for many women who would have been prescribed tamoxifen several years ago [8]. Because the use of aromatase inhibitors is likely to continue to increase and because the use of tamoxifen may become progressively more restricted to younger women, the potential side effects of anastrozole need to be evaluated, and the influence of age on tamoxifen side-effects needs to be closely examined. Of all the documented or putative effects of tamoxifen on the visual system [9], tamoxifen retinopathy may be the most widely known. However, tamoxifen retinopathy is uncommon [10] and possibly cumulative-dose-dependent [11]. In contrast, Eisner et al. [12] found that short-term tamoxifen users (i.e., ≤2 years of use) often have smaller optic cups than age-matched female control subjects, in a study of women who had been selected for the absence of any overt eye disease. This result was consistent with the possibility that tamoxifen often causes a subclinical degree of swelling at the optic cup, particularly since tamoxifen is a potent blocker (at least in culture) of swelling-activated chloride channels [13, 14], such as those in astrocytes [15, 16], and astrocytes are the predominant glial cell in the cup [17, 18]. If tamoxifen indeed causes optic-cup swelling via its ability to block swelling-activated chloride channels, the cup sizes of anastrozole users would be expected to more closely resemble those of female control subjects than those of tamoxifen users. The present study tested this prediction. Because of changing trends in adjuvant breast cancer treatment, the present study also assessed the effect of age on optic nerve head (ONH) parameters. The results support the inference that tamoxifen usage can lead to small cups, although apparently mainly for women older than about 50 years. All measurements for this study were obtained using confocal scanning laser ophthalmoscopy with the Heidelberg Retina Tomograph II (HRT). This commercial instrument was introduced in 1999 for routine clinical use, but until now, it has been used mainly to assess glaucoma and related conditions. Methods Subjects Three groups of amenorrheic women ages 40–69 years old were recruited for this study: (1) women using the standard dose of 20 mg tamoxifen daily as adjuvant therapy for breast cancer for at least 4 months but no longer than 2 years, (2) women using the standard dose of 1 mg anastrozole daily as adjuvant therapy for breast cancer for at least 4 months but no longer than 2 years, and (3) women not using any hormonally acting medications. This third group served as an approximately age-matched control group. The 2-year duration-of-use requirement was based on results from several studies of vision [19, 20] and the eye [12] indicating that short- versus long-term tamoxifen use can be distinguished operationally using a 2-year cutoff. The 4-month duration-of-use requirement was included to help ensure that the medications had time to act and that the body’s response had a chance to stabilize [21, 22]. All tamoxifen and anastrozole users had completed primary treatment for breast cancer, and all were fully active and able to perform their daily pre-cancer activities without restriction. None of the control subjects had positive breast cancer histories or had previously used breast cancer medication for any purpose. Similarly, none of the tamoxifen users had previously used a hormonally acting breast cancer medication other than tamoxifen, and none of the anastrozole users had previously used a hormonally acting breast cancer medication other than anastrozole. In addition, none of the subjects in any group had ever used the SERM raloxifene. With one exception, all analyses for this report are based on data from subjects who had been amenorrheic for at least 6 months at the time of testing. The one exception is an analysis comparing data from control subjects with data from an additional group of women (the fourth subject group overall) who met the same eligibility criteria as the control group, except for the absence of menses. All subjects, regardless of group, met a rigorous set of eligibility criteria for excellent ocular health in order to reduce the roles of confounding factors that could complicate the data. These are the same criteria we have used previously for several types of studies [12, 19, 23]. These criteria are: (1) 20/20 or better visual acuity in one eye and 20/25 or better visual acuity in the other eye, (2) no evidence or suspicion of eye disease on undilated direct ophthalmoscopic examination and on subsequent evaluation of individuals’ stereoscopic color fundus and ONH photographs by a glaucoma specialist (author JRS) who was masked from identifying subject information, (3) no history of eye disease or ocular hypertension, (4) no diabetes, (5) intraocular pressure (IOP) ≤22 mmHg on Goldmann applanation and no between-eye IOP difference >2 mm Hg, (6) no myopia >5 diopters, (7) no use of any medication (other than tamoxifen) known to affect vision, (8) no history of ocular surgery, and (9) normal color vision (i.e., no worse than a single minor transposition error on the D-15 test conducted under Macbeth illumination). There was no suggestion of papilledema for any subject (as for criterion #2). Demographic data for eligible subjects with usable HRT data (see second paragraph of Procedures) are given in Table 1 for each of the three amenorrheic subject groups plus the additional group comprised of women without breast cancer histories and still experiencing menses. The mean ages of the tamoxifen users were slightly lower than the mean ages of the anastrozole users and amenorrheic control subjects possibly because some tamoxifen users were not menopausal despite being amenorrheic [24]. For reference, the median age of natural menopause in American society is 51 years [25]. Table 1Demographic data for each of the subject groupsAmenorrheic subjects 40–69 years oldAmenorrheic subjects 51–69 years oldNot amenorrheic, no breast cancer history n = 19Anastrozole n = 20Tamoxifen n = 34Control, no breast cancer history n = 32Anastrozole n = 18Tamoxifen n = 25Control, no breast cancer history n = 29Age (years)58.2 (SD = 6.8)54.5 (SD = 5.2)57.2 (SD = 6.0)59.6 (SD = 5.4)56.5 (SD = 4.7)58.2 (SD = 5.1)48.0 (SD = 3.3)IOP (mmHg)15.7 (SD = 2.9)15.5 (SD = 2.3)15.6 (SD = 2.9)15.7 (SD = 3.0)15.5 (SD = 2.5)15.7 (SD = 3.0)15.5 (SD = 2.3)Duration of medication use (years)1.07 (SD = 0.45)1.31 (SD = 0.49)01.09 (SD = 0.471.36 (SD = 0.48)00IOPs are from subjects’ test eyes only All subjects were unpaid volunteers. Recruitment methods have been described previously [19]. All subjects were Caucasian, except for one Asian subject still experiencing menses. After being informed of the nature and possible consequences of the study, all subjects gave written informed consent to participate in this study. The study protocol adhered to the tenets of the Declaration of Helsinki and was approved by the OHSU Institutional Review Board and the OHSU Cancer Institute. For each subject, HRT data and IOP measurements were obtained at the same testing session. Subject recruitment and assessment lasted several years, until usable HRT data had been collected for 20 eligible anastrozole users. Procedures ONH scans were obtained using the HRT II according to standard techniques, after which a contour line is drawn on the average of three rapid-succession scans to define the margin of the ONH (i.e., of the optic disk) [26]. Stereo color photographs of the ONH were used to assist with contour-line placement, and extensive use was made of the ability of the HRT II software to rotate images in three dimensions. Contour lines were drawn with the grader masked from all other subject information, including subject group and age. Additional details have been presented previously [12]. However, we had previously placed the contour line as close as possible to the dark-to-light color change in the HRT reflectance image in order to define the margin of the ONH [12], but for the present study we ensured that the contour line was placed on stable rim tissue on top of the ring of Elschnig [27], which is often most discernable from the underside of the 3D image, as a ridge circumscribing the ONH. With these 3D-based contour lines, the mean measured disk areas for each of our subject groups were very nearly centered within the norms provided by Heidelberg Engineering on the HRT II printout. Thus, with one exception, all analyses of HRT values are based on the contour lines drawn according to the more recent set of criteria, which are intended to place the contour line exactly at the disk margin, rather than slightly within it. The one exception was made to assess the importance (or lack of importance) of the precise contour line placement for documenting between-group differences. Thus, we conclude the Results section by briefly reporting the results of analyses based on data obtained using the earlier contour-line placement criteria. Eyes were considered to have usable HRT data only if the average standard deviation for the height of each pixel in the 3 HRT scans did not exceed 50 μm, since average topographic standard deviation values above 50 μm are considered by Heidelberg Engineering to signify “low image quality” [28]. Image quality was equally good for each subject group. The mean average-topographic-standard-deviation for each of the three amenorrheic subject groups ranged from 20 to 22 (SD = 7–8) μm for all eligible subjects and also for subjects older than 50 years. Among women still experiencing menses, the mean was 23 (SD = 9) μm. Although HRT scans were obtained for each subject’s 2 eyes, only one eye per subject was used for data analysis for this paper. In this way, analyses could be limited to eyes with 20/20 or better acuity for a maximal number of subjects. Thus, test eyes were defined initially using the following steps applied in order as necessary: (a) the eye with the better acuity, even by 1 letter, than the fellow eye, (b) the eye with a lesser degree of spherical equivalent refractive error than the fellow eye, and (c) subject preference. In cases where the initially designated test eye had an HRT average topographic standard deviation value exceeding 50 μm, the fellow eye became the test eye for analysis purposes, but only if the fellow eye had 20/20 or better acuity and an HRT average topographic standard deviation value ≤50 μm. These are the same methods we used previously to designate test eyes [12]. Data analyses Comparisons of central tendency across more than two subject groups were made using Kruskal–Wallis non-parametric analyses of variance (ANOVAS) because the distributions of some of the HRT indices were significantly non-Gaussian (1-sample Kolmogorov-Smirnov test) in ways consistent with the literature [29]. Similarly, post hoc comparisons between pairs of groups were made using Mann–Whitney U tests. Relations of HRT indices to age were evaluated for significance using Spearman rank-order coefficients because the strongest age trends were not linear. None of the reported P-values have been adjusted for multiple comparisons, but the P-values required for statistical significance of post-hoc tests as determined using a step-down Bonferroni procedure [30] are provided in the legend to Table 2. All analyses were conducted using SYSTAT 10.2 (Richmond, CA). All P-values are for 2-sided tests. Table 2ONH indices: between-group comparisons (subjects 40–69 years old)ANOVA (Kruskal–Wallis)Anastrozole versus tamoxifen Tamoxifen versus control Anastrozole versus control Cup volumeP = 0.021P = 0.007P = 0.052P = 0.560 Maximum cup depthP = 0.016 P = 0.008P = 0.036P = 0.430Mean cup depthP = 0.028P = 0.011P = 0.050P = 0.645Cup areaP = 0.074P = 0.028P = 0.106P = 0.612Cup/disk area ratioP = 0.030P = 0.013P = 0.045P = 0.763Cup shapeP = 0.593P = 0.325P = 0.797P = 0.429Rim areaP = 0.115P = 0.173P = 0.053P = 0.492Rim volumeP = 0.084P = 0.059P = 0.061P = 0.940Disk areaP = 0.510P = 0.474P = 0.653P = 0.232Mean RNFL thicknessP = 0.718P = 0.654P = 0.438P = 0.735RNFL cross sectional areaP = 0.562P = 0.573P = 0.287P = 0.721Height variation contourP = 0.617P = 0.375P = 0.893P = 0.387Left-most data column––P-values for the non-parametric ANOVAS (Kruskal–Wallis) across all three amenorrheic subject-groups simultaneously. Comparisons based on the results for the larger contour circles (see Methods). Right 3 columns––unadjusted P-values for the post hoc comparisons (Mann–Whitney U tests) between pairs of subject-groups. The top 6 rows of variables are for cup data, the next 2 rows are for rim data, and the bottom 4 rows are for disk-margin data. The bold entries signify P-values that are considered statistically significant. For the ANOVAS, statistical significance was considered to be p ≤ 0.05. For the post-hoc comparisons, a step-down Bonferroni approach was used, so that the first (i.e., the most significant) comparison required a significant ANOVA plus P ≤ 0.05/3, the second most significant comparison required a significant first comparison plus P ≤ 0.05/2, and the third most significant comparison required a significant second comparison plus P ≤ 0.05. “RNFL” signifies retinal nerve fiber layer thickness. The height variation contour signifies the maximal minus the minimal retinal nerve fiber layer thickness, as defined in the text Results The optic cup volumes, maximum optic cup depths, and cup/disk area ratios were each significantly smaller for the tamoxifen users than for the anastrozole users. In contrast, none of the comparisons between anastrozole users and control subjects approached significance for any cup dimension. Table 2 provides P-values for the between-group comparisons for every summary ONH index. The cup-volume results are represented graphically as a function of age in Fig. 1, which presents the data from the three groups of amenorrheic subjects: tamoxifen users (filled symbols), anastrozole users, (shaded symbols), and control subjects (unfilled symbols). Fig. 1Graph of cup volume versus age for all amenorrheic subjects. Filled circles represent tamoxifen users, shaded circles represent anastrozole users, and unfilled circles represent control subjects The data in Fig. 1 indicate that among tamoxifen users, cup volumes tended to be relatively small mainly for those subjects older than about 50 years. Indeed, cup volumes decreased significantly with age (Spearman r = −0.41, P = 0.018) among tamoxifen users. However, a correlational analysis would necessarily reduce or obscure the significance of any age-related change if such change occurred predominantly within a narrow age range. Thus, although the corresponding results for cup area (Spearman r = −0.33), maximum cup depth (Spearman r = −0.30), and cup/disk area ratio (Spearman r = −0.34) were each non-significant when assessed using correlations, the cup volumes, cup areas, maximum cup depths and cup/disk area ratios were each significantly lower (P < .05) for subjects older than 50 years compared to subjects 50 years and younger when based on between-group comparisons of central tendencies. An age cutoff at 51 rather than 50 years yielded the same result. For each of these 2 age cutoffs, the difference in cup volumes between the older and younger tamoxifen users was highly significant (p ≤ .005). Among control subjects and anastrozole users, there were not enough younger subjects to make meaningful comparisons between these same 2 age classes. However, we also tested an additional group of women without breast cancer histories, comprised of subjects who met all the same eligibility criteria as the control group, except for the absence of menses. None of the ONH indices differed significantly between the amenorrheic control group (32 women with an average age of 57.2 years) and this additional group (19 women with an average age of 48.0 years), and there was no suggestion of any differences in the cup indices specifically. For example, the median cup volume equaled 0.077 mm3 for the 32 control subjects and 0.076 mm3 for the 19 subjects still experiencing menses. Similarly, the median cup/disk area ratio equaled 0.20 for the 32 control subjects and 0.21 for these 19 additional subjects. Among the 32 control subjects themselves, cup volume was not correlated with age (Spearman r = −0.08). Among anastrozole users, Spearman r = 0.27 (P = 0.255), but the regression line relating cup volume to age accounted for only 5% of the total variance, and for only 2% when the calculation was restricted to subjects older than 50 years. Because cup dimensions varied appreciably with age only for tamoxifen users, and because this variation appeared to be quite marked at about age 50 years, we recomputed the between-group comparisons of all the ONH indices for subjects older than 50 years. (None of the 19 women still experiencing menses were included in any of these comparisons). The results are shown in Table 3. As expected, all of the between-group differences that were significant when the data were compared for all subjects remained significant when the data were compared for only the older subjects, and many of the comparisons between tamoxifen users and control subjects became significant (e.g., for cup volume). In addition, rim volumes (i.e., the volume above and outside the cup but within the disk) now differed significantly between groups when compared for all three groups simultaneously, but not when compared for any two groups on subsequent post hoc analysis. Had we used a less conservative means of comparison (a parametric ANOVA followed by Tukey’s HSD test), the rim volumes of the tamoxifen users would have been considered to be significantly larger than the rim volumes of the anastrozole users and also than the rim volumes of the control users. This is probably the appropriate interpretation, since rim volumes, unlike cup volumes, were fairly normally distributed, allowing the use of parametric statistics. Table 3ONH indices: between-group comparisons (subjects 51–69 years old)ANOVA (Kruskal–Wallis)Anastrozole versus tamoxifen Tamoxifen versus control Anastrozole versus control Cup volumeP = 0.006P = 0.003P = 0.019P = 0.330Maximum cup depthP = 0.007P = 0.005P = 0.014P = 0.309Mean cup depthP = 0.016P = 0.007P = 0.038P = 0.463Cup areaP = 0.025P = 0.009P = 0.051P = 0.431Cup/disk area ratioP = 0.011P = 0.005P = 0.020P = 0.540Cup shapeP = 0.340P = 0.146P = 0.400P = 0.463Rim areaP = 0.113P = 0.161P = 0.049P = 0.511Rim volumeP = 0.030P = 0.018P = 0.035P = 0.678Disk areaP = 0.435P = 0.301P = 0.883P = 0.217Mean RNFL thicknessP = 0.542P = 0.350P = 0.336P = 1.000RNFL cross sectional areaP = 0.482P = 0.369P = 0.259P = 0.878Height variation contourP = 0.341P = 0.160P = 0.775P = 0.242Same as Table 2, except based on data from subjects older than 50 years) Table 4 presents the means and standard errors of the mean, along with the corresponding medians, for every ONH index for each of the three groups of subjects restricted to women older than 50 years. Note that the means and medians for many ONH indices (e.g., rim volume) were similar within groups, but that for some ONH indices (e.g., cup volume), means and medians were dissimilar, signifying a skewed distribution. Thus, Figs. 2–6 present between-group comparisons of medians with interquartile intervals plus overall ranges for cup volume (Fig. 2), cup area (Fig. 3), maximum cup depth (Fig. 4), cup/disk area ratio (Fig. 5), and rim volume (Fig. 6), in all cases for subjects older than 50 years. Although the cup areas of the tamoxifen users were not significantly smaller than those of the control subjects, their cup/disk ratios were significantly lower, probably because a major source of anatomic variability unrelated to tamoxifen use now was factored out. Table 4ONH indices: means and standard errors of the mean, and medians (subjects 51–69 years old)Anastrozole n = 18Tamoxifen n = 25Control n = 29Cup volume (mm3)0.107 ± 0.0170.043 ± 0.0150.090 ± 0.0170.1040.0120.073Maximum cup depth (mm)0.611 ± 0.0470.423 ± 0.0400.559 ± 0.0400.5910.3780.594Mean cup depth (mm)0.207 ± 0.0170.145 ± 0.0140.189 ± 0.0140.2060.1300.198Cup area (mm2)0.476 ± 0.0590.275 ± 0.0500.423 ± 0.0570.5230.1390.385Cup/disk area ratio0.207 ± 0.0220.123 ± 0.0210.193 ± 0.0220.2290.0790.192Cup shape−0.229 ± 0.015−0.199 ± 0.012−0.219 ± 0.013−0.229−0.205−0.210Rim area (mm2)1.724 ± 0.0661.895 ± 0.0711.697 ± 0.0611.7971.8671.732Rim volume (mm3)0.453 ± 0.0250.593 ± 0.0420.475 ± 0.0280.4610.5800.506Disk area (mm2)2.200 ± 0.0942.169 ± 0.0742.120 ± 0.0672.3162.1562.154Mean RNFL thickness (mm)0.245 ± 0.0100.264 ± 0.0140.246 ± 0.0080.2460.2620.242RNFL cross sectional area (mm2)1.281 ± 0.0591.366 ± 0.0141.266 ± 0.0491.2821.3350.242Height variation contour (mm)0.359 ± 0.0190.401 ± 0.0190.383 ± 0.0150.3750.3880.391Fig. 2Box plot of cup volume for women older than 50 years for each of the three amenorrheic subject groups. For each group, the horizontal line inside the box represents that group’s median value. The bottom and top of the box represent the upper and lower hinges, respectively, i.e., the first and third quartiles between which lie 50% of all the values. The whiskers represent the range of values that fall within a distance D from the hinges, where D equals 1.5 times the inter-quartile range. Asterisks represent individual points that fall between a distance of D and 2D from the hinges. Open circles represent individual points that are more distant outliersFig. 3Same as Fig. 2, except that box plots are for cup areasFig. 4Same as Fig. 2, except that box plots are for maximum cup depthsFig. 5Same as Fig. 2, except that box plots are for cup/disk area ratiosFig. 6Same as Fig. 2, except that box plots are for rim volumes When the between-group comparisons of the ONH indices were further restricted by excluding tamoxifen and anastrozole users using these medications for less than 1 year, every comparison that was significant (bold numerical entries in Table 3) remained significant. In addition, the duration of medication use was observed to matter little or not at all for the anastrozole users. For instance, the regression line relating cup volume to the duration of anastrozole use accounted for only 5% of the total cup-volume variance among all 20 anastrozole users, and for only 4% when the calculation was restricted to the 18 anastrozole users older than 50 years. These several sets of results indicate that the cup differences between the tamoxifen and anastrozole groups did not occur because the anastrozole users had experienced too short a period of medication use. The bottom 4 indices listed in Tables 2–4 represent biomorphometric assessments at the disk margin alone. These are: (1) disk area, (2) mean retinal nerve fiber layer (RNFL) thickness, (3) RNFL cross-sectional area, and (4) the height variation contour, calculated as the maximal minus the minimal RNFL thickness along the contour line. None of these indices differed significantly between groups, although there was some suggestion of greater RNFL thickness for the tamoxifen users (see Table 4). As stated in the Methods section, we also conducted analyses using the ONH indices derived from contour lines placed slightly interior to the anatomic disk margin rather than exactly at the disk margin. Every statistically significant result concerning the cup remained significant when based on the smaller contours. Conversely, whereas some of the cup-area comparisons involving tamoxifen users were not statistically significant when based on the larger contours (see Tables 2, 3), every corresponding comparison became significant when based on the smaller contours. In particular, the cup-area values for the older tamoxifen users (>50 years) became significantly lower than the cup-area values for the older control subjects (>50 years), presumably because using the smaller contours enhanced the importance of effects occurring predominantly within the interior of the cup. Discussion Because tamoxifen users and anastrozole users share similar medical histories, the results indicate that tamoxifen often leads to reductions of optic-cup size among middle-aged women older than about 50 years. Any effects of anastrozole at the ONH remain unproven, and if they exist at all, they are probably minor and opposite to the effects of tamoxifen. In contrast to tamoxifen [9, 10], there seem to be no reports in the peer-reviewed literature concerning any visual or ocular side effects of anastrozole. One study has found that anastrozole can cause regression of breast cancer metastases within the choroid [31]. Because all of the women in this study met a rigorous set of criteria for excellent ocular health, any cup-size changes presumably would be designated as subclinical. Data from several reports suggest that the color vision of tamoxifen users is often altered [10, 19, 20], although such alterations either are subtle or can be revealed only under specialized laboratory conditions, and there is no evidence that color vision changes are caused by effects occurring at the ONH. A recent survey found that about 13% of tamoxifen users reported experiencing vision changes that they attributed to tamoxifen [32], but the nature of these purported changes was not specified, nor were the subjects’ reports related to the results of any eye examination. Nevertheless, the subjects reporting vision changes had significantly higher serum levels of tamoxifen and a potent tamoxifen metabolite (N-DMT) than did the women not reporting these changes [32]. In contrast, reports of well-known side effects associated with tamoxifen use, such as hot flashes and vaginal dryness, were not related significantly to these same serum levels [32]. Cases of tamoxifen-induced optic neuropathy have been described in the literature [12, 33–38], but such cases usually are considered rare [9, 10, 38]. Perhaps tamoxifen users sometimes experience a degree of optic neuropathy that would not be detected or confirmed on a presenting eye examination owing to the wide range of normal cup appearances. If so, the results of this study point to a potential new application for intraocular imaging devices such as the HRT, which provides spatial resolution at the ONH on the order of tens of microns and is designed to detect longitudinal change [39]. Several types of high-resolution intraocular imaging device are now available clinically [39], and they allow high-resolution biomorphometric data to be collected in ways that are non-invasive and not intimidating to patients. The ease of data collection with these devices may allow patients who report tamoxifen-related vision change to be evaluated before and after cessation of tamoxifen use, and also after any rechallenge. Until this is done, and then only if changes in vision are shown to correspond to anatomical changes, might it be justified for the type of results described in this study to direct therapy. Presently, the only assured clinical application is to alert ophthalmologists and neurologists that assessment of the ONH for conditions such as glaucoma or optic neuritis may be complicated by the use of tamoxifen. A human-subjects study such as ours cannot definitively identify the mechanism(s) by which tamoxifen may reduce optic-cup size in vivo. Nevertheless, there is evidence to suggest that tamoxifen might reduce cup size via astrocytic swelling. First, tamoxifen is a potent blocker of swelling- (or “volume-”) activated chloride channels for astrocytes assessed in culture [16, 40], and astrocytes are abundant in the optic cup [18, 41]. Second, although astrocytes are present in the RNFL [17, 41], they comprise proportionally more of the neural volume as the cup is descended [41], and astrocytes along with ganglion cell axons are abundant at the base of the cup and also within the sides [42]. However, if the ganglion cell axons increased in volume, one would expect more visual dysfunction than is observed. In addition, there is evidence that tamoxifen-induced development of posterior subcapsular cataracts [10] may involve the blockage of volume-activated chloride channels [43], and that the utility of high-dose tamoxifen as salvage therapy for astrocytomas [44] may depend partly on the ability of tamoxifen to cause astrocytic swelling [45]. Based on photographic assessment of the ONH and retina by a neuro-ophthalmologist (author JF), there did not appear to be any difference in the vasculature of the three subject groups that would have led to the between-group cup differences quantified in this study and also observed in the ONH photographs when examined for groups of subjects rather than for isolated subjects. The presence of a steep reduction of cup size with age at about 50–51 years among the tamoxifen users suggests that any ability of tamoxifen to reduce cup size, by whatever means, depends on the absence or near absence of natural female hormones, given that the median age of natural menopause is 51 years [25] and given that tamoxifen elevates the estrogen levels of pre-menopausal women [46]. Hormone replacement therapy might, in principle, prevent or alleviate tamoxifen-induced reductions in cup size, but hormone replacement is generally contraindicated for breast cancer survivors, especially for women who have had hormone-receptor-positive tumors [47] (although diverse opinions exist [48]). Hormonal reduction by itself, whether occurring naturally or induced by anastrozole, does not appear to result in cup-size reductions of nearly the magnitude or the frequency inferred for the older tamoxifen users, at least not for the age range that we investigated. The presence of small cups among the tamoxifen users probably did not result from selectively low IOP levels, since the measured IOPs for each subject group were quite similar, differing by no more than 0.2 mmHg on average. Because the presence of small cups among tamoxifen users appeared to be unaccompanied by corresponding significant effects at the disk margin, the effects of tamoxifen on the ONH may be regarded as predominantly localized to the cup. For this study, we found no evidence of any age-related change of ONH indices for healthy women not using hormonally acting medications. However, a recent study of hundreds of subjects has documented the presence of ONH differences between healthy men and women older than 65 years [49]. In that study, several ONH indices were found to differ significantly between men and women, but the greatest percentage difference was for cup area, which averaged about 14% smaller for women than for men [49]. There is separate evidence that cup areas can depend slightly on the phase of the natural menstrual cycle [50]. These 2 sets of results along with our own results suggest that further studies be conducted to elucidate the effects of hormonally acting medications on the ONH. Such studies should include elderly women using tamoxifen, if possible, since elderly women have higher circulating levels of tamoxifen than do younger women [51].

J_Gastrointest_Surg-3-1-1915638

Long-term Results of a Primary End-to-end Anastomosis in Peroperative Detected Bile Duct Injury

The management of a bile duct injury detected during laparoscopic cholecystectomy is still under discussion. An end-to-end anastomosis (with or without T-tube drainage) in peroperative detected bile duct injury has been reported to be associated with stricture formation of the anastomosis area and recurrent jaundice. Between 1991 and 2005, 56 of a total of 500 bile duct injury patients were referred for treating complications after a primary end-to-end anastomosis. After referral, 43 (77%) patients were initially treated endoscopically or by percutaneous transhepatic stent placement (n = 3; 5%). After a mean follow-up of 7 ± 3.3 years, 37 patients (66%) were successfully treated with dilatation and endoscopically placed stents. One patient died due to a treatment-related complication. A total of 18 patients (32%) underwent a hepaticojejunostomy. Postoperative complications occurred in three patients (5%) without hospital mortality. These data confirm that end-to-end anastomosis might be considered as a primary treatment for peroperative detected transection of the bile duct without extensive tissue loss. Complications (stricture or leakage) can be adequately managed by endoscopic or percutaneous drainage the majority of patients (66%) and reconstructive surgery after complicated end-to-end anastomosis is a procedure with relative low morbidity and no mortality. Introduction Bile duct injury (BDI) after laparoscopic cholecystectomy (LC) is still a major problem in current surgical practice. BDI is associated with reduced survival, increased morbidity, and poor long-term quality of life (QoL).1,2 The incidence of BDI at laparoscopic cholecystectomy has been reported between 0.3 to 1.4%,3–5 depending on the criteria used to define the injury as well as the study population. Of these injuries, one-third is detected during the procedure.6 Measures to prevent and recognize BDI are outlined in many publications.6–8,9–11 The optimal treatment strategy and short- and long-term outcome has been published extensively.12–14 Controversy exists however about the management of peroperative detected BDI. The most important factor is the extent of tissue loss of the common bile duct, but also severity of inflammation and the size and diameter of the proximal duct. The peroperative management range from simple drainage and referral to a tertiary center to an end-to-end anastomosis (EEA) (duct to duct, with or without T-tube drainage) or a hepaticojejunostomy (HJ). It has been suggested that EEA is associated with a relative high stricture rate up to 70–80% and consequently a high incidence of secondary repair.15 Therefore, many tertiary centers prefer to perform a HJ instantly. A secondary repair after EEA should be associated with an increased risk of postoperative complications as the formation for strictures and stenosis.16,17 Others, however, consider EEA as a relative simple definitive repair, and also an optimal initial drainage procedure before reconstructive surgery in a secondary setting.18 Reports on large consecutive series to analyze the outcome of EEA are scary because this procedure is generally not performed in referral centers. One should realize that patients referred to such a center after previous EEA elsewhere are a negative selection of the EEA population. So far, a systematic analysis of a large group of patients with an EEA has not been performed and therefore this study was conducted. The aim of the present study was to analyze short- and long-term outcome in patients who are referred after failure of a primary EEA. Patients and Methods Patients Cohort and Data Collection Between January 1991 and January 2006, 500 consecutive patients were referred to the Academic Medical Center (AMC) in Amsterdam for the management of a BDI after cholecystectomy. Patient data was induced in a prospective database. All types of BDI were included, also minor injuries such as leakage from the cystic duct or ducts of Luschka. To define the location of BDI, the Bismuth classification was used.18 For the present study, the medical charts of all patients who underwent a primary EEA were retrospectively reviewed to analyze the initial operation reports and clinical data. Data from the referring hospital included: indication for cholecystectomy, type of initial procedure, location of injury, type of repair including the use of a T-tube, the postoperative diagnostic interventions, and the therapeutic interventions before referral. Data from the present center included: symptoms at referral, diagnostic work-up, type of treatment, short-term, and long-term complications. Endoscopic, Radiological, and Surgical Treatment for complicated EEA Endoscopic treatment was performed by balloon dilatation or catheter dilatation before stent placement. The biliary stent is placed over the guide wire bridging the stenosis. Two or more stents were inserted if possible. For multiple stent insertion, an endoscopic sphincterotomy was performed to facilitate stent placement. Stents were replaced after 6 weeks and subsequently exchanged every 3 months to avoid cholangitis. Percutaneous transhepatic catheterization was performed by injecting the contrast medium from the right intercostal approach. A right or left approach for the percutaneous transhepatic biliary drainage was chosen depending on ultrasound images illustrating the biliary anatomy, and the possibility of puncturing a dilated intrahepatic bile duct. Catheterization of intrahepatic bile ducts was performed in standard fashion. A guide wire was advanced through the biliary stricture into the duodenum. When this was achieved, a biliary drainage catheter was inserted. All drainage procedures were performed with the administration of broad-spectrum antibiotics. In case of a surgical reconstruction, the procedure was performed via a Roux-en-Y hepaticojejunostomy. The stricture in the CBD is transected and the hilar plate is opened. The hepatic ducts of different segmental bile ducts are mobilized and from there opened over the left hepatic duct. Intrahepatic segmental ducts are mobilized and if possible sutured together before one or two jejunal anastomosis are made. A closed suction drain is placed during operation and removed 24–48 hours after surgery. Percutaneous transhepatic drains, when inserted before surgery are left in place and removed after 10 days till 6 weeks, depending on the clinical course, the level of anastomosis and the surgeons’ preference. Outcome Follow-up data was obtained through outpatient records and the records of the general practitioner. The outcome of treatment was analyzed by the number complications and late restenosis during follow-up. Failure of treatment was defined as recurrent stenosis after stent therapy followed by surgery or recurrent stenosis after surgical reconstruction followed by additional therapy. Statistical Analysis Data from patient characteristics, management, and outcome show descriptive statistics in number of patients and percentages. Mean and median values are given with a minimum and maximum. Long-term stricture-free survival was analyzed by Kaplan Meier Survival Analysis. Data analyses were performed using SPSS® software (SPSS, Chicago, Illinois, USA). Results Patients’ Characteristics at Referral The referral pattern of BDI patients (n = 500) throughout the last 15 years are summarized in Fig. 1. From the total of 500 patients, 56 (11.5%) underwent a primary EEA. Patient characteristics are listed in Table 1. The laparoscopic cholecystectomy (n = 48, 86%) was converted in all patients. An open cholecystectomy was performed in eight patients (14%). In 49 patients (88%), the anastomosis was made over a T-tube. The tube was removed at the referring hospital or at the AMC after a mean of 52 days (range 2–145 days). Figure 1Referred patients for treatment of bile duct injury. Total number of referred patients (red), patients referred after a primary end to end anastomosis (blue), and patients referred after a primary biliodigestive reconstruction (green).Table 1Patient Characteristics Primary EEAn = 56%Age at cholecystectomyMean (years)52GenderFemale4377Indication for cholecystectomySymptomatic cholelithiasis4580Cholecystitis59Cholecystitis a froid61Type of initial operationOpen procedure814Laparoscopic to open procedure4886Anastomosis over T-tube4988Duration of T-tube in situDays, median (range)42(2–145) After the primary EEA, 19 patients (34%) underwent other therapeutical interventions before referral (Table 2) (Fig. 2). These patients underwent a range of one to three procedures before referral (median 2). The interventions included a relaparotomy in two patients (4%), percutaneous drainage of fluid collections in five patients (9%), endoscopicially placed stents in 12 patients (21%), a papillotomy in nine patients (16%), and percutaneous transhepatic drainage in two patients (4%). The median interval from the primary EEA to referral was 16 weeks (range 0–141 weeks). At referral, a biliary stricture was diagnosed in 38 patients (68%); in 10 patients (18%), bile leakage was diagnosed and combination of both in eight patients (14%). Symptoms at referral were cholestasis (n = 14, 25%), cholangitis (n = 10, 18%), and abdominal pain (n = 15, 27%). Three patients were referred because of uncontrolled sepsis (n = 2) and peritonitis (n = 1). According to the Bismuth classification, the majority of injuries (leakage of stricture) (n = 47, 84%) was located below the bifurcation. In nine patients (16%), the injury (mostly strictures) involved the bifurcation or the right or left hepatic duct (i.e., Bismuth classification grades IV and V). Table 2Referral Pattern Primary EEAn = 56%Time interval between injury and referral Weeks, median (range)16 (0–141)Intervention after EEA and before referral Explorative relaparotomy24 Percutaneous drainage59 Endoscopic stenting1221 Endoscopic papillotomy916 PTDa24Symptoms at referral Cholestasis1425 Cholangitis/fever1018 Abdominal pain1527 Abces/biloma47 Uncontrolled sepsis/peritonitis35Diagnosis at referralStenosis3868Leakage1018Combination of stenosis and leakage814Location of injury at referralb  I916  II2138  III1730  IV712  V24aPercutaneous transhepatic drainagebAccording to Bismuth classificationFigure 2ERCP showing successful (aggressive) stent therapy after primary EEA. a Stenosis of the common bile duct. b Stents in situ. c After stent removal within a year. Management after Referral Diagnostic work-up was performed by CT-scan (n = 9; 16%), endoscopic cholangiography (n = 38; 68%), and transhepatic cholangiography (n = 9; 16%). The definitive treatment of BDI patients after EEA is shown in the flow diagram (Fig. 3). After work-up, three patients (5.3%) were treated with percutaneous transhepatic cholangiographic drainage (PTCD) and 40 patients (71.4%) were treated endoscopically. Thirteen patients (23%) underwent reconstructive surgery after work-up; eight patients because of a complete stenosis of the CBD, in three patients reconstructive surgery was performed after failure of stent therapy at the referring hospital and in two patients because of a percutaneous fistula and persistent bile leakage. Figure 3Flow diagram of the success and failure rates after a multidisciplinary treatment of patients who underwent a peroperative end to end anastomosis for bile duct injury. Given percentages are calculated from the number of patients in the previous flow box. PTCD Percutaneous transhepatic catheter dilatation. Radiological and Endoscopic Treatment Three patients were successfully treated by PTCD. In two patients, a stenosis was treated by transhepatic dilatation and in one patient, bile leakage was treated by external transhepatic stent insertion. Forty patients (71.4%) were treated endoscopically (Fig. 2). In 37 patients (66%), stent insertion was successful and in three patients (5%), adequate drainage succeeded by papillotomy. The median number of stent replacements was five (range 1–15) with a median duration of treatment of 359 days (range 39–1,355). Complications occurred in nine patients (24%). Stent dislodgment (n = 3), clogging (n = 2), and cholangitis (n = 5) were mild complications and were successfully treated by stent exchange or administration of antibiotics. One severe complication occurred in a 75-year-old patient. After 4 years of stent therapy, the stent migrated and perforated the duodenum. Finally, the patient died due to multiple organ failure and sepsis. Surgical Treatment After referral and during the follow-up period, a new hepatobiliary anastomosis was performed by hepaticojejunostomy in 13 patients (23.2%). Mean duration of hospital stay was 9.1 ± 3.1 days. Postoperative complications occurred in one patient (7.6%) who underwent a PTC procedure after leakage of the anastomosis. No hospital mortality occurred in patients who underwent a reconstructive procedure after a previous EEA. Long-term Follow-up After a mean follow-up of 7.1 ± 3.3 years, seven patients (13%) have died. One endoscopically treated patient died due to a complication of treatment as described above. The other patients died due to malignancy (n = 4) and myocardial infarct (n = 2). The long-term results in patients treated with endoscopic and radiological treatment are as follows: from a total of 43 patients treated with endoscopic or PTCD procedures, 86% (n = 37) was successful. In three patients (7%), signs of restenosis occurred after stent removal after 2, 3.5, and 4 months. Continued stent therapy was successful in all three patients. Five patients underwent reconstructive surgery after prolonged endoscopic stenting. Postoperative complications occurred in two of the five patients and these patients received additional therapy for wound infection (n = 1) and postoperative cholangitis (n = 1). The long-term results of surgical treatment after EEA are as follows; from 13 patients who underwent a HJ after work-up, a stenosis of the anastomosis occurred in two patients (15%). Both patients underwent successful percutaneous transhepatic dilatation, respectively 9 and 35 months after surgery. The overall 5 years stricture free survival in the total cohort (n = 56) is 91%, shown by a Kaplan Meier curve in Fig. 4. Figure 4Kaplan–Meier plot showing proportion of patients without restenosis among 56 bile duct injury patients treated for complications after EEA. Discussion The present study describes a selected group of BDI patients, who were referred for treatment after a complicated EEA. This group of patients is a negative selection, representing the worst complications of EEA; otherwise, patients were not referred for additional treatment. So, this study does not provide any information about the success rate of EEA. The present study shows a long-term stricture free survival of 91% in EEA patients after treatment in a tertiary center. The analysis showed that even the majority of complications after primary AEE in a general hospital can successfully be treated by endoscopic and radiological interventions. In only one-third of the patients, a secondary surgical repair is necessary. The surgical reconstruction after EEA was associated with acceptable morbidity and without mortality. Around 40 to 45 patients are referred annually without any sign of decrease over the last years. Considering 15.000 LC’s per year in the Netherlands, we still consider 0.4–0.5% mentioned in the reviews as an underestimation of the real incidence of BDI, at least in the Netherlands.19 In 20% of the patients referred to the AMC, the injury was detected during the initial surgical procedure. This finding is similar to reports in literature.6,20 From the total of 500 referred BDI patients, 11.2% was referred for the treatment of a complication after peroperative EEA. Because referred patients only represent the complications after EEA, we do not know the real incidence of EEA procedures in BDI. Peroperative repair in BDI detected during surgery can be performed by EEA (with or without the use of a T-tube) or by a primary HJ. A HJ is a more complex procedure and one should be very careful not to further extend the injury into the intrahepatic ducts or subsequently damage the arterial supply (bleeding and clipping or ligation of right hepatic artery).21,22 The present study shows that if complications occur after EEA, these can successfully be treated by percutaneous or endoscopic balloon dilatation and/or stenting in the majority of patients. A HJ in the acute setting without dilated bile ducts is even more difficult and therefore consulting a surgeon with experience in reconstructive hepatobiliary surgery is recommended. In contrast with a primary HJ an EEA is a relatively simple procedure and can also be performed in less experienced hands. The risk to increase damage is smaller in an EEA procedure and with the use of a T tube instant bile drainage is realized. If indicated, reconstructive surgery by means of an elective HJ can be performed. It is strongly advised to perform a HJ after classification the injury and analyzing the biliary anatomy. Preoperative cholangiography (with the use of the T tube) will illustrate the location of the stenosis and the extension of dilation of the proximal bile ducts. A reconstructive procedure for stenosis of EEA has a satisfying outcome, as peroperative conditions are good after the inflammation has subsided and the bile ducts are dilated due to stenosis. In a situation in which peroperative bile leakage is due to (extensive) tissue loss, in particular, in patients with more proximal lesions at the bifurcation or intrahepatically, no primary repair should be performed. In this situation, adequate drainage of the upper right abdomen is strongly advised and the patient should be referred for elective reconstruction. Referral to tertiary center in this situation has a positive effect on outcome.2 End-to-end anastomosis is reported to be associated with a high incidence of recurrent jaundice due to stricture formation of the anastomotic area.15 Therefore, some authors suggest that EEA is almost never appropriate if the bile duct has been completely transected,15,23 while others favor this strategy when there is no extensive tissue loss.18 Stent therapy for iatrogenic bile duct strictures has changed during the last decade and therefore the long-term outcome after stenting has improved.24 A more aggressive approach with more stents and smaller time intervals between stent changes is favored. With this new approach, 80% of the patients who undergo an ERCP for postoperative bile duct stenosis, have a 10-year stricture-free survival.25 Although complications occur at a significant rate, these are usually mild. The only severe complication occurred in the present series, due to a migrated stent, was not reported in previous series.24,25 After stent removal, recurrent stenosis develops in 20% of patients within 2 years of stent removal.25 Therefore, endoscopic treatment should be the initial management of choice for postoperative bile duct strictures. Without signs of improvement after endoscopic stenting, reconstructive surgery is indicated in otherwise fit patients. Of interest is the evaluation of the long-term stricture-free survival after treatment for complications after EEA. After a mean follow-up of 7.1 years, restenosis after treatment developed in 9% of the patients. In all patients who underwent initial endoscopic therapy, restenosis occurred a relatively short time after stent removal, diagnosed within 2 to 8 months follow-up. Therefore, endoscopic treatment is not associated with a high rate of long-term restenosis after stent removal. In two patients, a restenosis occurred within 3 years after a hepaticojejunostomy. Symptoms were cholestasis and cholangitis. In both patients, transhepatic dilatation was successful to resolve the stenosis. The long-term stricture-free survival of 91% in the present series provides evidence for a good outcome after treating complicated EEA patients. If BDI is detected during surgery, in particular if there is no extensive tissue loss, the local anatomy is clear and there is no inflammation, EEA could be considered as a sufficient treatment strategy. Patients with postoperative complications (stricture or leakage) should be treated by a multidisciplinary team of gastroenterologists, radiologists, and surgeons. Postoperative complications can adequately be managed by endoscopic or percutaneous drainage in two-third of the patients. Reconstructive surgery after a complicated EEA is associated with low morbidity and no mortality.

Osteoporos_Int-4-1-2267483

Loss of treatment benefit due to low compliance with bisphosphonate therapy

Summary Among 8,822 new female bisphosphonate users, non-compliant bisphosphonate use was associated with a 45% increased risk of osteoporotic fracture compared to compliant use (MPR ≥80%). Classifying compliance into five categories, fracture risk gradually increased with poorer compliance. These results emphasize the importance of treatment compliance in obtaining maximal treatment benefit. Introduction Osteoporosis is a systemic condition characterized by low bone mass and microarchitectural deterioration of bone tissue, leading to enhanced bone fragility and, consequently, an increased fracture risk [1]. Osteoporosis has clinical and public health importance because osteoporotic fractures are one of the most common causes of disability and a major contributor to medical costs in many regions of the world [2]. Bisphosphonates are potent inhibitors of the osteoclast-mediated resorption of bone and widely used to treat osteoporosis and reduce the risk of osteoporotic fractures [3]. Long-term therapy with bisphosphonates is required to realise the full benefits of this treatment [3, 4]. This is supported by a recent PHARMO study that showed that the risk of fractures was reduced by 26% after one year persistent bisphosphonate use and up to 32% after two year persistent use [5]. However, in daily clinical practice, 50% to 75% of patients discontinue bisphosphonate use already in the first year of therapy [5–8]. Moreover, non-compliance with bisphosphonates has also been reported to be a frequent issue, with rates varying from 35% to 65% [6, 9]. Reasons for low compliance with bisphosphonate treatment are the stringent regimen to minimize the risk of oesophageal irritation, the existence of drug-related gastrointestinal side effects, and the fact that osteoporosis is often asymptomatic in early stages [10]. Caro et al. showed that patients who were compliant with their osteoporosis medication, including bisphosphonates, estrogens and calcitonin, experienced a 16% lower fracture rate compared to non-compliant users [11]. The objective of this study was to investigate the risk of osteoporotic fractures associated with low compliance with bisphosphonates in more detail. In addition to dichotomizing compliance, we also classified compliance into five categories. Patients and methods Setting Data were obtained from the PHARMO Record Linkage System (PHARMO RLS) which includes several databases, among which drug dispensing and hospitalization data of more than two million residents of the Netherlands. The drug dispensing histories contain data on the dispensed drug, the type of prescriber, the dispensing date, the amount dispensed, the prescribed dose regimens, and the duration of use. All drugs are coded according to the Anatomical Therapeutic Chemical (ATC) Classification. The hospital records include detailed information concerning the primary and secondary diagnoses, procedures, and dates of hospital admission and discharge. All diagnoses are coded according to the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM). Study cohort The source population included all new users of bisphosphonates in the period from January 1999 until July 2004. These patients were not dispensed any anti-osteoporotic drugs (bisphosphonates, raloxifene, tibolone, estrogens, hormone replacement therapy, calcitonin or teriparatide) for at least one year before their first bisphosphonate dispensing. All female new users of alendronate (10 mg daily or 70 mg weekly) or risedronate (5 mg daily or 35 mg weekly) (i.e. the dosages indicated for the prevention or treatment of postmenopausal osteoporosis), aged ≥45 years or with diagnosed postmenopausal osteoporosis (ICD-9-CM code 733.01), without gaps in registration in PHARMO RLS, and with a registration of at least one year before and one year after starting bisphosphonate treatment were included in the study cohort. All study patients were followed from the first bisphosphonate dispensing until the occurrence of the outcome of interest, death, end of registration in PHARMO RLS or end of the study period, whichever event was earliest. Compliance Compliance with bisphosphonates during follow-up may change and was measured over 90-day intervals using the Medication Possession Ratio (MPR). MPR was defined as the sum of days’ supply of all alendronate and risedronate dispensings during or overlapping the 90-day interval divided by 90 potential days of bisphosphonate therapy. For each dispensing only the days’ supply covering a specific interval were counted. E.g. of a 100-days supply dispensed at follow-up day 60, only 30 days were counted when calculating the MPR over the first interval (day 0–90) and 70 days for the MPR over the second interval (day 91–180). The maximum value of a MPR was set at 1. Subdividing follow-up in larger intervals was assumed too rough and subdividing in smaller intervals was not feasible regarding the usual prescription length of 90 days. Osteoporotic fractures The outcome variable osteoporotic fracture was defined as hospitalisation for an osteoporotic fracture during follow-up. Hospitalisations were selected based on primary discharge diagnosis for probable (ICD-9-CM codes 805.2 (vertebral thoracic), 805.4 (vertebral lumbar), 820 (proximal femur), 812 (proximal humerus), 813.4 (distal radius/ulna), 814 (carpus)) or possible (ICD-9-CM codes 823.0 (proximal tibia/fibula), 807.0 (rib), 807.2 (sternum), 808.0 (pelvis)) osteoporotic fractures. Covariates For all patients, a history of osteoporotic fracture was assessed. Furthermore, we determined use of non-steroidal anti-inflammatory drugs, analgesics, benzodiazepines and antidepressants in the year after starting bisphosphonate treatment as these drugs are associated with an increased risk of fractures [12–14]. Similarly, we assessed use of thiazide diuretics, ß-blockers and statins in the year after start as these drugs are associated with a reduced risk of fractures [15–17]. Statistical analyses The association between low compliance with bisphosphonates and the risk of fractures was analysed univariately and multivariately using time-dependent Cox regression analysis to account for changing compliance over time. In this analysis, at the time of each fracture, the cumulative compliance up to and including the corresponding 90-day interval of women who had experienced a fracture was compared to the cumulative compliance of those women who remained fracture-free at this time. The cumulative compliance of the fracture-free patients was measured up to the interval of corresponding fracture. In a first analysis, compliance was dichotomized and a MPR <80%, i.e. non-compliant bisphosphonate use was compared to a MPR ≥80%, i.e. compliant use. In a second analysis, compliance was classified into five categories: <20%, 20–49%, 50–69%, 70–89% and ≥90% (reference). A Wald test for trend was performed. Assuming that the effect of bisphosphonates is not immediate, and in line with Caro et al. [11], fractures occurring in the first 182 days of follow-up were excluded from the analyses. In a sub-analysis, this exclusion period was varied. Furthermore, an additional analysis was performed excluding women with only one bisphoshonate dispensing during follow-up; in these patients it is less sure that they actually used the drug. The multiple regression models included age, history of fracture and all covariates that were univariately associated with the risk of fractures and significantly contributed to the multivariate model, i.e. inclusion of the covariate resulted in a change of the compliance hazard ratio (HR) of 5% or over, starting with the most potent covariate. HRs and 95% confidence intervals (CI) were estimated using SAS V8.2 UNIX (Cary, NC, USA). Results The study cohort included 8,822 new female users of daily or weekly alendronate or risedronate, who contributed a total of 22,484 person-years of follow-up during the study period. More than half of the patients were older than 70 years of age (n = 4,708, 53%) and about one quarter of the patients used corticosteroids (irrespective of quantity) (Table 1). The first bisphosphonate was mainly prescribed by the general practitioner (64%). Only 197 patients (2%) were hospitalized for an osteoporotic fracture in the year before starting treatment. Table 1General characteristics of new bisphosphonate users between 1 January 1999 and 30 June 2004 (N = 8,822)CharacteristicN%Age class (years)45–5488910.155–693,22536.5≥704,70853.4Mean age (years) ± sd 69.4 ± 10.3Year of start1999 / 20002,25025.52001 / 20022,76931.42003 / 20043,80343.1Initial bisphosphonateDaily bisphosphonate4,22247.9Weekly bisphosphonate4,60052.1First prescriberGeneral practitioner5,70464.7Internist, rheumatologist, orthopedist2,31326.2Other prescribers8059.1Hospitalization for an osteoporotic fracture in the year before start 1972.2Use of corticosteroids in the year before start (irrespective of quantity)2,38927.1Follow-up time (months)12–232,93933.324–352,26825.7≥363,61541.0 During follow-up, 216 patients (2%) experienced an osteoporotic fracture, of which 40 patients during the first six months and 78 during the first year. About two third of all osteoporotic fractures were located at the proximal femur. The percentage of patients with a MPR < 80% (i.e. non-compliant patients) increased from 34% (3,018/8,822) after six months of follow-up to 42% (3,743/8,822) after one year, 51% (3,014/5,883) after two years and 60% (2,149/3,615) after three years of follow-up. Subdividing MPR in classes (Fig. 1), the majority of patients either had a MPR ≥90% or a MPR <20%, with the first group decreasing and the latter group increasing over time. Fig. 1Compliance with bisphosphonates, classified in five MPR categories, after six months, one year, two years and three years of follow-up. MPR: medication possession ratio Dichotomizing compliance, a MPR <80%, i.e. non-compliant bisphosphonate use was associated with a 40% increased risk of osteoporotic fracture more than six months after starting treatment (HR 1.41 95% CI 1.04–1.91, adjusted for age and history of fracture (see Table 2 for association between co-variates and fracture risk)) compared to a MPR ≥80%, i.e. compliant use. Varying the exclusion period, i.e. excluding fractures occurring in the first year of follow-up or including all fractures during follow-up, yielded similar adjusted increased risks (HR 1.50; 95%CI 1.06–2.13 and HR 1.45; 95%CI 1.10–1.91, respectively). Excluding women who received only one bisphosphonate dispensing during follow-up also did not change the results (adjusted HR 1.45; 95% CI 1.05–1.99, based on 156 fracture patients and 7,758 fracture-free patients). Table 2Association between co-variates and fracture risk more than six months after starting treatment Fracture patients (N)Fracture-free patients (N)N%N%HRcrude95% CITotal1761008,606100Age (years) at start 45–5474.087910.21.00reference 55–692916.53,18937.11.160.51–2.66 ≥7014079.54,53852.74.282.00–9.15Year of start 1999/20008648.92,15825.11.681.04–2.71 2001/20026335.82,69131.31.550.97–2.49 2003/20042715.33,75743.61.00referenceHistory of osteoporotic fracture52.81902.21.520.63–3.70Comedication in the year after start1 Analgesics6838.62,13624.81.821.34–2.46 Benzodiazepines8045.53,32438.61.300.96–1.74 Antidepressants2916.51,04612.21.521.02–2.26 ß-blockers4726.71,92922.41.350.97–1.89HR: hazard ratio, CI: confidence interval1 Univariate non-significant comedications (i.e. non-steroidal anti-inflammatory drugs, thiazides and statins) are not shown. Classifying compliance into five categories, fracture risk gradually increased with poorer compliance (p-value <0.05 for trend). Compared to a MPR ≥90%, the adjusted risk of fracture increased from 1.2 times for a MPR between 50% and 90% to 1.8 times for a MPR less than 20% (Fig. 2). Fig. 2The relationship between compliance with bisphosphonates, classified in five MPR categories, and fracture risk more than half a year after starting treatment. HRs are adjusted for age and history of fracture and are compared to a MPR ≥90%. MPR: medication possession ratio, HR: hazard ratio, CI: confidence interval Discussion This study indicates that non-compliant bisphosphonate users had an approximately 50% higher likelihood of osteoporotic fracture compared to compliant users. Classifying compliance into five categories, fracture risk gradually increased with poorer compliance to an 80% risk increase with very low compliance compared to very high compliance. These results emphasize the importance of treatment compliance in obtaining maximal treatment benefit. To quantify the loss of treatment benefit with low compliance, information on fracture risk in women not treated with bisphosphonates is necessary. The above mentioned percentage risk increases are compared to compliant use and cannot be translated into percentage loss of treatment benefit. We chose not to use untreated patients as reference because of the heterogeneity of patterns of treatment use that exist within this group. In a meta-analysis of 11 trials of alendronate for the treatment of postmenopausal osteoporosis, a consistent 50% reduction in fracture risk was seen compared to placebo [18]. For risedronate, a meta-analysis of 8 trials revealed a 25 to 35% reduction in fracture risk compared to placebo [19]. In other words, untreated women are 1.5 to 2 times as likely to experience a fracture compared to women using bisphosphonates. Assuming that the patients with a MPR ≥90% in our study are comparable to the treated patients in trials, the 1.6 to 1.8 times increased fracture risk we observed with a MPR <50% can be translated into a 80 to 100% loss of treatment benefit. Apparently, patients with the lowest compliance are essentially comparable to non-treated patients. At present, in addition to Caro et al., others have reported on the relationship between compliance with bisphosphonates and fracture risk [11, 20, 21]. Siris et al. [20] found a relative fracture risk reduction of 25% with compliant use of alendronate or risedronate, i.e. a 33% risk increase with non-compliant use. There was a progressive relationship between compliance and fracture risk reduction, commencing at a MPR around 50% and becoming more pronounced at a MPR ≥75%. This finding, like our results, suggests that low-compliant patients gain only marginal benefit from bisphosphonate therapy. Among women starting drug therapy for osteoporosis (of which two-third bisphosphonates) Weycker et al. [21] found that compared to women with a MPR <30%, the odds of fracture was similar (1.02) for women with a MPR between 30 and 69% and women with a MPR between 70 and 89%, but lower (0.70) for women with a MPR ≥90%. The latter corresponds to a 1.4 times increased risk of fracture with a MPR <30% compared to a MPR ≥90%. Caro et al. [11] reported that compared to a MPR >90%, fracture risk was significantly increased at lower compliance levels, i.e. up to 1.4 times with a MPR ≤50%. However, it is not clear whether this effect was gradual. In a study population of mainly users of hormone replacement therapy, Huybrechts et al. [22] found that compared to women with a MPR ≥90%, fracture risk was similar (HR 1.09) for women with a MPR between 80 and 90%, but higher for women with a MPR between 50 and 80% and women with a MPR <50% (HR 1.18 and 1.21, respectively). Our associations are about similar than reported by Siris et al., but stronger than found by Caro et al., Weycker et al. and Huybrechts et al. The most likely explanation for this is that our and Siris’ study population was restricted to bisphosphonate users, while the others also included users of other, less potent anti-osteoporotic drugs (30% to 65% of the study population). There is a previous PHARMO study on bisphosphonate use and risk of fractures, using persistence [5]. Besides a different measure of drug exposure, the current study differs from this previous study in that the study period was extended, different inclusion criteria were applied and new patients were included besides previously studied patients. Although persistence and compliance are measured differently, with persistence referring to the duration of uninterrupted drug use, they are closely related. Patients classified as non-compliant would also have been classified as non-persistent and vice versa. The results of our current study are in line with our previous study in which the risk of fractures was reduced by 26% after one year persistent bisphosphonate use and up to 32% after two years of persistent use [5]. Similarly, Gold et al [23] reported that persistent bisphosphonate use for at least 6 months was associated with a 26% lower fracture risk. Expressing these results the other way around, non-persistent bisphosphonate use is associated with an about 40% fracture risk increase. Among a population of mainly users of hormone replacement therapy, one year of uninterrupted therapy reduced the likelihood of fractures with significant odds ratios of 0.38 for hip fractures and 0.60 for vertebral fractures [7]. Our analyses may be limited by some issues common to observational database research in daily clinical practice. Most importantly, although we considered a large number of known confounders in our analyses, including history of fractures, there may be other confounding factors for which information was not available, such as bone mineral density (BMD). BMD is an important determinant of fracture risk [24] and BMD testing has been shown to be positively associated with persistence and compliance [25, 26]. Consequently, the results of our analyses are likely to be conservative in the sense that compliant patients may have had an increased fracture risk because of low BMD, and history of fractures only partially controlled for this. Furthermore, it is possible that patients have started using other anti-osteoporotic drugs, e.g. raloxifene, during follow-up and stopped bisphosphonate use. This means that part of the patients who were non-compliant with bisphosphonates may have been protected against fractures by another anti-osteoporotic drug and the observed increased fracture risk with non-compliant bisphosphonate use therefore may be underestimated. However, we know from our database that this concerns only about 7% of patients in the first year of bisphosphonate treatment. In addition, compliance with treatment was based on dispensing data. As it is unknown whether a patient actually used the drug, compliance may have been overestimated. However, it seems likely that patients who obtain prescription refills do take their medication. Excluding women who filled only one bisphosphonate dispensing during follow-up from the analyses did not change the results. Another limitation relates to the classified analysis. With the use of five compliance categories studying a trend is possible but it does not give information on the point at which there is a significant shift in fracture risk. Ideally, 10% MPR categories should be used. However, patient numbers per 10% MPR categories (data not shown) did not allow for this detailed analysis. To increase the treatment benefit of bisphosphonates, compliance should be improved. Related to the inconvenient, stringent intake regimen, dosing frequency is an important determinant of compliance with bisphosphonates. In recent earlier studies, once-weekly dosing was associated with better compliance and persistence compared to daily dosing. However, even with the weekly regimen, compliance and persistence were suboptimal in about half of the patients [6, 8, 9]. Bisphosphonates with less frequent dosing regimes than weekly, e.g. monthly or three-monthly ibandronate or annual zolendronate, are either available yet or upcoming and may further improve compliance and therefore treatment benefit. In a cross-over trial, women with postmenopausal osteoporosis preferred once-monthly ibandronate therapy and found it more convenient than once-weekly alendronate therapy [27]. There are no data yet on the actual impact of this once-monthly, and other formulations on compliance with bisphosphonate therapy and reduction of fracture risk in daily practice. Other keys to improve compliance are e.g. healthcare provider - patient communication and continuous reinforcement of the importance of treatment. Reminder systems may help the patient comply with therapy, especially with infrequent dosing regimes. However, results from a review show that current methods of improving patient compliance for several chronic health problems are mostly not very successful [28]. Furthermore, although a number of demographic and clinical variables are associated with compliance with bisphosphonates these predictors, either individually or together, have been shown to have low predictive value for identifying patients who will become non-compliant [25]. In conclusion, the results of this study show a statistically significant association between level of compliance with bisphosphonates and level of fracture risk, emphasizing the importance of treatment compliance in obtaining maximal bone protection.

Skeletal_Radiol-3-1-2042033

Whole-body imaging of the musculoskeletal system: the value of MR imaging

In clinical practice various modalities are used for whole-body imaging of the musculoskeletal system, including radiography, bone scintigraphy, computed tomography, magnetic resonance imaging (MRI), and positron emission tomography-computed tomography (PET-CT). Multislice CT is far more sensitive than radiographs in the assessment of trabecular and cortical bone destruction and allows for evaluation of fracture risk. The introduction of combined PET-CT scanners has markedly increased diagnostic accuracy for the detection of skeletal metastases compared with PET alone. The unique soft-tissue contrast of MRI enables for precise assessment of bone marrow infiltration and adjacent soft tissue structures so that alterations within the bone marrow may be detected before osseous destruction becomes apparent in CT or metabolic changes occur on bone scintigraphy or PET scan. Improvements in hard- and software, including parallel image acquisition acceleration, have made high resolution whole-body MRI clinically feasible. Whole-body MRI has successfully been applied for bone marrow screening of metastasis and systemic primary bone malignancies, like multiple myeloma. Furthermore, it has recently been proposed for the assessment of systemic bone diseases predisposing for malignancy (e.g., multiple cartilaginous exostoses) and muscle disease (e.g., muscle dystrophy). The following article gives an overview on state-of-the-art whole-body imaging of the musculoskeletal system and highlights present and potential future applications, especially in the field of whole-body MRI. Introduction The skeletal system is a frequent target of metastatic spread from various primary tumors like carcinoma of the breast, lung and prostate cancer. Moreover, primary malignancies may also originate from the bone marrow, such as lymphoma and multiple myeloma [1]. Therefore, it is highly important to accurately assess manifestations of malignant diseases within the bone marrow in order to facilitate adequate therapy and predict prognosis. Only pronounced destruction of bone with loss of bone mineral content exceeding 50% is readily visible in radiographic examinations [2]. Computed tomography (CT) is definitely more sensitive than radiography and it is the image modality of choice to evaluate the extent of destruction of trabecular and cortical bone and to assess stability and fracture risk. Magnetic resonance imaging (MRI), on the other hand, allows bone marrow components, such as hematopoietic and fat cells, to be visualized. Moreover, tumor infiltration into the spinal canal and paravertebral soft tissues is clearly depicted. Compared with other imaging modalities like radiography, CT or bone scintigraphy, it is the most sensitive technique for the detection of pathologies restricted to the bone marrow, even if trabecular bone is not destroyed [3, 4]. It has been reported that up to 40% of skeletal metastases occur outside the field of view covered by a routine assessment of the axial skeleton, underlining the importance of whole-body bone marrow imaging [5]. In the past, different requirements for patient positioning and coil set-up complicated the introduction of MRI as a practicable whole-body application. With multi-channel whole-body MRI (WB-MRI) scanners, however, head-to-toe assessment of the whole skeletal system has become a realistic option without compromises in image quality compared with dedicated examinations of limited anatomical areas. Beyond the assessment of malignant bone neoplasms, WB-MRI has recently been proposed for the whole-body imaging of systemic muscle disease and may prove useful for an evidence-based screening of patients suffering from diseases that predispose to bone malignancy (e.g., multiple cartilaginous exostoses). Technical aspects of whole-body MRI Due to its lack of ionizing radiation MRI seems suitable for whole-body imaging, but for a long time its primary application has been the assessment of focal pathologies within particular organs and body parts. The most severe challenges of WB-MRI in the past have been long examination times, mainly caused by time-consuming patient repositioning and changing of the array configuration. Initially, the sequential scanning approach for WB-MRI of the skeletal system consisted of separate scanning steps of T1-weighted and STIR (short tau inversion recovery) imaging at five body levels with at least one patient repositioning process using conventional head, neck, body, and spine array coils. For a complete whole-body examination, including dedicated imaging of the spine in sagittal orientation, a total room time of at least 60 min had to be taken in account. Steinborn et al. introduced for the first time this whole-body bone marrow scanning concept for the screening of bone metastases and, despite the considerable complexity of the examination, reported advantages in diagnostic accuracy for MRI compared with conventional WB imaging techniques, like skeletal scintigraphy [6]. Later, attempts to overcome FOV restrictions and increase patient comfort were based on a rolling platform concept mounted on top of the scanner table, making the patient glide in between a “coil sandwich” comprised of the body coil and the integrated spine coil (AngioSURF™/BodySURF™; MR-Innovation, Essen, Germany). However, with this approach considerable compromises in spatial resolution, especially in peripheral body regions like the head/neck and lower extremities, had to be tolerated [7]. With the introduction of multi-channel MR scanners, using a system of multiple phased array coils covering the whole body like a matrix, imaging of the total skeletal system without compromises in spatial resolution became possible. In particular, the combination of free table movement with parallel imaging acquisition techniques (PAT) resulted in substantially shorter room time and allowed to integrate otherwise time-consuming, but indispensable sequence types for bone marrow imaging (e.g., STIR sequences). The proposed imaging protocol for high resolution T1-weighted turbo spin echo (TSE) and STIR imaging from head to toe is performed with 1.3 × 1.1-mm and 1.8 × 1.3-mm plane resolution respectively (5-mm slices, matrix 384, PAT 2–3). Additionally, dedicated T1-weighted TSE and STIR imaging of the complete spine is performed (1.0 × 1.0-mm plane, 3-mm slices, matrix 384), which results in a total scan time of 43 min. A promising new concept for WB-MRI is the continuously moving table technique with the use of PAT. Recently, a SENSE reconstruction algorithm has been successfully applied on stationary receiver coils with arbitrary coil dimensions for continuously 3D gradient echo imaging of the complete body without significant constraints in image quality [8]. Zenge et al. have reported promising initial results for 3D whole-body continuous data acquisition using the rolling table concept as a new potential strategy for WB-MRI metastases screening, especially for large field-of-view imaging in short bore systems [9]. Technical aspects of whole-body CT Multislice CT (MS-CT) is frequently used in oncologic imaging, and in the detection of bone destruction CT is far more sensitive than radiography [10]. Moreover, it is unique in its ability to evaluate the extent of osseous destruction and to assess stability and fracture risk [11]. Whole-body imaging usually implies CT of the neck, thorax, abdomen/pelvis in axial orientation with multi-planar reconstructions and recalculation of bone window setting. Continuous hardware improvements for CT imaging from single-slice scanners to 64-slice scanners have resulted in larger FOV and faster acquisition times with high resolution. Examination time now is reduced to 1–2 min and improvement in resolution with the resulting formation of isotropic voxels allows high quality multiplanar reconstructions. In particular, the use of ultrathin collimation (0,5 mm) can provide excellent image quality for the neck and peripheral skeleton. Furthermore, with automated tube current dose modulation systems exposure of the patient to ionizing radiation could be reduced by 10–68% on average, depending on the anatomical region, without substantially sacrificing image quality [12]. Absolute dose reduction (described as volume CT dose index) ranges from 11.97/8.18 mGy for a pelvic examination to 23.28/7.45 mGy when a combined angular and z-axis modulation system is used. Multislice CT enables bone destruction to be detected as well as osteoblastic and mixed patterns with osteosclerosis and destruction. In cases of diffuse bone marrow infiltration inhomogeneous osteoporosis may be detected. However, as in radiography, this may mimic osteoporosis, unless areas of bony destruction are present. 3D reconstruction algorithms, like VRT (volume rendering technique), can be used to display complex fractures (e.g., pathological fracture) and further increases the reliability to detect occult vertebral fractures compared with plain radiographs. Furthermore, the 3D morphology of vertebral fractures may give indications concerning the nature of the fracture (e.g., tumorous vs. osteoporotic fracture). Clinical application of whole-body MRI Whole-body MRI is increasingly used in the field of oncologic imaging as an adjunct or alternative to established multi-modality approaches (e.g., radiographs, MS-CT, ultrasound, scintigraphy) for initial tumor staging or screening for tumor recurrence after curative therapy. Promising results have been reported for the detection of distant metastatic disease, especially in tumors that frequently metastasize to the bone, liver, and brain (Fig. 1) [13, 31]. Recently, WB-MRI has been proposed as a sensible application for a more integrated assessment of multiple myeloma and systemic muscle diseases [16, 17, 18, 19]. Fig. 1A 60-year-old patient with a malignant melanoma. a Coronal T1-weighted turbo spin echo (TSE) whole-body (WB) magnetic resonance imaging (MRI) indicates multifocal metastatic disease of the liver (arrow). b Whole-body positron emission tomography-computed tomography (PET-CT) shows multiple areas of pathologic [18F]-fluorodeoxyglucose (FDG)-uptake in the right atrium (arrow), the liver (arrowheads), and in the right iliac bone (arrowhead). c Contrast-enhanced CT reveals a metastasis within the right atrium. d T1-weighted TSE WB-MRI confirms a hypointense lesion in the right iliac bone indicating bone metastasis. e Axial fat-saturated contrast-enhanced MRI of the pelvis shows another bone metastasis in the right sacral bone (arrow). f, g The corresponding CT images in the bone window setting show extensive osteolysis within the right iliac bone; however, no morphologic changes are found in the right sacral bone Metastasis Based on morphological criteria in radiography, CT, and MRI, skeletal metastases are classified as osteolytic (approximately 50%), osteoblastic (35%) ,and mixed type (15%). For MRI bone screening, the combination of unenhanced T1-weighted spin echo and turbo-STIR sequences proved to be highly sensitive in discriminating benign from malignant marrow disorders [20]. On T1-weighted sequences tumor spread is identified by replacement of normal fat containing marrow, resulting in a hypointense signal . Fat-suppressed sequences, such as STIR, depict neoplastic lesions by virtue of the hyperintense signal due to increased content of water within the tumor cells (Fig. 2). However, osteoblastic metastases may be depicted in STIR sequences with variable signal intensities from hypointense in dense sclerotic lesions to hyperintense when more cellular components are present [21]. The unique soft-tissue contrast of MRI allows precise assessment of tumor infiltration within the bone marrow and even diffuse infiltration of the bone marrow with neoplastic cells, not associated with focal bone destructions or formation of new bone, is detected [22]. In some cases additional sequences, such as dynamic studies of signal enhancement after gadolinium injection, may be performed for accurate differentiation from benign bone marrow changes, such as hyperplastic bone marrow formation [23]. Finally, MRI allows for precise assessment of adjacent paraosseous structures, such as the spinal canal. Whole-body MRI compared with bone scintigraphy In clinical practice multi-modality algorithms are most commonly used when metastatic bone disease is suspected. They include radiography, bone scintigraphy, PET, CT, and MRI. In many institutions 99mTc-phosphonate-based bone scanning is performed as the method of initial bone marrow screening. However, scintigraphy provides only limited spatial resolution and at an early stage of disease lesions may remain invisible in the absence of an osteoblastic response [24]. Furthermore, misinterpretation of tracer uptake in healing fractures or degenerative disease may lead to false-positive findings. The diagnostic performance of MRI compared with bone scintigraphy for the detection of skeletal metastases has been examined in various studies and higher specificity and sensitivity in the early detection of skeletal metastases have been reported [6, 24, 25]. Steinborn et al. compared combined T1-weighted and STIR WB-MRI with bone scintigraphy using the sequential scanning approach in a lesion-by-lesion analysis [6]. WB-MRI reliably detected more confirmed skeletal metastases (91%) than bone scintigraphy (85%). Another study group reported higher sensitivity and specificity of WB-MRI (92%/90%) compared with scintigraphy (83%/80%) on a patient-by-patient basis [20]. Daldrup-Link et al. analyzed both methods for the detection of bone metastases in 39 children and young adults and observed a higher sensitivity of 82% for WB-MRI compared with 71% for scintigraphy, with this difference increasing, especially in medium-sized lesions between 1 and 5 cm [3]. Interestingly, both methods showed differences in performance according to lesion location. Most false-negatives in scintigraphy were found in the spine, while diagnostic problems for WB-MRI occur in the thoracic cage and skull, especially when coronal imaging orientation is used, a problem that is certainly increased by motion artefacts. These problems might be overcome when using fast turbo spin echo sequences for thoracic imaging in combination with axial slice orientation. Whole-body MRI compared with MS-CT Multislice CT is far more sensitive than radiography in the assessment of focal bone destructions. Krahe et al. compared radiography and CT examinations in 112 patients with metastases of the spine [10]. MS-CT identified 268 involved vertebrae while radiography depicted 88% of these lesions when the vertebral body was involved and only 66% when other parts of the vertebrae were affected. Intraspinal and paravertebral tumor extension was correctly assessed by plain radiography in only 23% and 33% of cases respectively. Recently, multidetector MS-CT has been proposed for whole-body screening of the skeletal system as an alternative to bone scintigraphy [26]. Groves et al. investigated 43 patients with suspected bone metastases using bone scintigraphy and 16-detector MS-CT and detected metastatic disease in 14 out of 43 and 13 out of 43 patients respectively, with an agreement of 84%, suggesting a similar diagnostic performance of both modalities. The authors concluded that in case of suspected skeletal metastasis, scintigraphic studies may not be needed, when an adequate whole-body MS-CT tumor staging with recalculation of bone window setting has been performed. Consecutively, this might shorten diagnostic pathways and save resources. However, it is not yet clear whether MS-CT is equal to MRI for the assessment of bony metastases. First results have demonstrated the superior detection rate of MRI (Fig. 1) [27]. Whole-body MRI compared with PET-CT In contrast to MRI and MS-CT, positron emission tomography (PET) using [18F]-fluorodeoxyglucose (FDG) provides functional information by tracing increased FDG uptake directly into the tumor cells. Recent studies indicate that whole-body FDG-PET increases specificity of bone screening compared with bone scintigraphy, although there is conflicting evidence whether there is a significant gain in sensitivity [28]. Still, FDG is a tracer that is not tumor-specific and may also accumulate in the presence of inflammation and thus lead to false-positive findings. Also, FDG is not suitable for several tumor entities due to poor tracer uptake, e.g., prostate cancer, myxoid tumors of the gastrointestinal tract, low-grade sarcomas or renal cell carcinomas [29]. A clear technical disadvantage of PET, similar to scintigraphy, is its poor anatomical resolution, which often makes the exact localization of a lesion difficult. Fused PET-CT scanners combine the functional data of PET with the detailed anatomical information of MS-CT scanners in a single examination and have further improved diagnostic accuracy. Various authors have reported a significant decrease in ambiguous lesions and an improvement in the specificity of PET-CT compared to that of PET alone in the detection of malignant disease [14, 30]. Only a few study groups have directly compared the performance of WB-MRI with FDG-PET-CT in the detection of skeletal metastases [15, 31]. Antoch et al. analyzed the accuracy of both modalities in 98 patients in terms of TNM-based tumor staging. Both imaging procedures revealed a similar diagnostic sensitivity to the detection of distant metastases (WB-MRI  93%, PET-CT 94%). Regarding bone metastases, the sensitivity was significantly higher when using WB-MRI (85%) instead of PET-CT (62%). Our own observations confirm these data. We examined 41 patients with suspected skeletal metastases with both FDG-PET-CT and WB-MRI using a multi-channel scanner with PAT [31]. In a lesion-by-lesion analysis 102 malignant and 25 benign bone lesions were confirmed by histology or follow-up. WB-MRI showed a significantly higher diagnostic accuracy than PET-CT (91% vs. 78%, Figs. 1, 2). In particular, lesions smaller than 5 mm were visualized by WB-MRI with a cut-off size of 2 mm for WB-MRI compared with 5 mm for PET-CT. Lesions less than double the size of the spatial resolution of the PET scanner (usually 6 mm) can especially lead to false-negative results. Additionally, 10 bone metastases in distal parts of the body were revealed by WB-MRI due to the larger field of view (the FOV of a routine PET-CT is restricted to the diagnostic spiral CT, usually ranging from the skull base to the proximal femora). However, specificity was higher in PET-CT (PET-CT 80% vs. WB-MRI 76%). Here, certainly, the additional metabolic information of PET plays the most important role in reliably discriminating between malignant and benign lesions (e.g., atypical haemangioma) [32]. Fig. 2A 28-year-old man with non-Hodgkin’s lymphoma. a Coronal whole-body STIR imaging shows suspect areas with hyperintense signal in the right clavicular region and both iliac bones. b Magnification shows a mass extending from the right clavicular fossa and infiltrating the right apical thoracic cage. c The lesion shows a pathological FDG uptake in PET-CT, indicating malignancy. d The T1-weighted sagittal image of the spine shows extensive multifocal lymphoma manifestations. e PET-CT underestimates the degree of tumor involvement by showing tracer uptake only in the thoraco-lumbar region On the other hand, at least in breast cancer, different patterns of FDG uptake have been reported in osteoblastic, osteolytic or mixed lesions, indicating that sclerotic lesions may be less FDG-avid [33]. The additional morphologic information of PET/CT compared with PET alone is certainly of great value in increasing diagnostic sensitivity. Also, significant improvement in diagnostic accuracy has been reported when 18F-fluoride is used for PET or PET-CT for the assessment of malignant skeletal disease. 18F-fluoride is a tracer that, similar to 99mTC-diphosphonate, specifically adsorbs onto bony surfaces with a predilection for sites of active bone formation [30]. An important indication in bone imaging is lesion monitoring after chemotherapy or radiation therapy. It has to be taken into account that on MRI necrotic bone metastases may remain virtually unchanged in morphology or signal characteristics, which may make evaluation of therapy response difficult. Compared with viable tumors, the contrast enhancement is frequently less pronounced and slower in tumors that have responded to a particular therapy. Reduction in tumor size may be delayed and is not a sensitive sign in the assessment of response. Tumor metabolism, and consequently FDG uptake, is highly susceptible to chemotherapy. On the CT image, osteolytic metastases often demonstrate typical sclerotic transformation. In this setting, future studies are needed to evaluate the sensitivity of FDG-PET-CT in assessing the response to chemotherapy and whether sensitivity is reduced in the first days after initiation of therapy. After radiation therapy MRI is particularly helpful, because irradiated lesions are easily distinguishable from new lesions because of the high signal of normal bone marrow on T1-weighted SE images, which is due to conversion of hematopoietic into fatty bone marrow. Multiple myeloma Multiple myeloma is a frequent neoplastic disease affecting the bone marrow. It is characterized by expansive growth of malignant plasma cell clones with consecutive destruction of the bony architecture. Predilection sites are the axial skeleton (spine and pelvis), but also the ribs, the shoulder region, skull, and proximal femurs, stressing the need for total body imaging for adequate assessment of the extent of disease. The bone marrow may either be diffusely infiltrated or there may be focal accumulations of atypical plasma cells (Fig. 3). Fig. 3A 70-year-old man with multiple myeloma. a The radiograph of the pelvis is inconspicuous. b Coronal MS-CT reconstruction of the pelvis in a bone window setting reveals a large area of destruction within the left iliac bone (arrow). c) STIR-WB-MRI confirms focal tumor manifestation within the iliac bone (arrow) and reveals multiple small nodular infiltrations within the sacral bone and pelvis. d MS-CT of the spine shows a compression fracture of Th9. e, f T1-weighted SE- and STIR imaging of the spine reveals diffuse myeloma infiltration of the spine. g Coronal STIR sequences of the pelvis show additional focal infiltration of the left femoral head missed on radiography and MS-CT In patients with myeloma the basic diagnostic work-up in many institutions includes radiographic examinations of the skull (two planes), the rib cage, the upper arms, the spine (two planes), the pelvis, and the upper legs. This diagnostic approach is still represented in the classic Salmon and Durie staging system of the disease, which includes radiographic, immunohistochemical, and serological factors of the disease, and defines the selection of adequate therapy [34]. However, diagnostic sensitivity of radiography in the detection of myeloma manifestations is rather low and thus allows diagnosis only at advanced stages of the disease when at least 50% of the bone mineral content has been lost [2]. In particular, a diffuse bone marrow infiltration pattern may cause diagnostic problems as it can easily be misdiagnosed as senile osteoporosis. In a study conducted by Baur et al. 55% of focal and 59% of diffuse infiltrations were missed by radiography [35]. With the use of contrast-enhanced sequences and calculation of percentage signal increases the sensitivity of MRI for diffuse myeloma infiltration, especially at earlier stages, can be further increased [36]. Schreiman et al. examined 32 patients with multiple myeloma using single-row CT and radiography [37]. Twelve patients showed osseous affections in both modalities. CT, however, usually demonstrated a more extensive involvement and in 6 out of 13 patients only CT detected myeloma involvement despite normal radiographs. Another study group recently proposed FDG-PET-CT as an alternative to radiographic imaging of the whole body [38]. PET-CT detected more lesions in 16 out of 28 patients (57%) and 9 of these patients had completely negative radiography. Ghanem et al. compared WB-STIR-MRI using the conventional radiographic skeletal survey in 54 patients with plasma cell neoplasms [16]. Whole-body MRI correctly revealed bone marrow infiltration in 74% of patients, while radiography depicted pathological changes in 55% of patients. Moreover, WB-MRI showed a larger extent of infiltration in 90% of concordant findings. In two studies, the diagnostic accuracy of MS-CT and MRI was analyzed. Mahnken et al. compared results of four-detector MS-CT and MRI examination of the spine and pelvis in 18 patients with stage III multiple myeloma [39]. A total number of 325 vertebral bodies were evaluated. In MS-CT 231 vertebral bodies were classified as “affected”, compared with only 224 vertebral bodies in MRI. These “false-negative” findings might have been a consequence of response to previous treatment. However, no information about previous therapy was provided in this study. On the other hand, MRI showed 5 affected vertebrae that were considered normal on MS-CT. The false-negatives on MS-CT might be due to early bone marrow infiltration without signs of osseous destruction. Furthermore, the surprisingly high sensitivity of MS-CT in this study may also be caused by pre-selection of a patient group with advanced disease. In our own study, 30 patients with multiple myeloma were examined with a 1.5-Tesla multi-channel WB-MRI scanner and a 16- and 64-detector MS-CT scanner [17]. MRI showed superior diagnostic accuracy to MS-CT. In particular, the diagnostic sensitivity of MS-CT was inferior to that of WB-MRI (Fig. 3). The high amount of false-negatives on MS-CT may be explained by the fact that early stages of the disease can be visualized on MRI, displaying bone marrow replacement before any destruction of trabecular and cortical bone occurs. False-positive results on MS-CT, on the other hand, may be due to misinterpretation of inhomogeneous osteoporosis as diffuse myeloma infiltration. Baur et al. were able to show that an extended staging system, including MRI of the spine, has a significant influence on the assessment of prognosis. Using the classic staging system of Durie and Salmon without MRI, 25 out of 77 patients would have been understaged, thus underlining the importance of incorporating the MRI bone marrow status into staging and therapy planning of this disease [40]. Muscle disease Diagnostic imaging of muscle disease (e.g., polymyositis or muscle dystrophy) represents a challenge as it requires high-resolution whole-body coverage of soft tissue structures to adequately assess the pattern, distribution, and extent of the disease. CT, scintigraphy, and ultrasound are not suitable for this purpose as all these modalities lack sufficient soft tissue contrast, spatial resolution or large FOV imaging options. Other diagnostic tests frequently used, like electromyography, are unspecific and muscle biopsies used as the gold standard have been reported to be false-negative in 10–25% of cases due to sampling errors [41]. MRI, with its precise delineation of fat, muscle, and bone is an ideal candidate for imaging of systemic muscle disease. Moreover, substantial dose exposure in a predominantly younger patient cohort commonly affected by muscle diseases can be avoided. Previously, MRI has been used for a targeted assessment of muscle disease [41, 42]. A combination of T1-weighted SE and T2-weighted pulse sequences with fat saturation, such as STIR, are indispensable for assessing increased fat and water content within the muscle tissue respectively (Fig. 4). Contrast enhancement with Gadolinium chelates also adds to diagnostic accuracy, especially when combined with fat suppression. Since various muscle groups and body parts may be involved in an unpredictable distribution, coverage of the whole body is important. Fig. 4a Whole-body MRI of a 60-year-old man suffering from severe progressive muscle dystrophy, symmetrically affecting the muscles of the rib cage, abdomen, pelvis, and lower extremities. b WB-STIR shows signs of diffuse reactive edematous processes in the left upper thighs and muscles of the distal lower limbs. c, d T1-weighted imaging displays fatty tissue replacement affecting most of the muscles of the pelvis. Residual adductor muscle tissue is shown with high contrast due to its hypointense signal. e Progressive muscle dystrophy has also occurred in the lower extremities, affecting all compartments O’Connell et al. previously described WB-STIR imaging for the diagnosis of patients suffering from polymyositis [18]. With the use of parallel imaging techniques WB-STIR imaging can be performed in under 13 min at an inplane resolution of 1.3 × 1.1 mm. With this method, symmetry of muscle involvement and the extent and severity of inflammation can be analyzed, so that an adequate target for biopsy can be identified. During therapy, WB-MRI is particularly helpful in monitoring the disease outcome by visualizing the muscle-to-fat tissue relation or to verify a decrease in inflammation. Also, various drugs, such as steroids, may have side effects that involve the muscles and result in various types of myopathy. Lenk et al. proposed coronal and axial WB-STIR imaging in combination with coronal T1-weighted SE imaging and sagittal T1-/T2-weighted imaging of the spine, as a sensible protocol for systemic muscle disease, resulting in an approximate total scan time of 45 min [19]. However, assessment of distal parts of the upper extremity may be limited. This is due to positioning of the arms on the pelvis and thighs. If the muscles of the forearm are in the focus of interest, additional examination with dedicated coil systems may be required. Patients with muscular dystrophy, an X-chromosome recessive disease, suffer from progressive destruction of muscle tissue with subsequent replacement by fatty and fibrous tissue. In addition to the morphological information concerning the distribution and extent of replacement of muscle by fatty tissue as provided by WB-MRI, exact quantification of the muscle-to-fat relation is of substantial diagnostic and prognostic importance. In particular, T1-weighted axial WB-MRI in combination with CAD (computer aided diagnostics) applications may be useful for the quantification of total body fat distribution. Interesting results have been reported by Brennan et al. who introduced a fast, automated approach to body fat measurement in healthy individuals as a useful alternative to the body mass index [43]. Future applications Eustace et al. have proposed WB-MRI for trauma assessment in patients referred for suspected child abuse [44]. For a rapid assessment of acute trauma, especially when multiple organs are affected, MS-CT provides fast and detailed systemic information on potentially life-threatening organ, vessel or bone injury and is the imaging technique of choice. Recently, an accelerated triage MS-CT protocol has been proposed by Koerner et al. for time-effective handling of mass casualty incidents using MS-CT. For a scenario of 15 multiple trauma patients admitted over a period of 2 h within a mass casualty incident a mean total time in the scanner room of 8.9 min was calculated, including patient preparation and image reconstruction [45]. Finally, there may be potential for WB-MRI in the secondary screening of congenital skeletal diseases predisposing to malignancies, like multiple cartilaginous exostoses or histiocytosis X (Fig. 5). However, only case reports have been published on the potential benefits or therapeutic impact in these specific patient groups. Fig. 5A 35-year-old man with multiple cartilaginous exostoses. a, b Whole-body MRI shows typical manifestations in the metaphyseal parts of the long bones of the upper and lower extremities. c Enlargement of the left knee joint shows an exostosis at the medial side of the femur and at the proximal tibia. d Deformation of the radius. e Exostoses and deformation of both femoral necks. In summary, no indications of malignant transformation (e.g., widening of the cartilaginous cap) were found in this patient Conclusion Whole-body imaging is increasingly successfully applied in musculoskeletal imaging, especially in the field of systemic malignant diseases affecting the bone and in diseases predisposing to malignant transformation. Whole-body MRI is able to depict bone marrow pathologies with high resolution and excellent soft tissue contrast by demonstrating signal alterations due to changes in its fat, water, and hematopoietic cell components with high resolution and excellent soft tissue contrast. Together with CT or PET-CT and its valuable additional metabolic information, it has great potential in the more comprehensive, more accurate, and earlier diagnosis of musculoskeletal diseases. Although further evaluation of the true potential of whole-body applications is awaited, they are promising tools aiding the more efficient management of patients suffering from systemic malignant or benign diseases of the soft tissue and bone.

Cell_Tissue_Res-3-1-2039796

Effect of thrombin peptide 508 (TP508) on bone healing during distraction osteogenesis in rabbit tibia

Thrombin-related peptide 508 (TP508) accelerates bone regeneration during distraction osteogenesis (DO). We have examined the effect of TP508 on bone regeneration during DO by immunolocalization of Runx2 protein, a marker of osteoblast differentiation, and of osteopontin (OPN) and bone sialoprotein (BSP), two late markers of the osteoblast lineage. Distraction was performed in tibiae of rabbits over a period of 6 days. TP508 (30 or 300 μg) or vehicle was injected into the distraction gap at the beginning and end of the distraction period. Two weeks after active distraction, tissue samples were harvested and processed for immunohistochemical analysis. We also tested the in vitro effect of TP508 on Runx2 mRNA expression in osteoblast-like (MC3T3-E1) cells by polymerase chain reaction analysis. Runx2 and OPN protein were observed in preosteoblasts, osteoblasts, osteocytes of newly formed bone, blood vessel cells and many fibroblast-like cells of the soft connective tissue. Immunostaining for BSP was more restricted to osteoblasts and osteocytes. Significantly more Runx2- and OPN-expressing cells were seen in the group treated with 300 μg TP508 than in the control group injected with saline or with 30 μg TP508. However, TP508 failed to increase Runx2 mRNA levels significantly in MC3T3-E1 cells after 2–3 days of exposure. Our data suggest that TP508 enhances bone regeneration during DO by increasing the proportion of cells of the osteoblastic lineage. Clinically, TP508 may shorten the healing time during DO; this might be of benefit when bone regeneration is slow. Introduction Distraction osteogenesis (DO) is a well-established technique, originally developed in orthopedic surgery for bone lengthening (Kojimoto et al. 1988) and later also used to treat hereditary malformations in the craniomaxillofacial region (Amir et al. 2006). DO is characterized by the formation of new bone between two osteotomized bone segments, which are separated by gradual traction. One of the clinical phases of the distraction technique is the bone consolidation period, defined as the time between the end of active distraction and the removal of the distraction device. It represents the time needed for a complete bridging of the distraction gap by bone and a further maturation of this bone. Typically, the bone consolidation phase takes approximately 6–12 weeks in the craniomaxillofacial region and 3–6 months in long bones (Fischgrund et al. 1994; Smith et al. 1999; Felemovicius et al. 2000; Amir et al. 2006). Although DO is believed to be superior to other bone augmentation techniques, many attempts have been made to improve the technique to enhance the bone regeneration process in the distraction gap (Hagiwara and Bell 2000; Li et al. 2002; Mofid et al. 2002; Takamine et al. 2002; Schortinghuis et al. 2005). Such improvements would provide the opportunity to shorten the bone consolidation period and hence minimize complications such as the development of non-union, infection or fracture. Various methods have been tested to promote bone formation in the distraction gap, e.g. electrical and mechanical stimulation (Hagiwara and Bell 2000; Mofid et al. 2002), transplantation of osteoblast-like cells (Takamine et al. 2002) or administration of growth factors such as bone morphogenetic proteins (Li et al. 2002) or fibroblast growth factor 2 (FGF-2; Okazaki et al. 1999). Another procedure to accelerate bone regeneration involves the application of thrombin-related peptide 508 (TP508; Ryaby et al. 2000; Sheller et al. 2004; Li et al. 2005a,b; Wang et al. 2005). TP508 is a synthetic peptide consisting of 23 amino acids, which represent the natural amino acid sequence of the receptor-binding domain of human thrombin (pro-thrombin amino acids 508–530; Sower et al. 1999; Stiernberg et al. 2000). Thrombin is an important factor in blood homeostasis, inflammation and wound healing (Coughlin et al. 1992). It plays a role in the formation of fibrin clots and platelet activation. Thrombin also stimulates the cell proliferation, migration and/or differentiation of various cell types, such as fibroblasts, endothelial cells and lymphocytes. All these cell types contain one or more thrombin receptors (Chen and Buchanan 1975; Belloni et al. 1992). TP508 appears to mimic and accelerate many of the effects of native thrombin in initiating wound healing through a non-proteolytic pathway (Sower et al. 1999). Histological studies have indicated that TP508 is chemotactic for neutrophils, lymphocytes and monocytic cells; it enhances the formation of granulation tissue, neovascularization and other wound-healing events (Sower et al. 1999; Ryaby et al. 2000; Naldini et al. 2004). TP508 has also been tested for its potential to enhance bone formation (Ryaby et al. 2000; Sheller et al. 2004; Li et al. 2005a,b; Wang et al. 2005). A single injection of 1 μg TP508 accelerates fracture healing in rat long bone fracture. The breaking strength of bones of young and aged rats injected with TP508 has been enhanced over that of controls (Ryaby et al. 2000). One recent study of TP508 treatment during DO has revealed a significantly greater bone mineral density in TP508-treated groups than in saline-treated controls (Li et al. 2005a,b). Histological analysis has also shown advanced bone consolidation and remodeling after TP508 treatment. In the present study, we report the effect of TP508 in the regulation of bone regeneration during DO in more detail. We have examined Runx2/Cbfa1, osteopontin (OPN) and bone sialoprotein (BSP) protein expression in tissue sections of rabbit long bones undergoing DO. Runx2 is an essential transcription factor for osteoblast differentiation and for extracellular matrix gene expression (Komori et al. 1997; Karsenty et al. 1999). In Runx2 knockout mice, both intramembranous and endochondral ossification are blocked because of the maturational arrest of osteoblasts (Komori et al. 1997). Runx2 can directly stimulate the transcription of the collagen type I and osteoblast-related genes, such as those for OPN and BSP, both major non-collagenous proteins found in the extracellular matrix of bone (Karsenty et al. 1999). We have hypothesized that TP508 treatment during DO enhances the expression of Runx2, OPN and BSP. Materials and methods Animal model of DO The paraffin blocks containing distracted rabbit long bones were those prepared previously (Li et al. 2005a,b) from 30 adult male New Zealand White rabbits (age 24 weeks, body weight 2.6–3.5 kg). External fixators were applied to stabilize the osteotomized mid-tibia for 7 days of the latency period and were activated for 6 days, once a day at a rate of 1.4 mm. This distraction rate was intentionally chosen in order to copy the poor bone healing condition as previously described (Li et al. 1997). During the experiment, seven animals were excluded in the study because of soft tissue complications (n = 2), pinhole fracture (n = 3) and anaesthetic death (n = 2). The remaining 23 rabbits were randomly divided into three experimental groups, each group consisting of minimally seven rabbits. TP508 (OrthoLogic, Tempe, USA) at two different concentrations (30 and 300 μg TP508) or saline were injected percutanously at the beginning and at the end of the lengthening phase. The first group (n = 8) received injections of 300 μl saline containing 300 μg TP508, the second group (n = 8) received injections of 300 μl saline containing 30 μg TP508 and the third group (n = 7) received injections of 300 μl saline alone and served as a control group. The injections were given in three different areas (100 μl in each area): the proximal, central and distal area of the distraction gap. Two weeks after the lengthening phase, all animals were killed. Tissue samples consisting of the distraction gap and 5 mm proximal and distal to the pre-existing bone were then harvested and immediately fixed in 95% ethanol for further examination. Tissues were decalcified with EDTA and embedded in paraffin. Serial sections (5 μm thick) were cut and mounted on poly-lysine-coated glass slides. The peripheral quantitative computed tomography and histological examination of these bones has been published previously (Li et al. 2005a,b). All animal experimental procedures were approved and performed following the guidelines for animal scientific procedures (Animal Scientific Procedures Act 1986, British Home Office). Antibodies The mouse recombinant monoclonal antibody Pebp2alphaA was kindly donated by Dr. K. Sasaguri, Kanawaga Dental School, Japan. This antibody has been used and characterized previously and reacts with all isotypes of Runx2 (Bronckers et al. 2001, 2005). The mouse anti-rat OPN monoclonal antibody (MPIIIB101) was obtained from the Developmental Studies Hybridoma Bank (developed under the auspices of the NICHD and maintained by the University of Iowa, Department of Biological Sciences Iowa City, IA 52242, USA). The mouse anti-human BSP (monoclonal antibody A4232.1) was purchased from Immunodiagnostik, Bensheim, Germany. Both anti-OPN and anti-BSP antibodies have been used and specified previously (Ibrahim et al. 2000; Bronckers et al. 2005; Zerbo et al. 2005). Immunohistochemical procedures Tissue sections were de-paraffinized in xylene and rehydrated through a graded series of aqueous ethanol solutions. Endogenous peroxidase was quenched by incubation in 3% H2O2 in 100% ethanol (5 min). A pre-incubation in hot citrate buffer (0.01 M, pH 6.0, 95°C, 20 min) was performed for antigen retrieval prior to incubation with 30% normal horse serum (30 min). Sections were then incubated overnight with primary monoclonal anti-Runx2 antibody (1:200), anti-BSP antibody (1:400) or anti-OPN antibody (1:600) at 4°C in a humidified chamber. After washes with phosphate-buffered saline (PBS), biotinylated horse anti-mouse IgG (ABC-peroxidase Elite kit; Vector Laboratories, Burlingame, Calif., USA) was added at a dilution of 1:200 and the sections were incubated at room temperature for a minimum of 60 min. The sections were washed again with PBS, incubated with ABC peroxidase complex for another 60 min followed by a maximum of 20 min in 3,3′ diaminobenzidine substrate solution (Vectastain kit) to visualize the immunoreaction, counterstained with methyl green for 2–3 min, mounted in 50% glycerol and covered with a glass coverslip. As negative controls, sections were incubated with non-immune mouse IgG in place of the primary antibody. Scoring Sections were examined by using a Leica DM RA microscope equipped with a Leica DC 200 digital camera. The central areas of the gap were selected as the region of interest (ROI). First, the boundaries of the distraction gap (indicated by the osteotomy lines) were determined and subsequently a grid was positioned over the ROI halfway between the two osteotomy lines. The grid was 609 μm in width and 479 μm in length. The ROI consisted of some newly formed bone (mostly in the group treated with 300 μg TP508) and fibrous tissue (Fig. 1a). Counting was performed by using the grid at a magnification of 200×, with a minimum of three sections, at least 30 μm apart, per animal. Some tissue blocks were of poor quality or no longer contained a complete intact gap area with an ROI; this reduced the number of animals per group to 5–6 animals. The total number of cells was counted (immunopositive and immunonegative cells). The sections were coded and randomized. Quantification was performed by one investigator (L.A.) who was unaware of the origin of the sections. Numbers of immunopositive cells were noted in absolute terms and were expressed as the percentage of all (immunopositive and negative) cells present in the ROI. Fig. 1Immunolocalization of Runx2 and osteopontin proteins in TP508-treated groups (a-c, e 300 μg TP508 group, d 30 μg TP508 group). a Overview of immunoreactions in a biopsy taken from rabbit long bone after distraction osteogenesis Runx2 staining (FT fibrous tissue, NB new bone, P periosteum, PC periosteal callus). A white grid was used to count the number of immunopositive and negative cells in the region of interest. ×50. b Runx2 staining in the nuclei and cytoplasm of the cells. Runx2 was strongly expressed in osteoblasts (Ob). Note that young osteocytes (Ocy), which had recently embedded in the bone matrix, expressed Runx2 protein. Cells adjacent to the osteoblast layer, presumably pre-osteoblasts (Pre-Ob), also expressed Runx2, but with lower intensity (B bone). ×400. c Immunostaining for osteopontin. Positive staining was found in the cytoplasm of the cells (white arrows). Note that expression was stronger towards the newly formed bone (NB). ×200. d Runx2 immunostaining in tissue of a rabbit injected with 30 μg TP508. Expression was much lower than that after injection with 300 μg TP508 and was similar to saline-injected control tissue (not shown). Fewer fibrocartilage-like cells (FCC) were stained than after injection with 300 μg TP508 (HC hypertrophic cartilage-like cells). ×200. e Negative control stained with non-immune IgG instead of primary antibodies to Runx2. Note the lack of positive staining. Counter-staining with methyl green. ×200. Bars 20 μm (b, d), 50 μm (c, e), 300 μm (a) Cell cultures The mouse osteoblast-like MC3T3-E1 cell line, derived from newborn mouse calvaria, was routinely maintained in α-MEM (Gibco, Paisley, UK) supplemented with 10% fetal bovine serum (HyClone), 120 μg/ml penicillin (Sigma, St Louis, Mo., USA), 100 μg/ml streptomycin sulphate (Sigma), 1.25 μg/ml fungizone (Gibco), 50 μg/ml sodium ascorbate (Merck, Darmstadt, Germany), 10 mM β-glycerophosphate (Sigma) and 300 μg/ml glutamine (Sigma) at 37°C in a humidified atmosphere of 5% CO2 in air. This cell line is capable of expressing Runx2 mRNA (Tsuji et al. 1998). To stimulate cells to express osteoblastic markers, they were first exposed to 10 nM dexamethasone in 25 cm2 flasks (Greiner BioOne, Solingen, Germany) for 1 week. Upon confluency (typically after 5 days), cells were harvested by using 0.25% trypsin and 0.1% EDTA in PBS, centrifuged at 600g for 10 min, washed and plated in 24-well culture dishes (Greiner BioOne) at three different cell densities (3×104, 1.5×104 and 7.5×103 cells/well) in order to obtain a comparable number of cells at the end of each experiment. After 1 day of culture, the medium was replaced by media containing various concentrations of TP508 (0, 10 or 100 μg/ml). Throughout the experiment, all media were supplemented with 10 nM dexamethasone. Cells were exposed to TP508 for 24, 48 and 72 h, after which times, the cells were collected. The experiment was carried out in quadruplicate and repeated once. RNA analysis and quantitative real-time polymerase chain reaction Total RNA from cultured cells was isolated by using TRIZOL reagent (Gibco) according to the manufacturer’s instructions. The RNA content was determined by measuring the absorbance in water at 260 nm by means of an Ultrospec III spectrophotometer (Amersham, Buckinghamshire, England). cDNA synthesis was performed by using 750 ng total RNA in a final reaction volume of 20 μl containing 5 U transcriptor (Roche Applied Science), 5× polymerase chain reaction (PCR) buffer, 4 U random primers (Roche), 20 U protector RNase inhibitor (Roche), 1 mmol each dNTP and 10 μl template. The reverse transcription step was performed on a Gene Amp 9700 Thermocycler (Applied Biosystems, Foster City, Calif., USA) at 55°C for 30 min followed by 85°C for 5 min. Real-time PCR was performed on the ABI PRISM 7700 sequence detection system (Applied Biosystems). The phosphobilinogen deaminase gene (PBGD) served as the endogenous reference (de Vries et al. 1999) to normalize Runx2 expression. The primers for the amplification of Runx2 mRNA were 5′-ATGCTTCATTCGCCTCAC-3′ and 5′-ACTGCTTGCAGCCTTAAAT-3′ (GenBank database, accession no. NM 001024630). The PCR primers for PBGD, amplified as the internal reference, were 5′-AGTGATGAAAGATGGGCAACT-3′ and 5′-TCTGGACCATCTTCTTGCTGA-3′ (accession no. BC 003861). For the amplification of the Runx2 and PBGD products, 37.5 ng cDNA was added to the PCR mixture containing SYBR Green PCR Master Mix consisting of SYBR Green I Dye, AmpliTaq Gold DNA polymerase, dNTPs with dUTP, a passive reference and buffer (Applied Biosystems) and 300 nM of each primer, in a final volume of 25 μl. The enzyme was activated by preheating the samples for 10 min at 95°C, followed by a two-step PCR procedure consisting of a denaturation step at 95°C for 15 s and an annealing and extension step at 60°C for 1 min for 45 cycles. Relative expression was calculated by using the comparative Ct method. Samples were normalized for the expression of PBGD by calculating ΔCt (CtRunx2−CtPBGD); subsequently, the ΔΔCt values were calculated as ΔCtsample−ΔCtcalibrator , where the calibrator was the control sample (without TP508 incubation). Relative expression of the Runx2 gene was expressed as 2-(ΔΔ Ct) (Livak and Schmittgen 2001). Statistical analysis Values obtained from the scoring of tissue sections and normalized mRNA levels determined by PCR analysis (presented as means and standard deviations) were analysed by using GraphPad Prism 4 for Windows. For the analysis of the immunostaining, an average value was calculated per animal for each protein and treatment; these average values were used to determine the mean and standard deviation for each treatment group (5–6 animals/group). Analysis of variance (ANOVA) was used to test for statistical significance, which was accepted when P<0.05 (two-tailed). Tukey’s post test was performed to compare individual pairs of groups if P<0.05. Linear regression analysis was used to examine the correlation between Runx2 mRNA expression and culture time. Results Immunohistochemistry demonstrated different localization patterns for Runx2, OPN and BSP proteins. Cells expressing these proteins were detected throughout the distraction gap and in the periosteum, the periosteal callus and the old (predistraction) bone (Fig. 1a). Runx2 staining was located in nuclei and cytoplasm, whereas strong staining for OPN and BSP was localized in the cytoplasm (Figs. 1b,c,2i,3c); weak staining for BSP and OPN was also seen in the extracellular bone matrix. Figures 1, 2, 3 present the different staining patterns and Table 1 summarizes the average intensity of immunostaining in the ROI. Fig. 2Runx2 (a, d, g), OPN (b, e, h) and BSP (c, f, i) expression in the distraction gap (B bone, FT fibrous tissue, FC fibrous cartilage-like tissue, asterisks central area of distraction gap, hatch layers of cells adjacent to the newly formed bone, arrows osteoblasts, arrowheads osteocytes) of the various groups (a-c 0 μg TP508 group, d-f 30 μg TP508 group, g-i 300 μg TP508 group). Inset (i): Newly formed bone in the gap. Original magnifications: a 50×; d, g 100×; b, e, h 100×; c, f, i, inset 200×. Bars 20 μm (c, f, i), 50 μm (b, d, e, g, h), 100 μm (a)Fig. 3Immunolocalization of Runx2, OPN and BSP in the 300 μg TP508 group. a Runx2 staining of a vessel wall (EC endothelial cells, SMC smooth muscle cells). Note that some cells near the blood vessel weakly express Runx2 (white arrows). ×400. b Osteopontin staining of a vessel wall. Note that the cells surrounding the blood vessel are immunonegative and only stain weakly for methyl green (white arrows). ×400. c BSP staining is restricted to the layer of osteoblasts (Ob) and hardly occurs in fibrous tissue × 400. d OPN staining is expressed in osteoblasts (Ob) and osteocytes of new bone (NB) and some fibrous tissue cells close to osteoblasts, presumably pre-osteoblasts (Pre-Ob). ×200. Bars 20 μm (b), 50 μm (c, d), 100 μm (a)Table 1Intensity scores of immunostaining in region of interest (± weak, + moderate, ++ strong, +++ very strong). The intensity of immunostaining was scored in the central area of the gap (region of interest) at 40× final magnification. Variation exists between sections in each animal and between animals within groupsMarkerGroup with 0 μg TP508Group with 30 μg TP508Group with 300 μg TP508RUNX2+ to +++ to ++++ to +++OPN± to ++± to ++++BSP± to +++±+ to +++ Expression of Runx2, OPN and BSP in the group injected with saline (control group) was similar to that in the group treated with 30 μg TP508 (Figs. 1d, 2a–f). The intensity of the staining in both groups was however much lower than that in the 300 μg TP508 group (Figs. 1c,2g,h,3c,d). Runx2 and OPN were detected in active osteoblasts, in fibroblasts next to the osteoblastic layer (presumably pre-osteoblasts), in osteocytes in the old (predistraction) and newly formed bone, in fibroblasts of the periosteum, in the central area of the gap, in endothelial and smooth muscle cells of the vessels and in chondrocyte-like cells next to new bone. We found two types of cartilage-like tissue in the group treated with 30 μg TP508 and in control group, i.e. fibrocartilage-like (collagen fibres between lacunae) and hypertrophic-like (evenly dispersed chondrocytes, without clear fibres; Fig. 1d). Higher numbers of cells associated with these two tissues were found in the group treated with 30 μg TP508 and in the control group than in the group treated with 300 μg TP508 (Fig. 2a,d,g). More fibrocartilage-like cells than hypertrophic cartilage-like cells were positive for Runx2 (Fig. 1d). Sections in which primary antibodies had been replaced with non-immune antibodies were negative (Fig. 1e). The Runx2- and OPN-positive blood vessel cells (both the endothelial and smooth muscle cells) were more evident in the group with 300 μg TP508 (Fig. 3a,b). They appeared to occur at locations with high bone formation activity. Connective tissue cells in close vicinity to Runx2-positive blood vessels also stained faintly for Runx2 but not for OPN. Immunostaining for BSP was more restricted than that for Runx2 and OPN and was seen in osteoblasts and young osteocytes (Fig. 3c,d). Weak expression was found in some of the cells considered to be pre-osteoblasts. Fibroblast-like cells in the central fibrous area of the distraction gap expressed Runx2 (Fig. 2a,d,g) and OPN (Fig. 2b,e,h) but were negative for BSP. The number and the intensity of Runx2 and OPN in the gap increased from fibrous tissue towards the newly formed bone (Fig. 1c). Weak BSP staining was seen in the area within the gap in which bone trabecules started to be formed (Fig. 2c,f,i). In the group injected with 300 μg TP508, immunostaining for Runx2, OPN and BSP was distributed throughout the regenerating areas. The immunostaining for Runx2, OPN and BSP in the connective tissue in the gap (endosteal area) was relatively high and as intense as in the periosteal callus tissue. In the 30 μg TP508 group and in the control group, staining was less intense in the central gap area than in the periosteal callus area. Quantitative immunohistochemistry was performed to investigate the effect of TP508 injection on the total number of cells and on the number of differentiated bone cells in the central area of the gap. The total number of cells (immunostained and unstained) in the ROI was not significantly different between the three groups. Mean values and the standard deviation of the total number of cells in the tissues treated with 0, 30 μg TP508 and 300 μg TP508 were 958±217, 1140±137 and 1161±282, respectively (ANOVA, P=0.5). This indicated that TP508 injection had no overall effect on total number of cells in the distraction gap. Next, we counted the number of cells, in the ROI, expressing each of the three antibodies tested. Absolute values of immunopositive cells in the 0, 30 μg TP508 and 300 μg TP508 groups were as follows: for Runx2, 263±61, 288±171, 560±115, respectively; for OPN, 299±201, 244±40, 677±216, respectively; for BSP, 255±111, 252±73, 480±31, respectively. We also expressed the number of immunopositive cells as a percentage of the total number of cells (Fig. 4). The number of Runx2- and OPN-expressing cells was significantly higher in the group treated with 300 μg TP508 than in the control group that received only saline (P<0.05; Fig. 4). The number of BSP-expressing cells was also higher in the 300 μg TP508 group than in the 30 μg TP508 group (P<0.05). No significant differences were found between the 30 μg TP508 group and the control group receiving saline (P>0.05). Fig. 4Effect of TP508 injection on the number of immunopositive cells in the central portion of the distraction gap (black bars Runx2-positive cells, grey bars OPN-positive cells, white bars BSP-positive cells). Data are presented as percentage cells immunopositive for each of the three proteins examined (means±SD). *The percentages of Runx2-, OPN- and BSP-positive cells in the 300 μg TP508 group were significantly higher than for cells with the corresponding proteins in the saline-treated control group; the percentage of BSP-positive cells in the 300 μg group was also significantly higher than that in the 30 μg group (ANOVA, P<0.05) The previous data suggested that TP508 increased the differentiation of osteogenic cells into osteoblasts in situ. To determine whether TP508 had any direct effect on osteogenic cells that were in the process of turning into osteoblasts, we performed in vitro experiments with the (mouse) osteoblast-cell line MC3T3-E1 after pre-culture in the presence of dexamethasone. After a 24-h stimulation with TP508, we found a significant (but slight) downregulation of Runx2 mRNA at 10 μg/ml TP508 (P=0.008) but no change at 100 μg/ml TP508 (Fig. 5). Fig. 5Quantitative reverse transcription/polymerase chain reaction analysis of Runx2 mRNA expression in mouse osteoblast-like (MC3T3-E1) cells cultured for 24, 48 and 72 h in the presence of TP508 (black bars control group with 0 μg/ml TP508, hatched bars 10 μg/ml TP508, white bars 100 μg/ml TP508). Data are presented as means±SD (n=8) and are representative of two experiments. *The 10 μg/ml TP508 group at 24 h of incubation was significantly lower than the control group (one-way ANOVA; Tukey’s multiple comparison test, P<0.01). Correlation analysis between mRNA and culture time: r2=0.94, P=0.06 for 10 μg/ml TP508; r2=0.99, P=0.06 for 100 μg/ml TP508 Discussion This study demonstrates a positive effect of TP508 on the synthetic activity of bone-forming cells in the distraction gap in rabbit long bones. After injection of a high dose of TP508, significantly more cells expressing Runx2, OPN and BSP are located in the central part of the distraction gap, predominantly consisting of soft connective tissue. We have also found that these cells stain more intensely for Runx2, OPN and BSP suggesting that TP508 not only proportionally increases the number of osteogenic cells in the distraction gap but also their expression level. The finding that TP508 enhances bone-forming activity in the soft tissue gap area is in agreement with previous morphometric data indicating that TP508 stimulates bone regeneration in the distraction gap (Li et al. 2005a,b). The underlying molecular mechanism of the stimulation of Runx2 expression in the distraction gap after TP508 treatment is not clear. Our in vitro experiment with osteoblast-like MC3T3-cells suggests that TP508 does not have a significant positive effect on Runx2 mRNA. However, we cannot rule out that the MCT3C3 cells are too far advanced along the osteoblast differentiation pathway to respond to TP508 for they had been pre-cultured in dexamethasone-containing medium before they were exposed to TP508. No definitive conclusions can therefore be drawn from these in vitro data. In vivo, TP508 may have stimulated the differentiation of osteogenic cells though the activity of other non-osteogenic cell types. Injection of TP508 into a bone fracture model upregulates FGF-2 mRNA in (not further specified) tissues at the fracture site (Wang et al. 2005). Since FGF-2 is known to be an upstream regulator of Runx2 (Franceschi and Xiao 2003), one explanation for the increased Runx2 expression in the gap is via an enhanced cytokine release by inflammatory cells, which in turn stimulates osteoblast differentiation (Okazaki et al. 1999; Stiernberg et al. 2000; Franceschi and Xiao 2003; Naldini et al. 2004). Alternatively, TP508 stimulates blood vessel growth in vivo (Norfleet et al. 2000; Li et al. 2005a,b; Vartanian et al. 2005; Wang et al. 2005) and this subsequently enhances bone formation. TP508 increases recruitment of human aortic and microvascular endothelial cells through chemotaxis (Norfleet et al. 2000; Li et al. 2002), upregulates angiogenesis-related genes (Wang et al. 2005) and acts directly on microvascular cells to accelerate sprouting (Vartanian et al. 2005). Vascularization provides the oxygen that is required for the metabolic activity of the osteoblasts (Tuncay et al. 1994; Salim et al. 2004); it provides homes for circulating stem cells that, under proper stimulation, differentiate into osteoblastic cells (Kuznetsov et al. 2001). In addition, vessel cells secrete paracrine factors that can regulate bone cell metabolism (Villars et al. 2000; Street et al. 2002). Thus, the angiogenic properties of TP508 might help to enhance bone growth. In the present study, we have also found Runx2 protein expression in endothelial cells, vascular smooth muscle cells and loose connective cells surrounding developing blood vessels at sites of active bone and tissue formation. Recent studies have presented evidence for transient Runx2 mRNA and protein expression during angiogenesis in embryonic human orofacial tissues (Bronckers et al. 2005), in human primary bone marrow endothelial cells and in endothelial cell lines (Namba et al. 2000; Sun et al. 2001). Thus, our data suggest that, with respect to Runx2 expression, the formation of blood vessels in regenerating adult tissues during DO resembles vessel formation in embryonic tissues. Of note, under certain conditions such as inflammation, cells associated with vessels (e.g. pericytes, endothelial cells and smooth muscle cells) or their precursors are capable of differentiating into osteoblast-like cells (Schor et al. 1995; Reilly et al. 1998; Kinner and Spector 2002) suggesting that the forming microvasculature is a potential source for osteogenic cells. The distribution and staining intensity of Runx2, OPN and BSP varies between the different groups. In the control group and in the group treated with a low dose of TP508, staining is more pronounced in the periosteal callus than in the tissue in the distraction gap. However, in the group treated with a high dose of TP508, we have found the same high staining in both periosteal callus and gap tissue. Periosteal callus formation is a common finding during fracture healing (Gerstenfeld et al. 2003). New woven bone is formed rapidly around the fractured segments, because of stimulation of the periosteum that contains many osteoprogenitor cells. In the same section, a high staining intensity in the periosteal callus can therefore be considered as an internal reference. Hence, as the staining for Runx2, OPN and BSP in the distraction gap is as intense as that in the periosteal callus at the high dose of TP508, this is a clear indication that the high dose of TP508 stimulates bone formation in the distraction gap. This observation is in accordance to a previous histological study that has demonstrated advanced bone consolidation in a group treated with a high dose of TP508 (Li et al. 2005b). TP508 might stimulate the migration of osteoblast precursors to the distraction gap (Li et al. 2005a). Li and colleagues have shown that human periosteal-derived osteoblasts and human microvascular endothelial cells are chemotactic to TP508 (Li et al. 2005a). Therefore, the presence of TP508 in the gap might attract osteoprogenitor cells in the periosteum to migrate into the distraction gap. Runx2 is a critical gene for osteoblast differentiation and for the function of the fully differentiated osteoblast (Komori et al. 1997; Karsenty et al. 1999). The Runx2 protein binds to the osteoblast-specific element 2 that is present in the promoter regions of all major extracellular matrix genes expressed in osteoblasts, including the genes for osteocalcin (OC), OPN and BSP and it regulates the expression of these proteins (Karsenty et al. 1999). OPN and BSP belong to the Small Integrin-Binding LIgand, N-linked Glycoprotein (or SIBLING) family (Fisher et al. 2001). Among other functions, members of this family are known to have a role in mineralization, as they bind strongly to hydroxyapatite (Fisher et al. 2001). In conclusion, we have demonstrated that the increase of bone regeneration by thrombin-related peptide TP508 is associated with an increase in the immunostaining for Runx2, an essential transcription factor of the osteoblastic lineage, and for the bone matrix proteins BSP and OPN. TP508 may thus be a candidate for enhancing bone regeneration when bone regeneration is slow, as occurs in elderly patients.

J_Med_Internet_Res-7-2-1550646

A Case Study of a Retracted Systematic Review on Interactive Health Communication Applications: Impact on Media, Scientists, and Patients

Background In October 2004, a flawed systematic review entitled “Interactive Health Communication Applications for People with Chronic Disease” was published in the Cochrane Library, accompanied by several press releases in which authors warned the public of the negative health consequences of interactive health communication applications, including the Internet. Within days of the review's publication, scientists identified major coding errors and other methodological problems that invalidated the principal conclusions of the study and led to a retraction. While the original study results and their negative conclusions were widely publicized in the media, the retraction seemed to go unnoticed. Introduction Publication of the Review On October 18, 2004, the Cochrane Collaboration, a organization which produces and disseminates systematic reviews of health care interventions [1], published a review entitled “Interactive Health Communication Applications for People with Chronic Disease” [2], which from this point on will be referred to as the “IHCA review.” The IHCA review was edited by the Cochrane Consumers and Communication Review Group [3]. Those who prepare reviews volunteer to work in one of many Collaborative Review Groups, with editorial teams overseeing the preparation and maintenance of the reviews. Interactive health communication applications (IHCAs) were defined in the IHCA review as “computer-based, usually Web-based, health information packages for patients that combine information with social-, decision-, or ‘behavior change'-support” [2]. The results of the IHCA review showed that IHCAs had a positive effect on knowledge and on social support, no effect on behavioral outcomes, and a negative effect on clinical outcomes. The principal conclusion of the review was “consumers whose primary aim is to achieve optimal clinical outcomes should not use an IHCA” [2]. This conclusion was the focus of a press release which the mass media widely circulated (as will be documented later). However, only days later, the IHCA review was found to be flawed and was retracted. Retractions The National Library of Medicine (NLM) is a leader in the bibliographic handling of retractions. The Medical Subject Headings (MeSH) contain the concept “retracted publication,” which identifies a citation previously published and now retracted through a formal issuance from the author, publisher, or other authorized agent. In January 2005, the PubMed query “Retracted Publication[Publication Type] AND 1971:2004[edat]” retrieved 619 retracted citations that entered PubMed between 1971 and 2004. Since the query “1971:2004[edat]” retrieves approximately 12.5 million citations, less than 1 in 10000 publications have been retracted. Friedman [4] studied 60 fraudulent articles by one scientist. Journals in which the scientist had published were notified of the fraud. Only 7 articles were subsequently tagged in PubMed with “Retracted Publication.” The delay between publication of a paper and its retraction often has deleterious effects [5]. Furthermore, journals and institutions are hesitant to issue a statement of errors in published work unless the author of the work confesses to the error, which authors may resist doing because such an admission can be career-damaging. While very few publications are officially retracted, the concern about factors related to retractions is substantial. The study of retractions itself might be indexed with MeSH concepts such as “scientific misconduct,” although the fraction of retractions that stem from error as opposed to scientific misconduct is not known. The query “Scientific Misconduct[majr] AND 1971:2004[edat]” in PubMed retrieved 1840 citations. This body of literature recommends that medical researchers constructively criticize the research practices of others in their institution to reduce the likelihood of misconduct [6]. The objective of this paper is to document the IHCA review as an event in the history of medical publishing, to identify the factors leading to the publicizing of a retracted publication, and to assess the implications. Methods The objectives of this research called for various study methods. The author employed the following three methods: (1) historical processes of collecting documents about a contemporary event and organizing them thematically; (2) ethnographic processes of author participation in the event, personal communication with other participants in the event, interpretation of communications, and construction of models; (3) content analyses based on bibliographic database and Internet searches, coding of the retrieved documents, and tallying of the code frequencies. The ethnographic method employs the extended case method, and the extended case method applies reflexive science to ethnography. Buroway describes reflexive science as follows: “Reflexive science starts out from dialogue, virtual or real, between observer and participants, embeds such dialogue within a second dialogue between local processes and extralocal forces that in turn can only be comprehended through a third, expanding dialogue of theory with itself” [7]. Various database and Internet searches were employed to study the impact of the review and to quantify the difference between mass media coverage of the original publication and its retraction. LexisNexis Academic databases of health news and general news were searched, as was Google. The queries were designed in an iterative process that began with keywords from the question to be addressed but refined the query based on study of the query retrieval results. The retrieved results were coded, and the coding language was also developed in an iterative process. First, the obvious codes “about the review” and “about the retraction” were introduced. Each retrieved document was classified into a single code by the author. If the retrieved document was not appropriately described by an existing code, then the coding language was augmented. The Web of Science was also queried to identify academic citations, but no citations were identified (data not shown). Most database and Internet searches were conducted in May 2005. To better understand how special the publicity accorded the IHCA review was, this study was extended to three other publications: 2 of these were retracted publications tagged as “Retracted Publication” (1 Cochrane review, but not eHealth related, and 1 non-Cochrane review, but eHealth related), and 1 was a meta-analysis with a scope similar to that of the IHCA review. These 3 reports were identified through PubMed searches. Results The following qualitative results on the impact of the IHCA review are organized into three main sections: scientist reaction, mass media reaction, and patient reaction. The section on scientist reaction considers Cochrane reviewers' reactions and how eHealth scientists responded to the IHCA review in the comment section of the Cochrane database. The mass media section provides the Cochrane retraction and then explores, via LexisNexis and Google results, the reaction of the mass media to the IHCA review. The patient reaction section shares dialogue from patient-patient online discussions that reveals the reactions of patients to the IHCA review. Scientist Reaction The Cochrane Collaboration allows anyone to submit comments to the published reviews. Two scientists' comments on the IHCA review appeared independently on October 28, 2004. Kummervold and Eysenbach criticized the IHCA review for both its protocol and its coding. Kummervold explained in detail how the coding of the meta-analysis was incorrect: “We can't get the numbers to add up, it looks like they are reversed in 8 of the 11 studies...” [8]. He delineated the facts and the interpretation for each of the 8 studies at issue; for example, regarding the HbA1c measurement in the Lehmann 2003 paper, he stated that Lehmann reported a reduction in HbA1c of 0.8 for the intervention group, and 0.1 for the control group, which should be interpreted as a positive result for the intervention group. Kummervold added: “We also find it strange that you focus so much on the overall estimates when there is so much heterogeneity in the material. The conclusion seems to be overstated” [8]. Eysenbach had similar comments, stressing that a formal meta-analysis of these heterogeneous studies was problematic, and that the three studies which contributed most to the “negative” result were in fact positive: “Apart from the fact that I do not think that it is legitimate to do a formal meta-analysis using papers measuring totally heterogeneous outcomes with different types of interventions, I also notice that the overall effect estimate is ‘negative' (eg, ‘favoring control') because of three studies…. However, when I read these three studies I cannot find that their result[s] are negative…. If my suspicion is correct, then this is quite a catastrophic error, and quite an embarrassment for Cochrane to let such an error slip through peer-review” [9]. On November 10, 2004, the Cochrane Consumers and Communication Review Group reacted to the discovered errors [10] with a notice that included the following: “The review will be withdrawn as soon as possible…. As the corrections to the review have not been completed yet, it would be premature to announce any reversal of the review's findings at this stage.… The original press releases regarding this review were made not by the Cochrane Collaboration itself but by University College London….” John Wiley & Sons (the publisher of the Cochrane Database) released to EurekAlert a retraction on December 6, 2004: “The review originally determined that…chronically ill people using interactive programmes had worse clinical outcomes than those who did not. Regrettably, errors in data analysis meant that these outcomes were reported incorrectly.... It is expected that the revised results will be published in April 2005” [11]. The April 2005 edition of the Cochrane Systematic Reviews did not mention the IHCA review. Royle, the chief executive officer of the Cochrane Collaboration, said that further review of the revised report was ongoing and no date could be given as to when the review might be published (personal communication, April 25, 2005). Mass Media Reaction The Cochrane Database of Systematic Reviews is not read by the typical consumer. However, Murray's employer, the University College London (UCL), worked with Murray to widely publicize the result. UCL posted a news bulletin on its website on October 18, 2004 that remained there as of May 25, 2005. The bulletin was titled “Knowledge may be hazardous to web consumers' health” and stated the following: “People who use their computers to find information about their chronic disease often wind up in worse condition than if they had listened to their doctor, according to a UCL review of studies on internet health.… One reason…might be because knowledge-seekers become so steeped in information from the Internet they make treatment choices on their own, contradicting advice from their doctors” [12]. Most significantly, the UCL bulletin was circulated to information intermediaries that are considered the main entrance to the world's mass media, including AlphaGalileo and EurekAlert. A search on LexisNexis Academic with the query “Elizabeth Murray AND health” for the period June 2004 to May 2005 revealed a total of 15 relevant press reports, in the following categories: Medical and Health News: There were 9 publications with titles such as UCL's press release title of “Knowledge may be hazardous to web consumers' health.” The publications appeared in places like Life Science Weekly, Law and Health Weekly, and Health and Medicine Week. General News–Major Papers: There were 5 relevant articles, such as one entitled “Why medical advice from the internet can be bad for your health” in the British The Daily Telegraph and another entitled “Medical Web sites may be unhealthy places to learn about ills” in the Omaha World Herald. Only 1 article was about the retraction, published in the Ottawa Citizen on October 18, 2004. Time Incorporated Publications: There was 1 article in the November 1, 2004 issue of Time entitled “Click to Get Sick?” [13]. Among the 15 results from the LexisNexis Academic database, only 1 newspaper report, authored by Tom Spears, dealt specifically with the retraction [14]. Spears, in personal communication with this author (November 18, 2005), said, “I was fairly stunned today to learn that it [IHCA review] has been withdrawn; I found out only because I was looking up the study for my daughter, a science student. Now I'm covering the sequel for tomorrow's paper.… I scan EurekAlert faithfully, as many reporters do, and never saw a hint of anything there.” To further test whether the media emphasized the false negative result but minimally covered the retraction, a content analysis on Google was performed on May 24, 2005. The query was “health AND Cochrane AND Murray AND (interactive OR web OR internet)” for English pages, within the past year. Of the first 200 retrieved hits, 170 pages were not related to the IHCA review. Of the remaining 30 pages, 23 (77%) were reports about the original publication that did not mention the retraction, and an additional page was a bibliography (at a UCL site) that included a citation to the IHCA review, again without mentioning the retraction. All reports (except the bibliography) used a title such as “Click to Get Sick?” and emphasized the negative impact on clinical outcomes of using the Web. The reports came from such reputable sources as the British Broadcasting Corporation and US News and World Report. In contrast, only 6 pages (20%) addressed the retraction: 2 were the original press releases now marked with “retraction” but still emphasizing in their particulars the negative health impact, 3 were Web pages at Cochrane sites, and 1 was an announcement from MedicalNews entitled “Updated press release to October 2004 Cochrane Review.” The latter was the only non-Cochrane–related page primarily addressing the retraction. The grey literature reported on the mass media. For example, The Neuroscience for Kids Newsletter summarized [15] the “Click to Get Sick?” Time article by Sanjay Gupta, and a Web archive for patient education at the Samaritan Health Center pointed patients to Gupta's article. This author wrote to Gupta and asked him to write about the retraction, but Gupta did not reply. NLM indexed the IHCA review and entered the citation for it (including its abstract) in PubMed on October 21, 2004. The “Retracted Publication” tag did not, however, appear in PubMed until March 24, 2005. Patient Reaction Some patients reported the news about the IHCA review to their patient-patient online discussion groups. In a neurology patient discussion group [16], a patient posted the entire BBC news story. Patients responded in two ways. Some rejected the IHCA review result and added strong comments, such as “I have gotten more help and answers for problems from knowledgeable people on this Internet Forum than I have from any of the multitude of doctors I have seen over the last 12 years.” Others accepted the conclusion but insisted that patients could filter bad information from good and benefit in the end from the web. These patients were not aware of the retraction of the IHCA review. This author reported the Time “Click to Get Sick?” article to two head-and-neck cancer patient discussion groups to which he belongs. A day later he reported the retraction from the Cochrane Database. One member of the discussion group replied: “Thanks for the update–the negative findings seemed odd to me when I read it, so I'm glad it's being revised.” This author, in his role as a cancer patient, also formally commented on the IHCA review at the Cochrane Database site [17]. The typical patient with a chronic disease has no formal medical training and is ill prepared to critique a meta-analysis of clinical trials. However, the typical patient is vulnerable to cultural pressures, as they are partially shaped by and reflected in the mass media. Comparison With Another Cochrane Retraction For comparison, a search for further retracted Cochrane reviews using the PubMed query “Cochrane Database Syst Rev[TA] AND Retracted Publication[PT] AND 1971:2005/5/25[edat]” was conducted. One reference, in addition to the IHCA review already discussed, was identified, which was a retracted review by Brewster et al [18] about antihypertensives. The retraction for the Brewster et al review is explained on the Cochrane website as follows: “This systematic review has been withdrawn temporarily because its contents are potentially misleading.” A search on LexisNexis with the query “Brewster AND antihypertensive” for the period November 2004 to May 2005 retrieved no articles in either the “General News–Major Papers” category or the “Medical and Health News” category. A search on Google for “Brewster antihypertensive” followed by an examination of the first 100 retrieved pages identified 23 relevant pages, which had a very different content pattern than the hits for the IHCA review. They all contained citations of papers from Brewster et al, who have published elsewhere on the same subject as in their review. The Brewster et al publication attracting the most attention was an article [19] in the Annals of Internal Medicine that was not retracted but has the same title as the Cochrane review. Thus, the only other retracted Cochrane review had a very different mass media, scientific, and web impact than the IHCA review. Comparison With Other Retracted Articles Related to eHealth To determine whether other articles on a similar topic to the IHCA review have been retracted, a search was first made for articles on a similar subject that had been MeSH indexed in PubMed. The article by Demiris [20] seemed relevant, and its two MeSH index terms were “Disease Management” and “Internet.” A search on PubMed for “Retracted Publication[PT] AND Disease Management[majr] AND Internet[majr] AND 1995:2005/5/25[edat]” returned no citations. When the search was broadened by removing the term “Disease Management,” 1 retracted reference was retrieved, entitled “The quality of surgical information on the Internet” [21]. As previously described in the Journal of Medical Internet Research, this article was retracted due to a case of cyberplagiarism, with large sections of the paper having been lifted from different websites [22]. A search on LexisNexis Academic with the query “McKinley and surgical and Internet” for the period 1995 to May 2005 revealed no relevant press reports, neither in the “General News–Major Papers” category (three hits were all not relevant to the McKinley article) or in the “Medical and Health News” category. A search on Google for English pages with the query “McKinley surgical Internet” revealed 96 irrelevant pointers in the first 100 results. Of the remaining 4 relevant hits, 1 was the article about the plagiarism [22], which precipitated the retraction of the McKinley et al manuscript, and 3 were academic references to the McKinley et al article, which did not note it being retracted. Thus, the only other retraction of a published article appearing in PubMed similar in topic (the Internet) to the IHCA review had a very different pattern of reactions than the IHCA review. A Similar Meta-Analysis on eHealth The IHCA review addressed a topic that the mass media found interesting. Have any other recent publications also been a meta-analysis on the impact of interactive applications on health, and, if yes, what was the mass media reaction? Using the query “Meta-analysis AND Web AND Chronic Illness” in PubMed, we found only 1 citation: Wantland et al [23] did a meta-analysis on Web-based health interventions that was published (in the Journal of Medical Internet Research) about the same time as the IHCA review. The paper concluded that “the effect size comparisons in the use of Web-based interventions compared to non-Web-based interventions showed an improvement in outcomes for individuals using Web-based interventions to achieve the specified knowledge and/or behavior change for the studied outcome variables.” What has been the impact of the Wantland et al paper and how does that compare to the impact of the IHCA review? The Wantland et al paper was not announced with a press release in EurekAlert. A search on LexisNexis Academic for newspaper articles about the Wantland et al paper retrieves no articles. The queries performed were similar to those performed for the IHCA review and included “Wantland AND health” for 2004 through 2005 in General News/Major Papers. A search was done on Google for “Wantland health Web” on May 24, 2005. Of the first 200 returns, 182 were not relevant. Of the remaining 18 hits, 15 pages contained academic citations to Wantland et al, 2 announced the appearance of the article, and 1 was a personal blog that commented on the article. Thus, most of the Google returns that gave Wantland et al citations are academic in character and very different from the mass media coverage afforded the IHCA review. Discussion As shown, the IHCA review provides a perhaps unprecedented case from which lessons should be drawn. Only one other Cochrane review (about antihypertensives) has been retracted, and that one received negligible mass media attention. The only retracted publication in PubMed that is indexed under the MeSH concept of “Internet” (the IHCA review did not have time to get indexed before it was withdrawn) received no newspaper coverage. The paper most similar to the IHCA review in topic and method (the Wantland et al report [23]) received considerable academic attention but no newspaper coverage. In other words, special circumstances must have come together for the IHCA review situation. This section next presents a framework based on tiers of response. The first tier is medical scientists. The second tier is the mass media spreading medical press releases. The third tier is the patient community reacting to the mass media and the scientists. First-Order Problem In an effort to critique the problem that occurred, one might build on the analysis of misconduct in toxicology by Purchase. Purchase [24] identified four roots of misconduct: Intention of the work Conduct of the studies Design and interpretation of studies Bias from conflict of interest In the case of the IHCA review, the intention was scientifically appropriate, namely to gain further insight about IHCAs through a systematic review. In the other three categories, fault can be found: The errors in the coding of data should not have been made. The coauthors Nazareth and Tai, who are credited with doing the coding, have good enough credentials to not lay the blame on lack of experience: Nazareth is a Professor at UCL and is Scientific Director of the British Medical Research Council's General Practice and Research Framework, and Tai has coauthored several articles over the past two decades that appeared in refereed medical journals. An explanation for the miscoding in terms of experience of the coders is not apparent. The design of the study has been criticized as lumping together studies which are too heterogeneous in their design, interventions, and outcomes [8,9]. The protocol might have been more rigorously vetted by the Cochrane Review Group, and the authors should have been more cautious in their interpretation of results and emphasized the weakness of the design in their publicity. The reporting of the work suggests possible bias. The authors and their employers have sensationalized a result that catches the media's attention. For some observers, the review appeared biased in that the authors, who are affiliated with medical institutions, concluded that patients should listen to their doctor, instead of seeking help on the Internet. Purchase [24] claims that a partial solution to this first-order problem is the institutionalization of quality controls. In the 1970s, good laboratory practice regulations were introduced, but comparable regulations do not exist for meta-analyses. For quality control of a meta-analysis the scientific community relies on the research team, the researchers' institution, and the referees. A medical research institution, such as the UCL Medical School, presumably embraces results from its researchers that can earn mass media coverage and is not the appropriate institution to prevent sensationalizing. Referees can not be expected to detect when laboratory data are intentionally modified [25]; however, in this case they could have been expected to detect when data available to them are miscoded. Problems with refereeing have been frequently noted and in particular for the Cochrane Database [26]. Open commentary, as exists for the Cochrane Database after a publication, is one way to identify flaws. Extending the open commentary to the refereeing phase might reduce the likelihood of something going to press with errors. A submitted article might be available to the public and a community of hundreds of registered scientists could be invited to make anonymous comment. Submissions online would require extensive online commenting that reached a consensus before a submission could be considered “published.” Other approaches to increase the commentary on the research process include refereeing the protocol phase [27], which is done by the Cochrane Collaboration but apparently not with the necessary rigor or topic expertise. Second-Order Problem The second-order problem is a press release and subsequent mass media coverage of the release. Winsten's classic study of science and the media shows how the truth is repeatedly misrepresented by journalists and researchers: “The most striking finding which emerged from the interviews [of medical journalists] is the dominant distorting influence of the competitive force in journalism.… As economic competition among hospitals has intensified, they have begun to compete aggressively for publicity.… With increasing frequency…scientists…are using the media to attach their names to important findings before their competitors do.… The result has been a spiraling competition, sometimes characterized by exaggerated claims” [28]. Online media have stimulated further competition [29]. The case of the IHCA review reflects these pressures. The UCL press release contained inaccuracies, even if the review would have been scientifically sound, in order to gain mass media attention. For instance, the subtitle of the press release was “Knowledge may be hazardous to web consumers' health.” In truth, the IHCA review was not about Web applications, per se, but about IHCAs, which are defined more broadly than “Web applications.” However, writing a news article about IHCAs is less likely to catch attention than an article about the Web. The UCL press release did not introduce and define the term IHCA, and Murray issued statements that implied the Web was the issue. By the time the information from the press release made it into the mass media, the material had been modified enough to lose any mention of IHCAs. For instance, the Time article said, “People who use the Web to get information about their chronic diseases often wind up in worse shape than before they logged on.” One way for researchers to prevent the mass media from misrepresenting the truth is for researchers to understand how the media work and to interact with the media accordingly [28]. Murray should have known that her words might be twisted to emphasize what would sell newspaper space and should not have wildly speculated. The reputations of the Cochrane Collaboration and UCL partially account for the wide dissemination of the original press release. Yet, neither organization has taken adequate steps to undo the impact of the media reporting on the IHCA review. The honesty of the press could be improved with the Internet [30]. Online health care mass media publications could allow the public to make comments on news articles. Rating techniques, such as employed at eBay and Slashdot, might be used to give prominence to quality feedback [31]. Third-Order Problem The third-order problem concerns the long-term impact of the mass media. While electronic publications might be erased from a computer or marked as retracted, this does not consistently happen. Furthermore, some of the mass media coverage of the IHCA review is on paper and sits on people's bedside tables with no practical way to be retracted [32]. Although this author did not (yet) find any citations to the IHCA review in Web of Science, previous studies have confirmed that a retracted scientific publication may continue to have impact without readers recognizing its retracted status. For instance, one study [33] tracking the citation pattern of 82 retracted articles revealed that, together, they were cited 733 times after their retraction, but only a small fraction of the citations referred to the retraction. In the case of the mass media, retracted publications might be read by people without them seeing the separate retraction notice. If and when the revised IHCA review is published, what could it say that would undo the effect of the original publication? If the conclusion is that IHCAs result in improved clinical outcomes, then the medical profession will want to closely study the protocol and might have grounds to discredit the conclusion. The media trumpeted the IHCA review conclusion partly because it was counterintuitive but was backed by top-notch institutions. If the conclusion becomes intuitive, then the media are unlikely to be interested in it. The reactions to the IHCA review in patient online discussions highlight the importance of virtual communities in helping patients deal with published information. Simple extensions to Web-based, patient, discussion systems could help patients connect to Web-based publications. For instance, when a patient posts a message to a Web-based discussion board, the Web system could parse the message and provide links from the message to relevant articles on the Web. Patients might follow the links and engage in discourse about the validity and implications of the literature. This might lessen the potential ill effects of publications that are wrong or misleading. Conclusions This special medical publishing event was marked by incorrect coding and a desire for maximum publicity. The IHCA review authors, their employers, and the Cochrane Collaboration were responsible for quality control, and failed. The mass media played their part by widely publicizing a sensational message but not reacting to the notice that that sensational message was false. The false result that patients are clinically harmed by interactive applications was very strongly delivered to patients worldwide. The broad lesson to be re-learned is that potentially sensational results should be carefully scrutinized before being sensationalized.

Virchows_Arch-4-1-2329729

The significance of lobular neoplasia on needle core biopsy of the breast

The management of a core biopsy diagnosis of lobular neoplasia is controversial. Detailed radiological–pathological review of 47 patients with cores showing classical lobular neoplasia was performed (patients with pleomorphic lobular carcinoma in situ (LCIS) or associated risk lesions were considered separately). Immediate surgical excision in 25 patients showed invasive carcinoma in 7, ductal carcinoma in situ (DCIS) in 1 and pleomorphic LCIS in 1; radiological–pathological review showed that the core biopsy missed a mass in 5, missed calcification in 2 and that calcification appeared adequately sampled in 2. Nineteen patients had follow-up of at least 2 years. Four patients developed malignancy at the site of the core biopsy (invasive carcinoma in three, DCIS in one); one carcinoma was mammographically occult, one patient had dense original mammograms and two had calcifications apparently adequately sampled by the core. In conclusion, most carcinomas identified at the site of core biopsy showing lobular neoplasia were the result of the core missing the radiological lesion, emphasising the importance of multidisciplinary review and investigation of any discordance. Some carcinomas were found after apparently adequate core biopsy, raising the question of whether excision biopsy should be considered after all core biopsy diagnoses of lobular neoplasia. Introduction Lobular carcinoma in situ (LCIS) was described by Foot and Stewart [25]. The term atypical lobular hyperplasia (ALH) was later introduced to describe a similar lesion, but with less marked changes. We prefer the term lobular neoplasia [30], which includes both LCIS and ALH, particularly for core biopsies in which there is limited tissue, making distinction between ALH and LCIS difficult. The relationship between lobular neoplasia and invasive carcinoma of the breast is controversial [28]. Lobular neoplasia is often multifocal and bilateral [4]. There is evidence that it is a risk factor for invasive carcinoma in both breasts: after a diagnosis of lobular neoplasia, there is an increased risk of subsequent invasive carcinoma in both breasts [30], and the majority are ductal in type. Recent evidence suggests that lobular neoplasia may, like ductal carcinoma in situ (DCIS), be a non-obligate precursor for invasive carcinoma at the same site. There is a higher risk of invasive carcinoma in the ipsilateral than in the contralateral breast. The proportion of subsequent carcinomas that are of lobular type is higher than in unselected series of carcinomas [45]. The morphology of the cells in lobular neoplasia and invasive lobular carcinoma is similar. The same truncating mutation in the E-cadherin gene has been found in invasive lobular carcinoma and adjacent LCIS [57]. The significance of lobular neoplasia in breast needle core biopsies is uncertain. Invasive carcinomas are sometimes identified after a core biopsy diagnosis of lobular neoplasia. Some studies suggest this occurs because the core biopsy missed the lesion and so does not explain the clinical or radiological abnormality [41]. Other studies suggest that carcinomas may be identified after the diagnosis of lobular neoplasia on core biopsy even if the core biopsy explains the clinical or radiological abnormality [20]. An important consideration in radiological–pathological correlation is whether classical lobular neoplasia is associated with calcification. The traditional view is that lobular neoplasia does not have a radiological correlate [4], but recent reports have described an association between classical lobular neoplasia and calcification. We also wished to investigate our impression that, in core biopsies containing classical lobular neoplasia, calcification is often associated with columnar cell change. Pleomorphic LCIS is a recently described variant of lobular neoplasia. The growth pattern and absence of E-cadherin expression resemble classical LCIS. However, the marked nuclear pleomorphism, frequent presence of central necrosis and calcification and expression of c-erbB-2 are similar to high-grade DCIS [50]. In the past, pleomorphic LCIS would usually have been diagnosed as high-grade DCIS. The main aim of this study was to assess the risk of invasive carcinoma and DCIS at the site of classical lobular neoplasia diagnosed on breast needle core biopsies. In view of the clear morphological differences, pleomorphic LCIS was considered separately. A second aim was to investigate the localisation of calcification in core biopsies containing lobular neoplasia. Materials and methods This project was discussed with the chair of the Nottingham University Hospitals Research Ethics Committee who considered that it was a service evaluation and, therefore, did not require formal ethical approval. A multidisciplinary review of the pathology and radiology of patients with a core biopsy diagnosis of lobular neoplasia from July 1998 to June 2006 was performed. The biopsies from July 1998 to June 2000 were included in a previous study [35]. The diagnosis of lobular neoplasia was made on haematoxylin and eosin sections. Immunohistochemistry for E-cadherin was performed for lesions with a differential diagnosis including DCIS or pleomorphic LCIS. Lobular neoplasia was subdivided into ALH and LCIS using the criteria described by Page et al. [44]. Pleomorphic LCIS was defined as a high-grade dyscohesive intra-acinar proliferation that was E-cadherin negative. Patients were excluded if they had synchronous or previous invasive carcinoma or DCIS in the same breast. Two patients with contralateral invasive carcinoma were included. Core biopsies containing invasive carcinoma, DCIS or an area suspicious of either diagnosis were excluded. Cores containing pleomorphic LCIS, an atypical intraductal epithelial proliferation (including atypical ductal hyperplasia), radial scar or a papillary lesion were considered separately. Thus, the main group of core biopsies studied contained classical lobular neoplasia with no other risk lesions: we termed this group “simple” classical lobular neoplasia. The core biopsies were reviewed for the following features: location of any calcification and presence of columnar cell change. Six cores were not available, so could not be reviewed. The frequency of columnar cell change was compared with 87 consecutive core biopsies reported by one observer (AHSL) as normal or benign. Results Simple classical lobular neoplasia on core biopsy Forty-nine core biopsies with lobular neoplasia from 47 women satisfied the entry criteria (0.3% of the 14,597 diagnostic core biopsies performed during this period). Two patients had two core biopsies from the same area of the breast containing lobular neoplasia: both are only counted once in the following results. The median age was 52 years (range 33 to 81). Twenty-six patients presented with mammographic screening abnormalities: 20 with calcification, 3 with a mass, 2 with calcification and distortion and 1 with a mass and calcification. Nineteen patients presented symptomatically: ten with a mass, seven with a thickening and two with a thickening and calcification. Two women had calcification identified by mammography performed in one patient as part of the investigation of a contralateral carcinoma and, in the other, as follow-up of a contralateral carcinoma. Eight core biopsies were freehand, 14 were ultrasound guided and 25 were performed under stereotactic guidance. There was an association between lobular neoplasia and columnar cell change (see Table 1). Calcification was seen in 25 of 41 cores with slides available for review. It was seen in lobular neoplasia in six biopsies (15%); in one the calcification was only seen in the lobular neoplasia, but in the other five, it was also present in other changes (columnar cell change in three, sclerosing adenosis in one, fibroadenoma in one). Of the other 19 biopsies with calcification, columnar cell change was the most commonly associated pathology (10 biopsies). Table 1Extent of columnar cell change in cores with lobular neoplasia and controlsColumnar cell changeCore biopsy with lobular neoplasiaControls (normal or benign core biopsy)None12 (29%)58 (67%)One lobule or duct9 (22%)14 (16%)At least two lobules or ducts20 (49%)15 (17%)χ2 = 18, P = 0.0001 Surgical excision within 2 months of core biopsy was performed in 25 patients. Excision showed malignancy in nine patients (see Table 2). In seven patients, pathology–radiology correlation showed that the core biopsy did not explain the radiological abnormality (a mass in five and calcification in two). Five of these seven patients were seen in the first 2 years of this study. Excision showed benign changes in 16, although one woman had invasive carcinoma of mixed ductal and lobular type diagnosed 33 months after the core biopsy in a separate quadrant of the same breast. Table 2Details of patients with invasive carcinoma, DCIS or pleomorphic LCIS at the site of the core biopsy showing simple lobular neoplasiaAgePresentationCore pathologyTime to excision/monthsFinal diagnosis (surgical procedure)Pathology–radiology correlation60Screen-detected massALH, CCCh, calcs113 mm IDC, G3, LN 0/4 (Mx)Core missed mass75Symptomatic massLCIS, UEH, CCCh26 mm IDC, G2, LN not examined (Mx)Core missed mass63Screen-detected distortion and calcification. Mass on USALH114 mm ILC, G2, LN 0/5 (WLE)Core missed mass54Screen-detected calcificationALH14 mm tubulobular carcinoma G1 + 50 mm DCIS, LN 0/6 (Mx)Core missed calcs43Symptomatic massLCIS19 mm IDLC, G2, LN 0/7 (Mx)Core missed mass68Screen-detected massALH112 mm ILC, G2, LN 0/6 (Mx)Core missed mass64Symptomatic mass, no mass on US. Mammographic calcificationALH, cyst, calcs17 mm DCIS (WLE)Calcs sampled by core54Screen-detected calcificationALH112 mm tubular carcinoma + 35 mm DCIS, LN 0/4 (WLE)Core missed calcs45Symptomatic cysts. Mammographic calcificationLCIS, calcs15 mm pleomorphic LCIS (WLE)Calcs sampled by core52Screen-detected calcificationALH, UEH, CCCh, calcs319 mm ILC, G2, LN 1/6 (Mx)Mammographically occult mass70Screen-detected calcificationALH, calcs2936 mm IDC, G3, LN 6/16 (Mx)Dense original mammogram51Screen-detected calcificationALH, CCCh, calcs2630 mm DCIS + extensive LCIS (Mx)Calcs sampled by core55Screen-detected calcificationALH, CCCh, calcs29IDC 8 mm, G3, 45 mm DCIS, LN 1/4 (Mx)Calcs sampled by coreUS ultrasound, ALH atypical lobular hyperplasia, LCIS lobular carcinoma in situ, UEH epithelial hyperplasia of usual type, CCCh columnar cell change, calcs calcification, IDC invasive ductal carcinoma, ILC invasive lobular carcinoma, IDLC invasive ductal and lobular carcinoma, DCIS ductal carcinoma in situ, G histological grade, LN axillary lymph nodes, Mx mastectomy, WLE wide local excision Nineteen patients who had not had a diagnostic surgical excision had follow-up of at least 2 years (range 25 to 105 months). None of these patients received radiotherapy to the breast. Malignancy was identified in seven women. In four, the malignancy was at the same site as the core biopsy (see Table 2), in one DCIS was found in a different quadrant of the same breast and two developed contralateral invasive carcinoma. Three women had no surgical excision and follow-up of less than 2 years. Neither of the two women with a previous history of contralateral invasive carcinoma developed carcinoma at the site of the core biopsy showing lobular neoplasia. No significant relationships were found between the diagnosis of malignancy (DCIS, pleomorphic LCIS or invasive carcinoma) at the site of the core biopsy and the following features: age of the woman, diagnosis of ALH or LCIS on the core biopsy, whether core biopsy or vacuum-assisted biopsy was performed and the method of guidance (freehand, ultrasound or stereotactic). Pleomorphic LCIS on core biopsy Both patients with pleomorphic LCIS on core biopsy presented with calcification detected by mammographic screening. In both, calcification was seen histologically in association with the pleomorphic LCIS. Both had a diagnostic surgical excision, which showed cribriform DCIS in one and further LCIS in the other. Atypical intraductal epithelial proliferation/radial scar/papillary lesion Eight patients had atypical intraductal epithelial proliferation, radial scar or a papillary lesion in addition to lobular neoplasia on the core biopsy. All eight presented with abnormalities detected by mammographic screening: calcification in five, calcification and distortion in one, mass in one and distortion in one. Seven patients had a diagnostic surgical excision, and one patient had her lesion removed with vacuum-assisted mammotomy. DCIS was found in one of three women with core diagnosis of atypical intraductal epithelial atypia and in one of two with core diagnosis of radial scar. Papillary DCIS with a 3-mm focus of invasion was found after a core diagnosis of papillary lesion with atypical intraductal epithelial proliferation. One patient with a core diagnosis of radial scar with atypical intraductal epithelial proliferation and one with a papillary lesion had benign findings at excision. Discussion Relation between lobular neoplasia and calcification An important question in the management of lobular neoplasia on core biopsy is the relationship between lobular neoplasia and calcification. Lobular neoplasia is not typically associated with any specific clinical abnormality, and there are no characteristic macroscopic features. Early papers suggested that classical lobular neoplasia does not have any diagnostic mammographic features [4]. In the past, surgical biopsies containing lobular neoplasia were frequently performed for mammographic calcification, but pathologically, the calcification was mostly associated with fibrocystic change and identified within the lobular neoplasia in about 20% [4, 11, 24, 55]. Calcification is associated with the lobular neoplasia in between 8% and 53% of core biopsies containing classical lobular neoplasia in different studies [1, 3, 14, 20, 41]. Benign calcification is frequently seen close to invasive carcinomas in surgical specimens [52]. It is well recognised that malignancy may be an incidental discovery on a biopsy for calcification that is associated only with benign disease [49]. The present study confirms the association between lobular neoplasia and columnar cell change [8, 41]. In older studies, the majority of surgical biopsies containing lobular neoplasia were performed for fibrocystic change [30]. In the present study, calcification was associated with lobular neoplasia in 15% of core biopsies, but commonly was also seen in fibrocystic change as well. Calcification was associated with columnar cell change in 32% of cores with lobular neoplasia. The association of calcification with columnar cell change is well recognised [27]. Our results support the view that, in biopsies containing lobular neoplasia, most of the calcification is associated with fibrocystic change, particularly columnar cell change. Importantly, the lobular neoplasia is incidental to the calcification in the majority of biopsies. A recent study of collagenous spherulosis found that 15 of 59 patients had associated LCIS [48]. In contrast, only one of the patients in the present study with lobular neoplasia had associated collagenous spherulosis. Relationship between simple classical lobular neoplasia and carcinoma In this study, 44 women with simple classical lobular neoplasia on core biopsy had surgical excision or follow-up of at least 2 years. Twenty-five women had immediate surgical excision, which showed invasive carcinoma, DCIS or pleomorphic LCIS in nine (36%). In the majority, the core biopsy was judged to have missed the clinical/radiological lesion. This emphasises the importance of clinico-pathological review and further investigation of any discordance. Most of the missed lesions occurred in the early part of the study, suggesting that the accuracy of radiological localisation of core biopsies may have improved later in the series. A smaller number of carcinomas occurred after lobular neoplasia on core biopsy apparently incidental to the clinical or radiological lesion. This raises the question of whether excision biopsy should be considered after all core biopsy diagnoses of lobular neoplasia. The frequency of malignancy if the patients with radiological–pathological discordance are excluded is 11% (2/18). Four of 19 patients with at least 2 years follow-up developed carcinoma at the site of the core biopsy. In one, the carcinoma was mammographically occult, although it was clinically palpable, in one the original mammogram was dense and two women had calcification apparently adequately sampled by the core biopsy. Definite comment cannot be made, but the clinical histories suggest that it is possible that the carcinoma may have been identified earlier in some of these patients if they had had an immediate surgical biopsy. If these women are added to those without radiological–pathological discordance in the above paragraph, the frequency of malignancy is 16% (6/37). Previous published studies of simple lobular neoplasia on core biopsy were reviewed [1–3, 6, 9, 10, 12, 14, 16–18, 20, 21, 23, 25, 26, 31, 33, 36, 37, 39, 41–43, 46, 47, 51, 54, 58, 59]. The data in most studies are not representative. Probably the major bias is that only about half of the lesions had a surgical biopsy and it is not clear how patients were selected for excision biopsy. Furthermore, most studies only report image-guided or stereotactic biopsies. A surgical biopsy was performed in 561 patients with a core biopsy diagnosis of lobular neoplasia with invasive carcinoma found in 50 or 49 and DCIS in 32 or 33 (82 [15%] in total). It is ‘impossible to know if some of the reported cases are those in which the targeted lesion has been missed and re-excision was done for failure to successfully sample the target’ [15]. In reviewing the previous studies, we have assumed that sampling was adequate for core biopsy performed for calcification if calcification was present in the core biopsy. Of those 52 malignancies with details available, 25 (48%) had a radiological–pathological discordance (missed mass in 22 and missed calcification in 3), and 27 had apparently adequately sampled calcification. One hundred four patients who did not have immediate surgical excision had reported follow-up. The only four patients (4%) with invasive carcinoma or DCIS on follow-up at the site of core biopsy are the four patients described in the present study. The absence of subsequent malignancy in other studies is very surprising given that lobular neoplasia is a well-established risk factor for later carcinoma. Nevertheless, the low rate suggests that the chance of invasive carcinoma or DCIS is lower in those not having immediate surgical excision. It is difficult to estimate the chance of finding invasive carcinoma or DCIS if all patients had an excision biopsy. The 15% rate found in the subset of patients who had immediate excision biopsy is probably an overestimate. If the cases with radiological–pathological discordance are excluded, the risk is probably about half of this, approximately 8%. If one assumes that half of the patients who did not have an immediate excision did not have invasive carcinoma or DCIS at the core site, this gives an estimate of about 4%. This represents an estimate of the lower limit of risk. Some studies suggest that the risk of finding invasive carcinoma or DCIS is a little higher if the core biopsy shows LCIS rather than ALH [13]. In view of the subjectivity of this distinction on core biopsy, it is probably of limited use in routine practice. It is clearly appropriate to investigate patients with simple classical lobular neoplasia on core biopsy that does not explain the clinical or radiological abnormality. If a definitive diagnosis cannot be made with further core biopsy, then an excision biopsy is indicated. The management of simple classical lobular neoplasia on core biopsy with no radiological–pathological discordance is less straightforward. As discussed above, the literature suggests that the risk of malignancy is in the range of 4% to 8%. This risk is comparable to that associated with radial scar or papillary lesion with no epithelial atypia [7, 34, 35, 40, 53]. Both radial scars and papillary lesions are excised in many centres according to current guidance [19]. The major difference with lobular neoplasia, however, is that both are usually clearly defined lesions on radiological examination and therefore easy to excise. There is a need for prospective studies with surgical excision of all lesions so that an unbiased assessment of the risk of simple classical lobular neoplasia on core biopsy can be made. Careful radiology–pathology correlation is essential. It may be possible to stratify the risk within this group using clinical, radiological or pathological features. Lobular neoplasia associated with radial scars, papillary lesions and atypical intraductal epithelial proliferations Most centres excise radial scars, papillary lesions and atypical intraductal epithelial proliferations diagnosed on core biopsy. In the present study, carcinoma was found on excision of three of the eight lesions with these lesions in combination with lobular neoplasia. This supports the practice of excising such lesions. Pleomorphic LCIS In the present study, both cores with pleomorphic LCIS had associated calcification. In contrast to the controversial nature of the association between calcification and classical LCIS, the literature suggests that there is a clear relationship between calcification and pleomorphic LCIS [29, 50]. There are limited data on the significance of pleomorphic LCIS on core biopsy. One of the two patients in the present study had DCIS in the excision specimen. Three of the six patients described in the literature [5, 20, 37] had invasive carcinoma on excision, but radiology showed a density in one and an architectural distortion in another. Although the follow-up data are limited, we believe that, until there is further evidence, it is prudent to recommend that pleomorphic LCIS is managed as DCIS because of the morphological similarity of pleomorphic LCIS to high-grade DCIS. The present study may underestimate the frequency of pleomorphic LCIS, as E-cadherin immunohistochemistry was infrequently performed early in the study period and some cases may have been diagnosed as high-grade DCIS. Some lobular neoplasia shows features overlapping with DCIS and is difficult to classify even with E-cadherin immunohistochemistry [32, 38]. Lobular neoplasia with comedo necrosis, but without marked nuclear pleomorphism, has recently been described [22]. The clinical significance of these rare clinico-pathological scenarios is uncertain. Until there is more evidence, it would be prudent to recommend surgical biopsy of such lesions because of the overlap of features with DCIS. Mass forming lobular neoplasia can occur rarely [56], but is an example of clinico-pathological discordance, so surgical biopsy is prudent. In conclusion, we recommend excision after a core biopsy diagnosis of lobular neoplasia in the following circumstances: Discordance of clinical or radiological findings with the pathological changes (this includes mass-forming lobular neoplasia)Lobular neoplasia with atypical histological features including pleomorphic LCIS, lobular neoplasia with necrosis, and when it is not possible to exclude DCIS despite E-cadherin immunohistochemistryIf there is an associated risk lesion, such as atypical intraductal epithelial proliferation, radial scar or papillary lesion. We suggest that further studies are needed to guide the management of simple classical lobular neoplasia, as the data on the risk of associated carcinoma remain unclear in this group of patients.

Anal_Bioanal_Chem-3-1-1592252

Identification of sulfation sites of metabolites and prediction of the compounds’ biological effects

Characterizing the biological effects of metabolic transformations (or biotransformation) is one of the key steps in developing safe and effective pharmaceuticals. Sulfate conjugation, one of the major phase II biotransformations, is the focus of this study. While this biotransformation typically facilitates excretion of metabolites by making the compounds more water soluble, sulfation may also lead to bioactivation, producing carcinogenic products. The end result, excretion or bioactivation, depends on the structural features of the sulfation sites, so obtaining the structure of the sulfated metabolites is critically important. We describe herein a very simple, high-throughput procedure for using mass spectrometry to identify the structure—and thus the biological fate—of sulfated metabolites. We have chemically synthesized and analyzed libraries of compounds representing all the biologically relevant types of sulfation products, and using the mass spectral data, the structural features present in these analytes can be reliably determined, with a 97% success rate. This work represents the first example of a high-throughput analysis that can identify the structure of sulfated metabolites and predict their biological effects. Introduction The study of drug metabolism is one of the critical steps in drug development [1–4]. At physiological conditions, drugs undergo a variety of biotransformations, which produce metabolites with different chemical structures. The resulting metabolites differ in both pharmacological and toxicological properties compared to their parent drugs [1]. Therefore, to identify and characterize the structures and properties of drug metabolites formed in vivo, an extensive investigation of the structures generated by biotransformation is required. Sulfate conjugation (sulfation) is one of the most important phase II reactions that occurs during the biotransformation of a variety of structurally diverse endogenous compounds, xenobiotics, and drugs [5–9]. During this process, a sulfonate moiety (SO3) is transferred from the donor 3′-phosphoadenosine-5′-phosphosulfate (PAPS) to the substrates through the catalysis of sulfotransferase enzymes, resulting in sulfated products [10]. Sulfation is generally considered as a detoxification pathway, because the sulfated products are more water-soluble; this facilitates their elimination from the body [11–13]. However, sulfation can also lead to the bioactivation of certain type of compounds which include benzylic, allylic alcohols and aromatic hydroxylamines. Sulfated products of these compounds can undergo loss of , resulting in reactive electrophilic carbocation or nitrenium ion intermediates, which covalently bind to cellular macromolecules, leading to mutagenicity and carcinogenicity [6, 14, 15]. The production of the electrophilic intermediates relies on the structural feature of sulfation sites, as shown in Fig. 1. Under physiological conditions, the sulfate anions in substrates a–c in Fig. 1 act as good leaving groups to form resonance-stabilized intermediates [6, 14]. For the other types of sulfated products in Fig. 1, this process is not favorable due to instability of cation products. Fig. 1Mechanism of forming bioactive carbocation or nitrenium ion intermediates from sulfated products. Substrates a, b and c undergo the bioactivation process while d–g do not. a sulfated benzylic alcohol; b sulfated allylic alcohol; c sulfated aromatic hydroxylamine; d) sulfated aliphatic alcohol; e sulfated phenol; f sulfated aliphatic amine; g sulfated aromatic amine The biological sulfation of tamoxifen is one example that demonstrates how the sulfation site dictates the biological fate of the molecule. Tamoxifen is a pharmaceutical that is widely used in the treatment and prevention of breast cancer. It is hydroxylated during phase I metabolism to generate two products: 4-hydroxytamoxifen (4-OH-TAM) and alpha-hydroxytamoxifen (α-OH-TAM). The sulfation of 4-OH-TAM and that of α-OH-TAM have completely different biological effects. The sulfation of 4-OH-TAM, which occurs at a phenol site (as in Fig. 1), leads to detoxification, whereas the sulfation of α-OH-TAM, which involves sulfation of a benzylic alcohol (as in Fig. 1a), leads to bioactivation [16, 17]. This example (along with many others) demonstrates the fact that the biological effects of sulfation can be determined by characterizing the sulfation site of the substrate. It is widely known that aromatic or aliphatic hydroxyl and amine groups are the major sites in molecules that can be sulfated during metabolism [18]. However, identifying the sulfation sites can be difficult because many potential sulfation sites can be produced or modified during phase-I biotransformations [5]. For example, hydroxyl groups can be added to a benzene ring or carbon chain through aromatic or aliphatic hydroxylation, generating new potential sulfation sites as in the tamoxifen example. An amine group can be changed to a hydroxylamine group through N-oxidation [1], which modifies the structural feature of the potential sulfation site. Since these new sites are formed in vivo, characterizing the biological effect of their sulfation can be problematic. It is thus essential to develop a method that can characterize the sulfation sites of unknown sulfated metabolites. With this information, the biological effect of sulfation can be estimated, and the properties of sulfated metabolites can be characterized. Analytical techniques such as NMR spectroscopy and mass spectrometry have been utilized to identify sulfated metabolites [19–31]. NMR is a powerful technique that can not only detect sulfated metabolites [20, 21, 24], but also help to determine the position of sulfation in the substrate molecules [19, 22, 23]. However, due to the relatively high sample requirements, the complexity of data analysis, and the need to characterize purified compounds, the application of NMR in identification and characterization of drug metabolites is limited in high-throughput metabolite profiling. Compared to NMR, mass spectrometry is more widely used in drug metabolite identification, due to its high selectivity and sensitivity, low detection limit, and ability to analyze mixtures [32]. It is known that mass spectrometry can identify sulfation by detecting the 80 Da mass increase in MS mode, or the characteristic ions in MS/MS mode [21, 25–31]. However, limited information about structural features of sulfation sites currently can be obtained from MS/MS analysis, with the well-known exception that the product ion m/z 97 () can be used to differentiate alicyclic sulfates from aromatic sulfates [25, 26]. In order to facilitate the identification of sulfation sites in pharmaceuticals, we developed a mass spectrometry-based protocol that differentiates between the biologically relevant sulfation sites. After completing the studies on several classes of sulfated products, a set of rules was developed to predict the sulfation sites. With these prediction rules, structural feature of sulfation sites can be determined by detecting MS/MS fragmentation pathways of their corresponding sulfated products. With the structural information of sulfation sites, the two key different biological effects of sulfation, “detoxification” and “bioactivation”, can be differentiated, based on literature precedence that links the type of sulfation to its biological fate. Based on that information, relevant properties of sulfated products can be estimated. Experimental Reagents Sulfates and sulfamates which include 4-nitrocatechol sulfate dipotassium salt, L-ascorbic acid 2-sulfate dipotassium salt, indoxyl sulfate potassium salt, β-estradiol 3-sulfate sodium salt, potassium 4-nitrophenyl sulfate, β-Estradiol 3-sulfate sodium salt, 5-Br-4-Cl-3-indolyl sulfate potassium salt, 4-methylumbelliferyl sulfate potassium salt, 2-aminoethyl hydrogen sulfate, poly (vinyl sulfate) potassium salt, D-glucose 6-sulfate potassium salt, chondroitin disaccharide Δdi-6S sodium salt, N-acetylglucosamine 6-sulfate sodium salt, N-cyclohexylsulfamic acid, 3-hydroxypropyl-sulfamic acid monopotassium salt, D-glucosamine 2-sulfate sodium salt, and butyl-sulfamic acid were purchased from Sigma-Aldrich (St. Louis, MO). One sulfamate, 4-methylphenyl- sulfamic acid, was purchased from Scientific Exchange, Inc. (Center Ossipee, NH). (R)(+)-α-phenethylsulfamic acid was purchased from Norse Laboratories (Newbury Park, CA).The benzylic alcohols, allylic alcohols, hydroxylamines which include (R)-(+)-α-methyl-2-naphthalenemethanol, benzyl alcohol, (R)-1-phenyl-2-propen-1-ol, 4-chloro-2-methylbenzyl alcohol, 3-ethoxybenzyl alcohol, 2-ethoxybenzyl alcohol, 4-ethoxybenzyl alcohol, crotyl alcohol, furfuryl alcohol, N-methylhydroxylamine hydrochloride, N-isopropylhydroxylamine hydrochloride, N-cyclohexylhydroxylamine hydrochloride, N, N-diethylhydroxylamine and N-benzoyl-N-phenylhydroxylamine were purchased from Sigma-Aldrich (St. Louis, MO). These benzylic alcohols, allylic alcohols, and hydroxylamines were utilized to synthesize their corresponding sulfated products. The other reagents used in the sulfation of alcohols and hydroxylamines were also purchased from Sigma-Aldrich (St. Louis, MO). Sulfation of benzylic, allylic alcohols and tertiary hydroxylamines Sulfated products were prepared based on slight modifications to a previous protocol [33] by dissolving 1.0 equivalent (5.0 mmol) of the substrate in 5.0 ml dimethylformamide (DMF) and adding sulfur trioxide-dimethylformamide complex (DMF-SO3 5.5 mmol) and pyridine (5.5 mmol). The reaction mixture was stirred at 40 °C for 1 h. H2O (35 ml) was added to the product solution, and the product was extracted with 3×15 mL ethyl acetate (EtOAc). Evaporation of the solvent afforded the corresponding sulfated product. Sulfation of secondary hydroxylamines To a dry THF (10 ml) solution of NaH (10.0 mmol) was added secondary hydroxylamines (5 mmol) at 0 °C. Then the solution was stirred at room temperature for 2 h, followed by reaction with DMF-SO3 (5.0 mmol) overnight. The corresponding sulfated products were obtained by evaporation of THF. Sample pretreatment Sulfated compounds that were purchased were dissolved to a final concentration of 1.0×10−4 M with 50% MeOH/ 50% H2O, and directly injected into the mass spectrometer. Sulfated benzylic and allylic alcohols and tertiary hydroxylamines were prepared by diluting 10 μl of the liquid product by 100 fold with 50% MeOH/50% H2O, and then injected to the mass spectrometer. Sulfated secondary hydroxylamines were prepared by dissolving 10 mg solid product in 1.0 ml 50% MeOH/ 50% H2O, followed by injection into the mass spectrometer. Mass spectrometry The mass spectra were recorded using a Quattro Ultima (Waters Corp., Milford, MA) triple quadrupole mass spectrometer equipped with an electrospray ionization source. Data was acquired in the negative ion mode using a capillary voltage of 2.80 kV and a cone voltage was 45 V. The source temperature and desolvation gas temperature were 80 and 150 °C, respectively. Argon is used as the collision gas, and the pressure in collision cell is 1.7E(−3) mbar. The collision energy applied in MS/MS for all the samples was 35 eV. Results and discussion The structures of the sulfated products are in Fig. 2. These compounds are representative of different types of sulfated products that are produced by sulfation at different sites. Mass spectrometric studies were conducted in the negative ion mode since sulfated products are deprotonated at physiological pH [11, 15]. The common product ions from the MS/MS experiments are summarized in Table 1. As the table indicates, common product ions include m/z [M-H-80], m/z 80, m/z 96 and m/z 97. The presence of m/z [M-H-80], m/z 96 or m/z 97 ions are quite dependent on the site of sulfation. The presence of the negative ion with m/z 80, however, is common in the fragmentation of almost every type of sulfated products, so it is less helpful in identification of sulfation sites. Thus this ion, m/z 80, is used to identify sulfation sites when no other characteristic ions can be detected. Based on the MS/MS data, each type of sulfated product possesses a characteristic fragmentation pathway, which is described below. An example of MS/MS data for each type of sulfated product is provided in Fig. 3 with the characteristic ion labeled by an asterisk. Fig. 2Structures of selected sulfated compounds a sulfated aromatic alcohols or enols; b sulfated aliphatic alcohols with β hydrogens on sp3carbons; c sulfated benzylic or allylic alcohols; d sulfated hydroxylamines; e sulfated aminesFig. 3Characteristic (-) ESI-MS/MS data for compounds in this study: a sulfated aromatic alcohol, from group a; b sulfated aliphatic alcohol with β hydrogen on sp3carbon, from group b; c sulfated benzylic alcohol, from group c; d sulfated aliphatic hydroxylamine, from group d; e sulfated aromatic hydroxylamine, from group d; f sulfated aliphatic amine, from group e; g sulfated aromatic amine, from group eTable 1Characteristic fragmentation of selected sulfated productsGroupNo.Chemical nameAbundance of characteristic ions in MS/MS (%)m/z [M-80]-m/z 80m/z 96m/z 97a14-nitrocatechol sulfate100a–––2L-ascorbic acid 2-sulfate31a–––3Indoxyl sulfate24a52––44-nitrophenyl sulfate100a–––5β-estradiol 3-sulfate100a–––65-Br-4-Cl-3-indolyl sulfate47a59––74-methylumbelliferyl sulfate100a–––b82-aminoethyl hydrogen sulfate –1001867a9Poly (vinyl sulfate)–––97a10D-glucose 6-sulfate–––100a11Chondroitin disaccharide Δdi-6S –––31a12N-acetylglucosamine 6-sulfate–––93ac132-naphthalenemethanol, α-methyl- sulfate–6.77.3a10014Benzyl sulfate–16100a–15(R)-1-phenyl-2-propen-1-sulfate–26100a12164-Cl-2-methylbenzyl sulfate––100a95173-ethoxybenzyl sulfate–10013a6.4182-ethoxybenzyl sulfate–10017a8.5194-ethoxybenzyl sulfate–10020a8.720crotyl sulfate–5591a5021furfuryl sulfate––53a39d22N-methyl-hydroxylamine-O-sulfonic acid–10073a9.223N-isopropyl- hydroxylamine-O-sulfonic acid–68100a5.124N-cyclohexyl-hydroxylamine-O-Sulfonic acid–27100a1125N,N-diethyl-hydroxylamine-O-sulfonic acid–92100a1926N-benzoyl-N-phenyl-hydroxylamine-O-sulfonic acid––100a–e27N-cyclohexylsulfamic acid–100a––284-methylphenyl-Sulfamic acid–100a––293-hydroxypropyl-sulfamic acid–100a––30D-glucosamine 2-sulfate–100a––31butyl-sulfamic acid–100a––32(R)(+)-α-phenethylsulfamic acid–100a––*A threshold of 5.0% is used for the relative abundance of characteristic ions. The long dash (–) means that the characteristic ion can not be detected or the relative abundance is below 5.0%aRepresents the specific characteristic ion of each group that can help to identify the sulfation site Group a: sulfated aromatic alcohols or enols Compounds 1–7 are sulfated aromatic alcohols or enols. For these compounds, the sulfate group is attached to an sp2 carbon (benzene ring or double bond). These sulfated products share the same fragmentation pathway by undergoing the neutral loss of 80 Da, and forming a characteristic ion at m/z [M-H-80], as seen in Fig. 3a. After undergoing the neutral loss of SO3, phenoxide or enolic anions are formed, and the negative charge is resonance stabilized, as shown in Scheme 1 [33]. Scheme 1Sulfated aromatic alcohols or enols (group a) dissociate to produce a characteristic ion with neutral loss of 80 Da. Group b: sulfated aliphatic alcohols with β hydrogens on sp3carbons The second group of sulfated products originates from the sulfation of aliphatic alcohols: compounds 8–12. For these species, the sulfate group is attached to an sp3 carbon. All sulfated products of this type can produce the characteristic ion m/z 97 in MS/MS, and one example is shown in Fig. 3b. The ion m/z 97 is generated when the proton from the β carbon is transferred to the sulfate moiety, and the C-O bond is broken to form a bisulfate anion [33]. See Scheme 2. Scheme 2Sulfated aliphatic alcohols (group b) produce the characteristic ion m/z 97 This rearrangement occurs via an energetically accessible six-membered ring transition state, and a stable neutral product is formed. As a result, this fragmentation is predominant in the MS/MS data of this group of sulfated products. The mechanism shows that the availability of a β hydrogen on an sp3 carbon is required for this fragmentation. Unlike compounds in group a, neutral loss of 80 Da is not observed from this group of compounds. This is likely due to the fact that the product ion that would be generated from loss of 80 Da (SO3) is not resonance stabilized, so the loss is not favorable. Compound 8 is a special case because it gives a characteristic ion with m/z 96, in addition to m/z 97. The relevant mechanism for this loss is addressed in the discussion of compounds in group d. Group c: sulfated benzylic or allylic alcohols In the third group of sulfated compounds (13–21), each contains a sulfate group attached to benzylic or allylic carbon. All the compounds in this group produce a characteristic ion with m/z 96. An example of an MS/MS data for this type is shown in Fig. 3c. The characteristic ion is produced by homolytic cleavage, producing benzylic or allylic radicals. The radical is resonance stabilized, which is demonstrated in Scheme 3 [33]. Scheme 3Sulfated benzylic and allylic alcohols (group c) produce the characteristic ion m/z 96 In addition to m/z 96, the negative ion m/z 97 can also be obtained for some of the compounds in this category. For example, compounds 13 and 16 both have an abundant peak at m/z 97, due to the availability of a hydrogen on the sp3 carbon in close proximity to the sulfate group. Even though compound 13 is a benzylic sulfate, a hydrogen attached to the β sp3 carbon is present. Therefore, just like compounds in group b, it undergoes a rearrangement to form the ion, m/z 97. Compound 16 possesses a methyl group in the ortho position, and the availability of hydrogen attached to the sp3 carbon in methyl group enables a similar rearrangement (eight-membered ring) to form the product ion, m/z 97, as illustrated in Scheme 4. Even though β hydrogens are available in other benzylic and allylic sulfates in this group, the fragmentation to produce m/z 97 is not as favorable, since the hydrogens are on the sp2 hybridized carbons. The neutral loss of 80 Da is not favorable either, because the product ion that would be generated from such a loss is not resonance stabilized. Scheme 4Fragmentation mechanism producing m/z 97 from compound 16 Group d: sulfated hydroxylamines The fourth group of sulfated products originates from sulfation of hydroxylamines, and as a result, the sulfate group is directly attached to the amine: See compounds 22–26. All the sulfated products of this group can produce the characteristic ion m/z 96, and examples of MS/MS data are shown in Fig. 3. A characteristic ion with m/z 96 can be obtained for sulfated products originating from both aliphatic (Fig. 3d) and aromatic (Fig. 3e) hydroxylamines. For these sulfated compounds, homolytic cleavage is favorable. This is due to the small electronegativity difference (about 0.5) between N and O, compared to electronegativity difference of around 1.0 for C –N and O-S [34]. The small electronegativity difference causes the electrons to be split equally (homolytic cleavage) between the oxygen and nitrogen when the bond breaks, instead of both electrons moving onto the oxygen or nitrogen, which would be the case during heterolytic cleavage. In Scheme 5, the mechanism of this homolytic cleavage is depicted. An amine radical is formed as the product. Scheme 5Sulfated hydroxylamines (group d) produce the characteristic ion m/z 96 This mechanism can also be used to explain the presence of the ion with m/z 96 in the MS/MS spectrum of compound 8. Scheme 6 shows how this ion is produced from compound 8. In this case, homolytic cleavage can be used to distribute a single electron onto the nitrogen. Compounds that contain a nitrogen two carbons away from the sulfation site, like this one, could also produce m/z 96 as a product ion. Scheme 6Fragmentation mechanism producing m/z 96 from compound 8 Group e: sulfated amines The last group of sulfated products, sulfamates, originates from the sulfation of amines. Compounds 27–32 belong to this type. The only characteristic ion for this type of sulfated products is the negative ion, m/z 80. This characteristic ion can be obtained from fragmentation of sulfated products originating from both aliphatic amines (Fig. 3f) and aromatic amines (Fig. 3g). This group of sulfated products has another type of bond, a N-S bond, that has a small electronegativity difference between the two atoms (0.5) [34], making the homolytic cleavage between the nitrogen and sulfur favorable. The mechanism that shows production of m/z 80 is illustrated in Scheme 7. No other fragmentation pathway is as favorable as this homolytic cleavage pathway, making m/z 80 the predominant ion in all of the MS/MS data for these compounds. Scheme 7Sulfated amines (group e) produce the characteristic ion m/z 80 Prediction rules For each type of sulfated product, there is a predominant fragmentation pathway that results in a characteristic product ion during MS/MS fragmentation. The specific characteristic ion for each group is obtained from every sulfated product in that group, as shown in Table 1: the ion with the neutral loss of 80 Da can be seen in MS/MS spectrum of every sulfated product in group a; an ion with m/z 97 can be detected in all the MS/MS data of group b compounds; an ion with m/z 96 can be obtained from MS/MS data of all the sulfated products of both group c and group d. A negative ion with m/z 80 can always be found in MS/MS data of sulfated products in group e. We have shown that characteristic fragmentation pathways of sulfated products are dependent upon the structural features of the sulfation sites. Conversely, relevant structural information of sulfation sites can be obtained by determining characteristic fragmentation pathways or characteristic ions. To facilitate this information transformation, a set of prediction rules is described as follows: If the characteristic ion with a neutral loss of 80 Da can be detected, this indicates that the sulfate group is attached to an sp2 carbon, the sulfated product is therefore produced by the sulfation of a phenol or an enol.If the characteristic ion with m/z 96 can be detected, the sulfated product is produced by sulfation of a benzylic or allylic alcohol, or a hydroxylamine.If the characteristic ion with m/z 97 can be obtained, and it is more abundant than the ion, m/z 96 (if it is detected), the sulfate group is attached to an sp3 carbon. This is the sulfated product produced by sulfation of aliphatic alcohols with an available β hydrogen attached to an sp3 carbon;If none of the ions, m/z [M-H-80]−, m/z 96, or m/z 97 can be detected, and only the ion m/z 80 is observed, then the sulfated product is produced by sulfation of an amine. There are occasions when the conditions in rule 2 and 3 might both be observed. For example, both of these apply to compound 13, which suggests that the sulfation site is not only an alcohol with available β hydrogen (on an sp3 carbon), but also a benzylic or allylic alcohol; and this estimation is consistent with the actual structure of compound 13. Among the selected 32 sulfated products, 31 of them follow the proposed prediction rules. The exception, compound 8, possesses an amine group in the β position. The amine at this position enables the compound undergo the fragmentation pathway to produce ion with m/z 96 (see Scheme 6). In this case, rule 2 incorrectly predicts the sulfation site for compound 8. As a result, the obtained data in Table 1 demonstrates a 97% (31/32=0.97) success rate of these prediction rules. Potential application of the prediction rules in characterizing unknown sulfated metabolites As one of the major phase II biotransformations, sulfate conjugation is involved in the metabolism of an enormous range of substrates [5]. In general, sulfation is a detoxification or deactivation pathway, but it can also bioactivate the substrate molecules when sulfation sites are benzylic, allylic alcohols (group c) or aromatic hydroxylamines (group d) [6, 8, 14]. According to the prediction rules developed herein, sulfation at any of these sites will generate products that produce the ion m/z 96 in MS/MS, so the structural feature of the sulfation sites can be determined by rule 2. Sulfation of other sites, like phenols, aliphatic alcohols or amines, will cause other characteristic ions to appear in MS/MS data. Therefore, the sulfation sites can be determined by rule 1, 3 or 4. In these cases, sulfation facilitates detoxification. With the proposed prediction rules, biological effects of sulfation can be determined, and the properties of sulfated metabolites can be estimated. Figure 4 demonstrates how the proposed method works to characterize unknown sulfated metabolites. Since high-quality MS/MS data is a necessary prerequisite of using this method, if the metabolite is present in a complex biological matrix, some mass spectral optimization may be necessary. After obtaining reliable (-)ESI-MS/MS data on the unknown sulfated metabolite, the prediction rules are used to determine sulfation site by identifying the characteristic ions present in the MS/MS data. If the MS/MS data matches rule 1, 3 or 4 (instead of rule 2), the sulfation sites can be phenols (or enols), aliphatic alcohols (with β hydrogen on sp3 carbon) or amines. In these cases, the sulfation undergoes a detoxification pathway and the sulfated metabolite is more easily eliminated, compared to the parent drug. If the MS/MS data is consistent with rule 2, the sulfation site should be benzylic, allylic alcohols or hydroxylamines. This would indicate that the sulfation might follow a bioactivation pathway, and lead to carcinogenic sulfated products. One potential caveat to this second condition is that a few nontoxic sulfated metabolites that have structures similar to compound 8, might also produce the characteristic ion with m/z 96, following the mechanism in Scheme 6. Another possible drawback of this characterization is that the diagnostic ion m/z 96 does not discriminate between sulfated products of aliphatic and aromatic hydroxylamines; and thus far only aromatic hydroxylamines have been proven to be toxic. However, distinguishing between aromatic and aliphatic hydroxylamines is not critically important because sulfated aliphatic hydroxylamines are rather uncommon metabolites. While it is possible that a few compounds could be misassigned as “toxic”, none of the sulfated metabolites that are known to be bioactivated would be misdiagnosed as “nontoxic”, as demonstrated herein. Fig. 4The method for characterizing unknown sulfated metabolites Conclusion A method was developed to determine the structural features of sulfation sites, by detecting the characteristic fragmentation pathway of the corresponding sulfated products in (-) ESI-MS/MS. By summarizing MS/MS data from five different types of sulfated products originating from different sulfation sites, their characteristic fragmentation pathways and characteristic ions were determined. Based on this information, a set of prediction rules was developed to transfer information about the fragmentation pathway of sulfated products to the structural features of the sulfation site. As a result, the proposed prediction rules can be applied in drug metabolite profiling to characterize sulfation sites, to further estimate the biological effect of sulfation, and to evaluate relevant properties of sulfated metabolites.

Anal_Bioanal_Chem-4-1-2270354

Neutron activation analysis and X-ray Rayleigh and Raman scattering of hair and nail clippings as noninvasive bioindicators for Cu liver status in Labrador Retrievers

The heritability of chronic hepatitis in the Labrador Retriever is studied with the aim of identifying the related gene mutation. Identification of cases and controls is largely based on instrumental neutron activation analysis (INAA) Cu determination in liver biopsies. The burden for these companion animals may be reduced if nail clippings and hair (fur) could serve as a noninvasive indicator for the hepatic Cu concentrations. No correlation was found between hepatic Cu concentrations and Cu concentrations in hair and nail samples. However, hair and nail samples were also analyzed by X-ray tube excitation, taking advantage of the X-ray Compton, Rayleigh, and Raman scattering which reflects the organic components such as the type of melanin. Principal component analysis provided first indications that some differentiation between healthy and sick dogs could indeed be obtained from hair and nail analysis. Introduction Chronic hepatitis in dogs is a histological diagnosis, characterized by the presence of fibrosis, inflammation, and necrosis of the liver cells. Symptoms of chronic hepatitis are quite variable but the most common signs are lethargy, loss of appetite, and diarrhea. Pets may also drink and urinate more. As the disease progresses, many dogs develop yellowish gums, eyes, and skin (icterus/jaundice), and a swollen abdomen which is filled with fluid (ascites). In some cases toxins affect the nervous system and the dogs become blind and obtuse. This can progress to seizures, coma, and death. The disease is characterized, amongst others, by copper accumulation in the liver. Previous studies have shown elevated hepatic Cu concentrations in the order of 1,000–2,000 mg kg−1 compared with normal Cu levels of < 200–500 mg kg−1. Such hepatic copper accumulation can result from increased uptake of copper, a primary metabolic defect in hepatic copper metabolism, or from altered biliary excretion of copper. Several studies have already been conducted at Utrecht University on inherited copper toxicosis and hepatic copper storage in companion dogs such as Bedlington Terriers [1] and Dobermanns [2]. Currently, one of the programs is investigating the genetic basis of the disease by examination of family members from Labrador Retrievers with copper-associated chronic hepatitis. The selected dogs undergo blood sampling, and retrieval of a liver biopsy (by use of the Menghini technique). The liver biopsy is examined histologically and stained for copper granules. As this project deals with companion animals from the Labrador population in The Netherlands and not with test animals, cooperation of the dog owners is crucial. Following the Delft group’s involvement and over 20 years’ experience investigating nail clippings as a bioindicator for the trace element status in man [3, 4], a feasibility study was undertaken to assess the usefulness of nail clippings and hair (fur) as a noninvasive indicator for the Cu liver levels with the aim of replacing the transcuteaneous full-needle biopsy. Two approaches were tested. The most straightforward was to verify if Cu levels in nail clippings and hair would reflect the Cu liver levels. The other approach was initiated by the observation that a copper deficiency may result in hair/coat color changes. This is ascribed to the activity of cuproenzymes, which catalyze the biosynthesis of melanin from L-tyrosine. A copper deficiency causes, among others, pigmented hair on the head and face to lose its normal color and get a “washed out” appearance and become gray [5]. In 2005, Bueno et al. demonstrated that X-ray Rayleigh peaks and especially the peakless part in the X-ray Raman scattering continuum in X-ray spectra obtained by tube excitation carry information on the organic constituents of biological materials [6, 7]. A study of Poodles in which all spectral information was processed by dedicated principal component analysis (PCA) resulted in a clear differentiation of all dogs examined on the basis of their hair color and, moreover, the separation of healthy and sick dogs on the basis of the hair analysis. Hair pigmentation is due to two major types of melanin, viz., eumelanin and phaeomelanin (Fig. 1). The type of melanin manifesting in the hair can be attributed to the hair color genotype [8, 9], whereas the intensity and distribution of the granules may also be affected by diseases in which sulfur metabolism is involved. Fig. 1Chemical structures of eumelanin and phaeomelanin Both approaches were tested here, i.e., (i) Cu determination via instrumental neutron activation analysis (INAA) in liver biopsies, nail clippings, and hair samples; and (ii) X-ray spectrum acquisition after X-ray tube excitation, followed by full spectrum data (i.e. all channel contents) processing by PCA, including the often neglected peakless part of the X-ray scattering region resulting from Compton, Rayleigh and Raman effects. Experimental Liver biopsy samples, nail clippings, and hair were collected from approximately 100 dogs with chronic hepatitis as well as from healthy animals. Nail clippings were taken from phalanx I (the thumb nail), since this nail is not in contact with the ground, thus minimizing contamination problems. Nail clippings could not be collected from all animals as in some cases breeders had removed the nail permanently shortly after birth, since sometimes such a thumb nail hinders the dog in working trials. Hair was sampled from the median abdominal wall, just cranial to the belly button. The liver biopsy samples were freeze-dried at the veterinary clinic. The nail clippings and hair samples were analyzed without cleaning to prevent changes in the nail and hair morphology that might affect the scattering information from the X-ray fluorescence (XRF) analysis. It should be noted that a Cu contamination of the nail clipping and hair is highly unlikely; the nail of the thumb nail is never in contact with, e.g., soil, and external contamination of the hair was also unlikely. INAA The needle biopsies were transferred into polyethylene capsules and lyophilized, resulting in sample masses varying typically between 1.5 and 50 mg dry weight. Nail clippings had masses of 5–50 mg, and the typical mass of the hair samples was approximately 100–200 mg. The Cu levels were established by INAA via the determination of 66Cu. This is the radionuclide of choice although 64Cu is also produced; however, the associated 511-keV peak is insufficient for reliable Cu determination, whereas the 1,345-keV peak provides insufficient sensitivity. Quantification of all element mass fractions was based on the single comparator method [10]. Samples were irradiated one-by-one for 3 min in the Hoger Onderwijs Reactor in a thermal neutron fluence rate of 1.7 × 1017 m−2 s−1. A Zn flux monitor was included in each irradiation; the neutron fluence rate was determined by measurement of the induced 69mZn activity. The activated biopsy, nail, and hair samples were measured 30 s after irradiation for 3 min at a 1-cm distance from the end-cap of a side-looking 12% Ge detector. The samples were rotated during counting. The Zn monitors were measured in the same geometry for 60 s after a 4-min decay time. The samples were processed in batches of 14 samples, one internal quality control sample, and a blank. A sample of NBS 1577b Bovine Liver (certified Cu content 160 ± 8 mg kg−1) was used for internal quality control. Details of the irradiation facility, gamma-ray spectrometer, and associated quality assurance have been published elsewhere [11]. The spectrometer is equipped with ORTEC DSPEC-PLUS electronics, which allows for adequate dead-time correction [12]. It is obvious that, particularly in the nail and hair samples, several other elements such as Ca, Mg, and S were also quantified on the basis of their short halflife radionuclides, measured simultaneously with 66Cu. X-ray tube excitation The measuring procedure consisted of weighing approximately 100 mg of each hair sample and 10 mg of each nail sample directly into appropriate cells and submitting them to blank rhodium X-ray tube radiation, in triplicate. A common laboratory energy dispersive XRF instrument (EDX 700, Shimadzu) was used under the same irradiation conditions: 50-kV applied voltage in the tube, 25% dead time, 10-mm beam collimation, and 100-s irradiation time. It should be noted that the methodology of using the region in the X-ray spectrum resulting from scattering implies that only the spectrum’s shape is needed for the data processing by PCA. There is no need for quantification towards Cu mass fractions, and hence no reference materials were used for quality control as was the case in the hair, nail, and liver biopsy analysis by INAA. The chemometric method [13] used for data analysis was PCA, processed via the software The Unscrambler, version 9.2, from Camo. Results and discussion Cu could not be determined in all available nail samples because of the small sample masses and the detection limit of the analytical protocol selected. Detection limits on the basis of 66Cu were typically in the order of 25 ng, corresponding to a minimum detectable Cu mass fraction of 5 mg kg−1 in a 5-mg nail clipping sample. Cu levels in hair and nail are given in Fig. 2, together with the associated Cu liver levels. Uncertainties are not depicted for clarity reasons; typical uncertainties (combined standard uncertainty) were approximately 5–20% for the liver analyses, 15–30% for the nail, and 5–10% for the hair analyses. It is clear from Fig. 2 that the Cu levels in hair and nail do not reflect the Cu liver levels. This indicates that the gene(s) responsible for the Cu accumulation in the liver is(are) only active in the liver cells and not in the cells at the nail base or in the hair pocket. Fig. 2Cu levels in nail clippings and hair of Labrador Retrievers as a function of the Cu levels in liver biopsies taken from the same dogs: ○ hair, ● nail; uncertainties are omitted for clarity (see text) Principal component analysis (PCA) results of the processing of the channel contents of the X-ray spectra of all hair samples are given in Fig. 3, and the related factor loadings of PC1 and PC2 are shown in Fig. 4 (similar results were found for the processing of the channel contents of the X-ray spectra of the nail samples, although these are not shown here). The PC1 loading graph indicates the dependent variables that explain most of the data variance, and the PC2 loading graph indicates variables that explain the second data variance. PC1 and PC2 are orthogonals and therefore independent of each other. Fig. 3PCA analysis of all hair samples, classified here by the hair color: □ yellow, * chocolate, ■ blackFig. 4PC1 loadings for hair samples (top). PC2 loadings for hair samples (bottom) The variables included in the PCA analysis were from the channel contents between 0 and 40 keV, i.e., covering the entire spectum recorded.The symbols in Fig. 3 reflect the hair color. The differentiation with respect to coat color results most likely from differences in type of melanin in Labradors with a black, liver (chocolate), or yellow-colored coat. Table 1 shows the genotypes of the hair color inheritance of Labrador Retrievers [8], in which each character represents a specific gene (denoted in pairs, of course). An uppercase character represents a dominant gene, a lowercase character represents a recessive gene; no character in a gene pair implies that it is not relevant for the genotype if the locus is occupied by a dominant or recessive gene. As produces black without any tan on the dog and there are other genes that can modify the black to liver (chocolate Labrador). If As is present, in most cases the dog will be able to produce only eumelanin pigment. Some shades of liver (chocolate), though a eumelanin pigment, overlap some shades of tan, a phaeomelanin pigment. The e, recessive red, overrides whatever gene is present at the A locus to produce a dog which shows only phaeomelanin pigment in the coat. The related melanin types are also included in Table 1 and the color differentiation by the principal component analysis is most probably based on the differences in melanin corresponding to the different genotypes. The wide spread of the groups might reflect homozygotes and heterozygotes for the genes involved. Table 1Labrador genotypes [8] for hair color inheritance and corresponding types of melaninColorGenotypeMelanin typeBlackAs–  B–  E–EumelaninLiver (chocolate)As–  bb  E–EumelaninYellow, liver nose– –  B–  eePhaeomelaninYellow, black nose– –  bb  eePhaeomelanin The PCA also revealed a differentiation to gender, as is further exemplified in Fig. 5 for nails. This is in agreement with earlier work for Poodles in which a similar gender separation was observed [6]. Fig. 5PCA analyses of all nail samples, classified by gender These results (Figs. 3, 4, 5) indicate that PC1 allows for a separation of dogs by gender and PC2 for separation by color, particularly when using hair samples. The loading graphs of the PCs (Fig. 4) show that sulfur and Rh scatter (Compton, Rayleigh, and Raman) explain the variance in PC1. This is most likely attributed to variations in (sulfur bondings in) keratin structures. This effect contributes to almost 100% of the data variance. The rest of it, PC2, is not very large (approximately 1–10%) and allows for separations due to differences in sulfur, and to some extent with melanin. The physiological cause of the variation of the keratin structures in hair and nail by gender is not yet fully clear, but there are several indications that the structure of hair is different for males and females. It is well known that there is a relationship between testosterone and hair formation (andropogenic hair). Andropogenic hair has a different growth rate and a higher weight than other hair. It has also been observed that the type of sulfur is different for females and males: rhombic (alpha) sulfur predominates in female hairs, monoclinical (beta) sulfur in male hairs [14]. Less information is available about nails. Genetic variation in the proteins of the human nail has been observed [15], as well as a sex variation in lipid composition of human fingernail plates [16]. Another possible explanation for the separation of dogs in terms of genders in the hair analysis might be found in the results of Scott et al. [17] who concluded that the human melanocortin 1 receptor (MC1R) is regulated by, amongst others, specific endocrine sex hormones and by UV radiation. In confirmation of the latter study, Broekmans et al. [18] found that male subjects were more sensitive to UV irradiation than female subjects. It was already known that eumelanin is photoprotective, whereas phaeomelanin may contribute to UV-induced skin damage due to its potential to generate free radicals in response to UV radiation [19]. Skin melanin content, which was positively associated with hair color in men, was the main phenotypical determinant of sensitivity to UV irradiation. The PCA results could also be plotted as a function of the INAA Cu liver biopsy values. Results for hair and nail are given in Fig. 6 for a differentiation towards dogs with Cu liver levels of < 600 mg kg−1 or > 600 mg kg−1. A Cu biopsy level of > 600 mg kg−1 is considered to be positively indicative for liver Cu accumulation. In addition, plots were also made in which the hair and nail samples are identified by correlating with Cu biopsy values > and < 200 mg kg−1 as well as at > and < 400 mg kg−1. At these lower Cu levels no clear differentiation is obtained, as could be expected, which confirms that the PCA is not producing an artifact, accidentally coinciding at the 600 mg kg−1 Cu liver biopsy level. Fig. 6PCA results, nail X-ray spectral data vs. INAA Cu liver biopsy values (top): ■ dogs with Cu liver levels 0–600 mg kg−1, □ dogs with Cu liver levels > 600 mg kg−1 , N = 60. PCA results, hair X-ray spectral data vs. INAA Cu liver biopsy values (bottom): ■ dogs with Cu liver levels 0–600 mg kg−1, □ dogs with Cu liver levels > 600 mg kg−1, N = 43 In all PCA, it could be justified to remove some outliers on the basis of high leverage values and high energy variance residuals in PCs. High values of leverage and residuals are caused, for instance, by imperfect quality and positioning of the sample during the X-ray irradiation. In addition, outliers may sometimes result from the presence of dogs with strong parental relationships [6]. The strength of using the information of the scattered radiation is emphasized by the role of sulfur in the separation towards gender and Cu accumulation. Total sulfur, as determined by INAA via 37S, hardly shows any correlation with INAA Cu liver values (correlation coefficients 0.03 and 0.32 for nail and hair, respectively). However, the INAA results provide only the total sulfur content, i.e., sulfur present in keratin, melanin, and any other compound, whereas the PCs of the channel contents of the X-ray spectra, especially due to the peakless scattered region, apply to sulfur directly related to keratin and/or melanin. This explains the much better correlation of “keratin/melanin” sulfur with Cu liver values compared with total S with Cu. The results of our study show that a fairly sharp differentiation can be obtained between dogs with low and high Cu liver levels if PCA is performed on a dataset obtained from all channel contents of X-ray spectra obtained by X-ray tube excitation of nail or hair samples together with the spectral information of often neglected X-ray (Rayleigh, Raman) scattered radiation. A practical application of this technique would start with construction of a dataset using samples of known origin, encompassing all varieties expected. New and unknown samples can then be projected on this model, and subsequently classified. This approach will be further validated with new samples to be collected from Labrador Retrievers still being observed and treated along medicinal and dietary approaches. Conclusions Hair and nail clippings are widely used as bioindicators for the trace element status of man and, to a much lesser extent, also for animals. The results of this study emphasize that it is quite risky to assume a priori that the trace element levels in hair and nail reflect differences between classes of individuals, e.g., “healthy” and “sick” people or animals. In this study related to chronic hepatitis, the Cu levels in liver biopsies are not reflected by Cu levels in hair and nail. The X-ray Rayleigh scattered and X-ray Raman continuum contain valuable information about the organic, low-Z elemental matrix composition of an object. In this study, first indications were obtained that the Cu accumulation possibly may have an effect on the nail and hair melanin, thus offering a potentially favorable outlook for noninvasive monitoring of the Cu liver status. An additional advantage of the approach is that a measurement by X-ray tube excitation takes only about 2 min per sample so that, once a database has been built up, the technique might be valuable for large-scale screening purposes.

J_Gastrointest_Surg-3-1-1852374

Risk of Fecal Diversion in Complicated Perianal Crohn’s Disease

The purpose of the study was to determine the overall risk of a permanent stoma in patients with complicated perianal Crohn’s disease, and to identify risk factors predicting stoma carriage. A total of 102 consecutive patients presented with the first manifestation of complicated perianal Crohn’s disease in our outpatient department between 1992 and 1995. Ninety-seven patients (95%) could be followed up at a median of 16 years after first diagnosis of Crohn’s disease. Patients were sent a standardized questionnaire and patient charts were reviewed with respect to the recurrence of perianal abscesses or fistulas and surgical treatment, including fecal diversion. Factors predictive of permanent stoma carriage were determined by univariate and multivariate analysis. Thirty of 97 patients (31%) with complicated perianal Crohn’s disease eventually required a permanent stoma. The median time from first diagnosis of Crohn’s disease to permanent fecal diversion was 8.5 years (range 0–23 years). Temporary fecal diversion became necessary in 51 of 97 patients (53%), but could be successfully removed in 24 of 51 patients (47%). Increased rates of permanent fecal diversion were observed in 54% of patients with complex perianal fistulas and in 54% of patients with rectovaginal fistulas, as well as in patients that had undergone subtotal colon resection (60%), left-sided colon resection (83%), or rectal resection (92%). An increased risk for permanent stoma carriage was identified by multivariate analysis for complex perianal fistulas (odds ratio [OR] 5; 95% confidence interval [CI] 2–18), temporary fecal diversion (OR 8; 95% CI 2–35), fecal incontinence (OR 21, 95% CI 3–165), or rectal resection (OR 30; 95% CI 3–179). Local drainage, setons, and temporary stoma for deep and complicated fistulas in Crohn’s disease, followed by a rectal advancement flap, may result in closing of the stoma in 47% of the time. The risk of permanent fecal diversion was substantial in patients with complicated perianal Crohn’s disease, with patients requiring a colorectal resection or suffering from fecal incontinence carrying a particularly high risk for permanent fecal diversion. In contrast, patients with perianal Crohn’s disease who required surgery for small bowel disease or a segmental colon resection carried no risk of a permanent stoma. Introduction Crohn’s disease was initially described as a nonspecific inflammatory bowel disease, affecting mainly the terminal ileum and characterized by a subacute or chronic necrotizing and cicatrizing inflammation.1 Eventually, gastrointestinal Crohn’s disease became recognized as a full-thickness disease of the gastrointestinal wall that may affect the entire gastrointestinal tract, including the perianal region.2–6 Perianal lesions are common in patients with Crohn’s disease.7–13 Clinical manifestations vary from asymptomatic skin tags to severe, debilitating perianal destruction and sepsis. Asymptomatic perianal lesions require no treatment, but because they become painful and disabling, they may require surgical treatment. Surgical management needs to be conservative and should focus on the drainage of septic sites, preserving sphincter function and palliating symptoms.10–13 Medical management has had some success in improving symptoms, but as yet, it has not been able to eliminate most perianal complaints permanently.14,15 Fecal diversion was successfully used to achieve remission in colonic Crohn’s disease. Moreover, it was utilized to allow severe perianal disease to settle, thereby avoiding proctectomy.16–19 However, restoring the intestinal passage carries the risk of recurrent perianal disease activity, possibly resulting in a decreased quality of life compared to the situation with fecal diversion. We investigated the overall risk of a permanent stoma in patients with severe perianal Crohn’s disease and tried to identify risk factors predicting permanent stoma carriage. Methods Patients A total of 102 consecutive patients with Crohn’s disease presented with the first manifestation of a perianal fistula or a perianal abscess in our outpatient department between 1992 and 1995. Patients were investigated in Trendelenburg’s position by perianal inspection, proctoscopy, rectoscopy, and rigid sigmoidoscopy. Endoanal ultrasound was performed in case of suspected perianal abscess formation, and MRI was conducted of the pelvic floor in case of complicated fistulizing disease or intrapelvic abscess formation. All patients were documented prospectively. Follow-up data of 97 patients (95%) were available by a standardized questionnaire mailed to the patients and by a standardized chart review. There were 50 female and 47 male patients (ratio 1.06:1) with a median age of 23 years (range 8–51 years). Patients were evaluated with respect to the recurrence of perianal abscesses, fistulas, or surgical treatment of Crohn’s disease over the years. The median interval between the first diagnosis of Crohn’s disease and last follow-up was 16 years (range 8–37 years). Four patients had isolated small intestinal disease, 11 patients had isolated colonic disease, and 82 patients had small intestinal and colonic disease. The abscess location was categorized as subcutaneous, intersphincteric, deep perianal, ischiorectal, and above the pelvic floor. Abscess formations were categorized into simple (subcutaneous, intersphincteric, deep perianal, and ischiorectal, circular extension less than 90°, pelvic floor not involved) and complicated (circular extension more than 90° [horse shoe abscess] or pelvic floor involved). Fistulas were classified according to Parks et al.20 into subcutaneous, intersphincteric, extrasphincteric, transsphincteric, rectovaginal, and suprasphincteric, as described previously.21 We divided fistulas into simple fistulas (no more than two perianal openings) and complex fistulas (rectovaginal, three or more perianal openings). A variety of factors, such as sex of the patient, perianal fistula, rectovaginal fistula, abscess formation, anal stricture, fecal incontinence, or abdominal surgery were evaluated with regard to their predictive character for permanent stoma carriage by univariate and multivariate analysis. Statistical Analysis Kaplan–Meier analysis using a log-rank test was used for comparing risk rates over time. Factors that might influence permanent fecal diversion were analyzed using the chi-square test. Fisher’s exact test was used for univariate analysis and multiple logistic regression and Wald’s test for multivariate analysis. Subgroup analysis was performed for rectovaginal fistulas, as present in female patients only. Data are given as numbers of cases and percentages or median and interquartile ranges. A P < 0.05 was considered as significant. Results Overall Risk of Permanent Fecal Diversion Thirty of 97 patients (31%) with perianal Crohn’s disease eventually required a permanent stoma. Nineteen patients were female and 11 male (ratio 1.73:1). The median time from first diagnosis of Crohn’s disease to permanent fecal diversion was 8.5 years (range 0–23 years; Fig. 1). Increased rates of permanent fecal diversion were observed in 53% of patients with previous temporary fecal diversion, in 54% of patients with complex perianal fistulas, in 54% of female patients with rectovaginal fistulas, and in 83% of patients with fecal incontinence (Table 1). Patients who required subtotal colon resection (60%), left-sided colon resection (83%), or rectal resection (93%) needed a permanent stoma at high rates (Table 2). Figure 1Relative proportion of patients without permanent stoma in complicated perianal Crohn’s disease (n = 97) during follow-up.Table 1Patients with First Manifestation of Complicated Perianal Crohn’s Disease (n = 97) and the Rate of Fecal Diversion During Follow-up PatientsTemporary StomaPermanent StomaOverall9751 (53%)30 (31%)Abscess formation7532 (43%)21 (28%)Simple abscess formation33 (44%)13 (39%)7 (21%)Complex abscess formation42 (56%)23 (55%)15 (36%)Fistulas8851 (58%)26 (30%)Simple perianal fistulas42 (48%)17 (40%)8 (19%)Complex perianal fistulas46 (52%)34 (74%)23 (50%)Rectovaginal fistulas26 (54% of ♀)18 (69% of 26)14 (54% of 26)Overlap between abscess formation and presence of fistulas exists, and abscesses led to fecal diversion in combination with fistulas only.Table 2Abdominal Procedures and the Rate of Fecal Diversion During Follow-up in Patients with Complicated Perianal Crohn’s Disease (n = 97) Patient NumberPermanent Stoma (% of Patients)Small Bowel ProceduresStricturplasty70Small bowel resection because of stenosis230Small bowel resection because of enteroenteric fistula70Anastomosis resection because of inflammatorystenosis300Stoma revision40Large bowel proceduresSegmental colon resection140Right-sided colon resection204 (20)Ileocecal resection6824 (35)Subtotal colon resection3521 (60)Left-sided colon resection65 (83)Rectal resection1312 (92)Patients that required small bowel resections carried no risk of fecal diversion, whereas patients with colon resections carried an increased risk of fecal diversion. Abscess Formation Seventy-five of 97 patients (77%) had at least one perianal abscess at first presentation or during follow-up. Surgical therapy for abscesses consisted of seton drainage, mushroom catheter drainage, or incision and drainage, as described previously.21 A permanent fecal diversion because of recurrent abscess formations, always in combination with fistula problems and perianal sepsis, became necessary in 21 of 75 cases (28%). Simple abscess formations occurred in 33 patients, and 7 of those patients (21%) required permanent fecal diversion (Fig. 2). Complex abscess formations were present in 42 patients, with permanent fecal diversion being necessary in 14 cases (33%; Table 1). Figure 2Relative proportion of patients without permanent stoma in complicated perianal Crohn’s disease with simple (n = 42) and complex (n = 46) fistulas during follow-up. The presence of a complex fistula significantly increased the risk of permanent fecal diversion (P < 0.001). Perianal and Rectovaginal Fistulas Perianal fistulas affected 88 of 97 patients (91%), including 26 female patients with rectovaginal fistulas. Symptomatic fistulas with abscess formation were treated by simple drainage procedures and provided with a seton. Azathioprine therapy was tried if purulent secretion persisted. Infliximab was rarely used in recent years at the discretion of the referring gastroenterologist. Overall, permanent fecal diversion became necessary in 26 of 88 patients (30%). Forty-two patients (48%) had simple fistulas, whereas 46 patients (52%) had complex fistulas. Of these, permanent fecal diversion was documented in 14 patients (54%), whereas simple fistulas eventually required a permanent stoma in only 8 cases (16%). Twenty-six of 48 female patients had rectovaginal fistulas, of which 14 (54%) eventually required permanent fecal diversion (Table 1). Thirty-four of 97 patients (35%) developed anal strictures, whereas fecal incontinence was documented in 12 patients (12%) during the follow-up. In 14 of 34 patients (41%) with anal stricture, a permanent fecal diversion became necessary, whereas 10 of 12 patients (83%) suffering from fecal incontinence required a permanent stoma. Abdominal Surgery Eighty-three of 97 patients with perianal Crohn’s disease (86%) underwent abdominal surgery at least once during follow-up because of Crohn’s disease activity, with a total of 227 abdominal procedures being performed. In 68 of 227 operations (29%), intestinal stenosis required surgery. Segmental colon resection was performed 108 times (46%), whereas subtotal colectomy for fulminant colitis was performed 35 times (15%). One patient required surgery because of a fistula carcinoma. Patients with complicated perianal Crohn’s disease who required a small bowel resection carried no risk of permanent fecal diversion, whereas the majority of patients with left-sided colon resection, subtotal colon resection, or rectal resection needed permanent fecal diversion (Table 2). Patients that needed abdominal surgery repeatedly carried an increased risk for a permanent stoma (permanent stoma rate: three or more abdominal operations, n = 20, 50%; less than two abdominal operations, n = 20, 15%; P < 0.043). Patients with permanent fecal diversion had undergone previous abdominal surgery three times as often as those patients without (P < 0.05; Table 3). Eventually, 17 patients (18%) needed proctectomy, of which 13 were female patients with rectovaginal fistulas. In 14 patients with a permanent stoma, a rectal stump remained in place and was controlled at yearly intervals. Patients with complicated perianal Crohn’s disease and large bowel resection carried a significantly increased risk of permanent fecal diversion compared to patients with complicated perianal Crohn’s disease and small bowel resection (48 vs 0%, P < 0.001) (Fig. 3). Table 3Abdominal Procedures per Patient and Percentage of Patients with Abdominal SurgeryAbdominal ProceduresMedian(25–75%)Patients with Abdominal Surgery (%)P ValuePermanent fecal diversion3(2–4)100P < 0.05Without permanent fecal diversion1(1–3)Temporary fecal diversion3(2–4.25)80P < 0.05Without temporary fecal diversion1(0–2.5)Anal stricture2(1–4)60n.s.Without anal stricture2(1–3)Incontinence2(2–4.5)10n.s.Without incontinence2(1–3)Patients with Crohn’s disease and complicated perianal fistulas who required temporary or permanent stoma had more abdominal procedures than patients who did not need a stoma, whereas patients with anal stricture or fecal incontinence had no increased abdominal procedure rate (the number of abdominal procedures per patient is given as median with interquartile ranges). Abdominal surgery and fecal diversion correlated on univariate analysis (P < 0.05).Figure 3Treatment algorithm for patients with complicated perianal Crohn’s disease. Temporary Fecal Diversion Temporary fecal diversion, done mostly by loop ileostomy, became necessary in 51 of 97 patients (53%). Increased rates of temporary fecal diversion were observed in 55% of patients with complex abscess formations, including those with severe perianal sepsis, in 77% of patients with complex perianal fistulas, and in 69% of female patients with rectovaginal fistulas. In 24 of 51 patients (47%) the temporary stoma could be removed after perianal disease had subsided. Fistulas were closed in these patients by rectal advancement flaps,23 whereas cutting setons, fistulectomy, or infliximab infusions were not employed in these patients. Risk Factors Predictive of a Permanent Fecal Diversion Univariate Analysis Complex perianal fistulas (P < 0.04), fecal incontinence (P < 0.001), and rectovaginal fistulas in female patients (P < 0.001) carried an increased risk for a permanent stoma. Patients with high rates of abdominal surgery had a significantly increased risk for fecal diversion, whereas the frequency of abdominal surgery was not influenced by the presence of anal stricture or fecal incontinence (Table 3). The need for subtotal colon resection (P < 0.001), rectal resection (P < 0.001), or temporary fecal diversion (P = 0.001) also resulted in an increased risk of permanent fecal diversion. A variety of factors did not increase the risk of a permanent stoma, such as sex of the patient, anal stenosis, perianal abscesses, and abdominal surgery, excluding subtotal colon and rectal resection (Table 4). Table 4Risk Factors for Permanent Fecal Diversion by Univariate and Multivariate Analysis in Patients with Complicated Perianal Crohn’s DiseaseRisk FactorsUnivariate AnalysisMultivariate AnalysisPermanent Fecal Diversion (%)P ValueOR95% CIP ValueRectal resection92P < 0.001305–179P < 0.002Fecal incontinence83P < 0.001213–165P < 0.02Subtotal colectomy60P < 0.001Rectovaginal fistulas54P < 0.001Temporary fecal diversion51p = 0.00182–35P < 0.02Complex perianal fistulas54P < 0.0452–18P < 0.03 Multivariate Analysis According to multivariate analysis, complex perianal fistulas (P < 0.03), fecal incontinence (P < 0.02), temporary fecal diversion (P < 0.02), or rectal resection (P < 0.002) still carried a significantly increased risk for permanent fecal diversion (Table 4). Discussion Perianal disease is a feature of Crohn’s disease that afflicts approximately one third of patients, but prevalence rates vary between 8 and 90%.19 Patients with colonic involvement will have perianal disease in more than 50% of cases, whereas patients with small bowel disease are affected in less than 20%.24 Perianal Crohn’s disease can present as minor lesions, such as skin tags, skin excoriations, and fissures, which rarely need surgical treatment. In contrast, perianal fistulas often result in abscesses or perianal sepsis, requiring urgent surgery, and fistulas and abscesses are considered as suppurative complications of perianal Crohn’s disease.25 We conducted surgery as conservatively as possible because sphincter function should be preserved as much as possible.19,22,23 Throughout the study period, 77% of patients developed abscess formations, which is considerably more than the 48% rate reported previously with a shorter mean follow-up of 32 months.22 The incidence of perianal fistulas in Crohn’s disease is somewhat lower, affecting 10–34% of patients.26 Healing rates of 60–70% in patients with Crohn’s disease and perianal fistulas were reported in earlier series.21,26–28 However, these series contained a large proportion of simple fistulas, whereas most of our patients had complex transsphincteric or rectovaginal fistulas and fistulas that did not heal through conservative treatment. Perianal fistulas that resulted in abscess formation were drained surgically, and in case of massive purulent secretion or perianal sepsis, oral antibiotics were used additionally. Setons were put into place to prevent recurrent abscess formation, a treatment that is well known as a possible means for controlling perianal sepsis and preventing recurrent abscess formation.19,21–23,29 We did not use cutting setons nor fistulotomy for transsphincteric fistulas, as fecal incontinence may succeed.19,22,23 Infliximab was successfully shown to reduce fistula secretion, and outer fistula openings may heal.30 However, fistula tracks persist with varying degrees of residual inflammation, which may cause recurrent fistulas and perianal abscesses.31 Rectal advancement flaps can be used to close the inner opening of transsphincteric or rectovaginal fistulas when perianal disease has abated and the rectal mucosa is not inflamed. However, fistulas eventually recurred in about one third of patients after a mean follow-up of 19 months, and some patients developed new fistulas.23 Hyman32 reported an initial 79% healing rate in a series of 14 patients with Crohn’s disease and perianal fistulas, but observed longer term success in only 50% of patients. Our current approach is to use rectal advancement flaps for transsphincteric fistulas while delaying surgery if there is an acute flare of Crohn’s disease, in particular, if proctitis is present.23,33 Perianal fistulas in Crohn’s disease closed by a rectal advancement flap have a recurrence rate of 50–60%, but can be attempted repeatedly.23,34 If a second rectal advancement flap fails, the failure rate increases up to 75%,34 and a temporary stoma is suggested to the patient before a third attempt is made. By this approach, we were able to heal perianal fistulas in 24 out of 51 patients (47%) who required a temporary stoma. In a previous study, diversion was the most powerful factor influencing healing according to multiple regression analysis.23 The chances of a symptomatic fistula recurrence were increased without a stoma (52% vs 14%, 21), and patients with Crohn’s colitis carried an increased risk of fistula recurrence.23 A particular problem is rectovaginal fistulas, which almost always open at the dentate line.19 They occur in 3–10% of patients with Crohn’s disease.19 In our series, 54% of female patients had rectovaginal fistulas, indicating that a large patient proportion had complex perianal fistulas; referral bias might have contributed to this high proportion of rectovaginal fistulas, as our outpatient clinic for Crohn’s disease is well known regionally. Rectovaginal fistulas carry a poor prognosis,18 and a 70% recurrence rate was observed 24 months after a rectal advancement flap was performed.23 In the experience of Keighley et al.35, 11 of 13 patients with rectovaginal fistulas required proctectomy with a permanent stoma. In our series, 8 out of 26 rectovaginal fistulas healed by a rectal advancement flap. Eighteen patients required a temporary stoma, of which four could be removed after a rectal advancement flap was tried again successfully (overall healing rate 12 of 26, 46%), resulting in a permanent stoma rate of 14 of 26 patients (54%). Healing rates of rectovaginal fistulas associated with Crohn’s disease vary widely, and low patient numbers are usually reported.18 In the Cleveland clinic, 16 of 37 rectovaginal fistulas (43%) healed using an endorectal advancement flap, but patients not having Crohn’s disease were included in the study.36 In low anovaginal fistulas, higher healing rates of up to 68% were reported in patients with Crohn’s disease.37 Recurrent rectovaginal fistulas were shown to heal after repeated rectal advancement flap procedures were conducted at about the same rate even in Crohn’s disease, but surgery should be delayed for at least 3 months after a previous repair,38 which is standard policy at our clinic. It is interesting to note that Halverson et al.38 reported that the presence of a diverting stoma significantly increased the fistula recurrence rate (67 vs 50%). After years of perianal or transvaginal pus secretion and recurrent abscess formations, patients are often tired of the suffering involved. Antibiotics, azathioprine, or 6-mercaptopurine may be used successfully in suppurative perianal Crohn’s disease.15,18,38,39 In our department, this is routinely tried if perianal disease does not settle after surgical drainage is provided. However, azathioprine or 6-mercaptopurine are not tolerated by all patients and may also fail to relieve perianal Crohn’s disease. In these cases, fecal diversion can be offered as a relatively minor procedure with low morbidity,18 and might be performed laparoscopically in selected cases.40 However, overall healing rates of perianal disease are only around 40%, with 87% of those being diverted eventually retaining their stoma.18 We diverted 51 patients, of which 27 retained there stoma (53%), which compares favorably, but indicates a considerable risk for a permanent stoma if diversion is tried, and being a significant risk factor according to multivariate analysis in our series. Perianal disease in combination with fecal incontinence might necessitate fecal diversion as well. Factors that may contribute to fecal incontinence in patients with Crohn’s disease are reduced stool consistency, sphincter injuries after abscess incisions, and keyhole deformities after laying open perianal fistulas. The latter was reported to result in fecal incontinence in 5 of 27 patients with perianal fistulas in Crohn’s disease,41 which is why we did not use this technique or cutting setons in transsphincteric fistulas. We always incise perianal abscesses along a circular perianal line, meticulously preserving sphincter integrity. Nevertheless, fecal incontinence became prevalent in 12% of our patients during follow-up and was a significant predictor of permanent fecal diversion by multivariate analysis. Rectal advancement flap procedures may influence continence,22 and a 9% rate of worsened continence was reported thereafter.33 In particular, the risk of fecal incontinence was increased after flap repairs in patients with previous surgical fistula repairs,33 but only 2 out of 12 patients with fecal incontinence had a previous rectal advancement flap repair in our study. A high rate of abdominal procedures was associated with an increased risk for a temporary or a permanent stoma, possibly reflecting high intestinal disease activity of Crohn’s disease in these patients. All 30 patients that ended with a permanent stoma had frequently undergone abdominal procedures (median of three procedures). Because of the suffering involved and unsuccessful previous attempts to heal perianal fistulas, about half of our patients eventually opted for temporary ileostomy, being brought into contact with a stoma nurse and stoma carriers with Crohn’s disease before, and 31% of all patients remained with a permanent stoma. In general, the need for a stoma was reduced during recent decades in Crohn’s disease,43 but the overall long-term risk for a stoma in patients with Crohn’s disease who require abdominal surgery was reported between 30 and 40%.44,45 Fecal diversion remains an option to subside perianal disease activity, with an early response rate of about 70–80%.46–48 Unfortunately, 75% of those patients eventually experienced a relapse, and the restoration of intestinal continuity was achieved in 10% only.48 In 130 patients with surgically treated perianal Crohn’s disease, a permanent stoma became necessary in 24%, and this rate increased to 53% in patients with Crohn’s colitis.45 In this light, considering that all of our patients had complicated perianal disease and almost all (93 of 97 patients, 96%) had colonic involvement, we judge a 31% permanent stoma rate as a success in a subset of patients carrying a particularly high risk for a stoma. Local drainage, setons, and, if not successful, a temporary stoma after a rectal advancement flap, resulted in closure of the stoma in 47% of the time, which is an improvement compared to previous studies where restoration of intestinal continuity was reported in only 10–40% of patients, with most studies providing the lower end percentages.44,46–48 If perianal Crohn’s disease was the indication for creating a stoma, stoma closure was successful in 2 out of 15 patients only (13%).47 Eventually, proctectomy became necessary in 18% of patients, which is in the range others have reported.24,44,49,50 However, if the presence of colonic disease was considered separately, seven out of eight patients with complicated perianal Crohn’s disease needed proctectomy, again indicating that an 18% proctectomy rate in our high-risk population is rather low. All of our patients who needed proctectomy had colonic disease, and 13 of 17 had a rectovaginal fistula. In a recent study investigating the quality of life in patients with Crohn’s proctocolitis, patients in remission had a health-related quality of life similar to controls. Patients with active disease had a reduced quality of life, and only the symptom index negatively predicted a reduced quality of life, whereas neither previous colonic surgery nor the presence of a stoma did.51 Accordingly, carrying a stoma does not necessarily mean a reduced quality of life, although a slight reduction in two out of eight domains (physical and emotional role) was detected.51 However, in patients with perianal disease, the cumulative abscess rate during 3 years of follow-up and the symptomatic fistula recurrence rate with a mean follow-up of 22 months were both greatly reduced with a stoma (76 vs 13% and 52 vs 14%), indicating reduced suffering and need for surgery.21,22 In summary, patients with complicated perianal Crohn’s disease, colonic involvement, and a high rate of abdominal procedures carried a significant risk for a permanent stoma. However, if the risk factors identified were taken into account, the rate of patients eventually requiring a permanent stoma seemed low and showed a decrease compared to previous studies, indicating that multiple treatment episodes and complex surgery, including temporary fecal diversion, might eventually heal at least some patients in this high-risk population, as was also observed by others.50

Plant_Mol_Biol-3-1-1805040

Genome-wide transcriptional analysis of salinity stressed japonica and indica rice genotypes during panicle initiation stage

Rice yield is most sensitive to salinity stress imposed during the panicle initiation (PI) stage. In this study, we have focused on physiological and transcriptional responses of four rice genotypes exposed to salinity stress during PI. The genotypes selected included a pair of indicas (IR63731 and IR29) and a pair of japonica (Agami and M103) rice subspecies with contrasting salt tolerance. Physiological characterization showed that tolerant genotypes maintained a much lower shoot Na+ concentration relative to sensitive genotypes under salinity stress. Global gene expression analysis revealed a strikingly large number of genes which are induced by salinity stress in sensitive genotypes, IR29 and M103 relative to tolerant lines. We found 19 probe sets to be commonly induced in all four genotypes. We found several salinity modulated, ion homeostasis related genes from our analysis. We also studied the expression of SKC1, a cation transporter reported by others as a major source of variation in salt tolerance in rice. The transcript abundance of SKC1 did not change in response to salinity stress at PI stage in the shoot tissue of all four genotypes. However, we found the transcript abundance of SKC1 to be significantly higher in tolerant japonica Agami relative to sensitive japonica M103 under control and stressed conditions during PI stage. Introduction Rice is one of the most economically important crops of the world. Rice yields can be reduced by up to 50% when grown under moderate (6 dS m−1) salinity levels (Zeng et al. 2002). Soil salinity is a serious constraint to rice cultivation under irrigated agriculture in countries like Pakistan and elsewhere due to poor quality of water (Abdullah and Ahmad 1982). Rice is the major crop of many coastal regions which are prone to sea water ingress during high tide and of rainfed systems which experience increasing salinity in the root zone due to a rising water table. Despite significant yield losses under saline conditions, rice is potentially a good candidate for cultivation in areas prone to coastal flooding due to its ability to tolerate standing water in the field for long periods. In California, salinity is an increasing concern for rice cultivation due to irrigation practices that involve use of recirculating water systems and the requirement of water holding after pesticide application during early growth stages (Scardaci et al. 1996). The susceptibility of rice to salinity stress varies with growth stage. Rice is relatively salt-tolerant at germination, and in some cases is not affected significantly by up to 16.3 dS m−1 of salinity (Heenan et al. 1988; Khan et al. 1997). Rice becomes very sensitive at the young seedling stage, which impacts the stand density in salt-affected fields (Lutts et al. 1995). It is relatively less sensitive during late reproductive stage (grain ripening). From an agronomic perspective, sensitivity of yield components is of primary importance. Salinity significantly reduces the tiller number per plant, spikelet number per panicle, fertility, panicle length and primary branches per panicle (Heenan et al. 1988; Cui et al. 1995; Khatun et al. 1995; Zeng et al. 2002). Reductions in tiller number per plant and spikelet number per panicle were reported to be the major causes of yield loss in one cultivar of rice under salinity stress (Zeng and Shannon 2000). The number of spikelets per panicle was determined to be the most sensitive yield component. This component is determined at early reproductive stage, around panicle initiation (PI) (Hoshikawa 1989; Counce et al. 2000). The loss in spikelet number per plant was most significant when the stress was imposed before PI (3-leaf stage) or between PI and booting stage (∼16 days after PI) (Zeng et al. 2001). This loss of potential spikelets is attributed to degeneration of primary and secondary branches and flower primordia (Zeng et al. 2001). Considerable genetic variability for this yield parameter was reported based on evaluation of several diverse genotypes for salinity stress response (Zeng et al. 2002). Considerable variation for salt tolerance at critical stages in the cultivated gene pool has also been reported by other workers (Yeo and Flowers 1982; Moradi et al. 2003). This genetic variability can be utilized for improvement of salt tolerance by focusing on specific yield components. However, the underlying genes conferring tolerance during PI are at present unknown. In the current study, we have focused on the salinity stress imposed during the PI stage considering the significant impact it has on rice yields and a lack of prior information on molecular components of the response to salinity stress at this critical stage. The genotypes selected for this study have contrasting physiological responses to salinity stress at PI based on several seasons of experimentation at the U.S. Salinity Laboratory, Riverside (Zeng et al. 2002). One of the most salt-tolerant genotypes identified by Zeng et al. (2002) was IR63731, an indica line developed from a cross between tolerant land race Nona Bokra and sensitive cultivar IR28. This genotype had the most favorable combination of salt tolerance at early vegetative and PI stages. This genotype possesses good yield potential under salinity stress due to high tillering ability and panicle weight (Zeng et al. 2002). Another genotype, Agami, was ranked high for its ability to yield high under salinity stress. Agami is a salt-tolerant japonica subspecies cultivar grown in Egypt. Several studies have reported the popular California japonica cultivar M103, to be very sensitive at all stages (Zeng et al. 2002, 2003). Therefore, for the current work we have used IR63731 and Agami as the tolerant genotypes representing indica and japonica subspecies and IR29 and M103 as the sensitive genotypes for each of the subspecies. Previously, we focused on transcript accumulation responses of rice to salinity stress imposed during the early vegetative stage in two indica subspecies genotypes which differed in salt tolerance as manifested by higher tillering capacity and maintenance of favorable ion homeostasis (Walia et al. 2005). The sensitive genotype used for that study was IR29, a widely used sensitive standard which is susceptible at the vegetative and reproductive stages. IR29 was the sensitive parent of a recombinant inbred population developed at the International Rice Research Institute (IRRI) for mapping salt tolerance quantitative trait loci (QTL) in rice using Pokkali as the tolerant parent. A salt-tolerant recombinant inbred line (RIL) FL478 from this population was used as the tolerant genotype in our prior study (Walia et al. 2005). The transcriptome of rice under salinity stress at PI has been largely uncharacterized. Considering the sensitivity of rice yield to salinity stress when imposed around PI, it is important to identify specific genes that are involved in the response to stress at this stage. Responsive genes are expected to include those with a role in salt adaptation and tolerance, and which respond to salt injury or constitute non-adaptive responses. The use of genotypes with well-characterized phenotypic responses to salinity stress at a particular growth stage can help in identifying potential candidate genes which may be important for heritable salt tolerance. Here we perform whole genome transcript profiling of genotypes from two major subspecies of cultivated rice chosen for their representative tolerance or sensitivity to salinity stress during the early reproductive stage, specifically at PI. By including indica genotype IR29 in the present study on PI, we now have mRNA expression level responses at both of the two most salt sensitive, yield-determining stages for this genotype (early vegetative as per Walia et al. 2005, and early reproductive in this study). Physiological characterization of the lines was also performed as part of the present study, and gene expression results were interpreted in the physiological context. We also incorporated information from known salt tolerance QTL in rice with the objective of correlating gene expression with gene location on the rice genome. Material and methods Plant materials for expression studies Seeds of rice (Oryza sativa L.) genotypes IR29, IR63731-1-1-4-3-2 and Agami were obtained from G. B. Gregorio at IRRI of the Philippines and then propagated through quarantine at the George E. Brown, Jr. Salinity Laboratory, Riverside, CA, USA. Seed for M103 was supplied by California Cooperative Rice Research Foundation Inc., at Biggs, California. These genotypes were selected from the germplasm collections at the two sites based on their differences in salt tolerance in terms of agronomic and physiological performance. Extensive characterization of these lines was performed by Zeng et al. (2002, 2003) at George E. Brown, Jr. Salinity Laboratory, Riverside, CA, USA. Plant culture and salinity treatments The experiment was conducted in the greenhouse at Riverside, California [33°58′24″ N latitude, 117°19′12″ W longitude] between May and July 2005. Plants were cultured in tanks (122 × 61 × 46 cm) filled with sand and irrigated with nutrient solution (Yoshida et al. 1976). The nutrient solution consisted of NH4NO3 (1.43 mM), NaH2PO4 · 2H2O (0.37 mM), K2SO4 (0.5 mM), CaCl2 (1.00 mM), and MgSO4 · 7H2O (1.6 mM). Nutrient solution pH was maintained between 5.0 and 6.5 by adding sulfuric acid as needed. Irrigation solutions were prepared in 1600 L reservoirs and pumped to provide irrigation to the sand tanks. Overflow irrigation was returned to the reservoirs through drainage by gravity. Each reservoir provided irrigation water to three sand tanks (replicates) three times daily for 30 min per irrigation cycle. Four rows of each genotype were planted per tank for control tanks. An extra row at each end of the tank was used as the border row. Each genotype was planted in separate tanks which were marked for salinization later in the experiment. The water level was maintained at the sand surface until emergence of the seedlings from the sand and then increased to 6–8 cm above the sand surface. Air temperature ranged from 30°C to 42°C during the day and 20–24°C during night. Relative humidity ranged from 40% to 80%. Light averaged 1104 μmol m−2 s−1 at noon. Weak plants were removed from the tanks. Plant growth stages were closely monitored. Rough estimates of the days to PI stage for each genotype were available from growing the genotypes for several years in the same greenhouse conditions and time of the year during previous experiments (Zeng et al. 2002). Salinity treatment was applied by adding NaCl and CaCl2 (5:1 molar concentration) in two steps over a period of 3 days to each genotype about 9–10 days before PI (Fig. 1). Electrical conductivities (ECw) of nutrient solutions were monitored using an EC meter daily (Hanna Instruments, Woonsockett, RI). The salinity in the treatment reservoir was allowed to stabilize for 5 days to a final electrical conductivity of 7.0 dS m−1. The non-saline control reservoir with Yoshida solution had an ECw of 1.1 dS m−1. Two plants were randomly selected each day for each genotype when the plants were approaching PI. The main culms were dissected under a microscope to observe the development of young panicles. The day when we observed any plant of a genotype with panicle about 0.5 mm long was defined as the PI date for that genotype under the treatment in which it was growing (control or salt stressed). Salinity delayed PI in the sensitive genotype M103, but not in other three lines. Fig. 1Experimental design of salinity stress treatment. Salinity stress was applied 9–10 days before panicle initiation (PI). A final salinity level of 7 dS m−1 was reached by a two-step addition of NaCl and CaCl2. Conductivity in the control tanks is represented by a dotted line and the treated tank by a solid line. Diamonds represent the PI and ovals represent the harvest time points. The crown and growing point of the main shoot were harvested 4–5 days after reaching PI Phenotypic characterization and gene expression studies Plants were characterized for phenotypic responses to salinity stress on the day of harvest for RNA extraction. For phenotypic characterization the shoot tissue from six plants was harvested for ion analysis. Plants were washed with deionized water, dried in a forced air oven (70°C), and then ground into fine powder. Shoot Na+ and K+ concentrations were determined on nitric-perchloric acid digests by inductively coupled plasma optical emission spectrometry (ICP, Perkin-Elmer Co., Norwalk, CT, USA). Transpiration rate of the youngest fully expanded leaf was measured between 10.00 h and 12.00 h one day before harvesting the tissue for RNA using a LI-COR 6400 Photosynthesis System (LI-COR Biosciences, Lincoln, NE). The following conditions for leaf gas measurements were used: photosynthetic photon flux density, 1200 μmol m−2; chamber CO2 concentration, 380 μmol CO2 mol−1; leaf temperature, 27°C; and chamber vapor concentration, 20 mmol H2O mol−1. The plants were harvested 4 or 5 days after reaching PI. The main shoot was dissected for RNA extraction to obtain the growing point and crown tissue which was snap frozen. Approximately 12 plants were harvested per genotype per tank and tissue pooled to make each biological replicate for RNA extraction. Three biological replicates (one per array) were used for each treatment. Two plants from each tank were allowed to grow to maturity to verify that plants survived the imposed salinity stress. RNA extraction and processing for GeneChip analysis RNA samples were processed as recommended by Affymetrix, Inc. (Affymetrix GeneChip Expression Analysis Technical Manual, Affymetrix, Inc., Santa Clara, CA) at the Core Instrumentation Facility at the University of California, Riverside by Barbara Walter. Total RNA was initially isolated from frozen shoot tissue using TRIzol Reagent. The RNA was purified using an RNeasy spin column (Qiagen, Chatsworth, CA) and an on-column DNase treatment. Eluted total RNAs were quantified with a portion of the recovered total RNA and adjusted to a final concentration of 1 μg/μl. All RNA samples were quality assessed prior to beginning target preparation/processing steps by running out a small amount of each sample (typically 25–250 ng/well) onto a RNA Lab-On-A-Chip (Caliper Technologies Corp., Mountain View, CA) that was evaluated on an Agilent Bioanalyzer 2100 (Agilent Technologies, Palo Alto, CA). Single-stranded, then double-stranded cDNA was synthesized from the poly(A)+ mRNA present in the isolated total RNA (10 μg total RNA starting material each sample reaction) using the SuperScript Double-Stranded cDNA Synthesis Kit (Invitrogen Corp., Carlsbad, CA) and poly(T)-nucleotide primers that contained a sequence recognized by T7 RNA polymerase. A portion of the resulting ds cDNA was used as a template to generate biotin-tagged cRNA from an in vitro transcription reaction (IVT), using the Affymetrix GeneChip IVT Labeling Kit. Fifteen micrograms of the resulting biotin-tagged cRNA was fragmented to strands of 35–200 bases in length following prescribed protocols (Affymetrix GeneChip Expression Analysis Technical Manual). Subsequently, 10 μg of this fragmented target cRNA was hybridized at 45°C with rotation for 16 h (Affymetrix GeneChip Hybridization Oven 320) to probe sets present on an Affymetrix Rice Genome array. The GeneChip arrays were washed and then stained (SAPE, streptavidin-phycoerythrin) on an Affymetrix Fluidics Station 450 followed by scanning on a GeneChip Scanner 3000. Rice Genome array The Rice Genome array (Affymetrix, Santa Clara, CA) contains probe sets designed from ∼48,564 japonica and 1,260 indica sequences. The sequence information for this array was derived from NCBI UniGene Build #52 (http://www.ncbi.nlm.nih.gov/UniGene), GenBank mRNAs and 59,712 gene predictions from TIGR’s osa1 version 2.0. Gene models which had any indication of transposable elements were removed from the list of TIGR predicted genes. The array is believed to represent about 46,000 distinct rice genes. About 26,000 of these are 3′ anchored Unigene EST and mRNA clusters, including known rice full length cDNA clones and 19,431 are solely from TIGR gene predictions. To obtain annotations for the salt-regulated probe sets, we extracted the target sequence of identified probe sets from the sequence information file (.sif) for the rice genome array. The target sequence extends from the 5′ end of the 5′-most probe to the 3′ end of the 3′-most probe. The target sequences were then searched using BLASTn against the TIGR rice pseudomolecules, release 3 (http://www.tigr.org/tdb/e2k1/osa1) and the TIGR Arabidopsis database, version 5. Statistical analysis of array data The triplicated array data set was analyzed using GeneChip Operating Software (GCOS 1.2) and DChip (Li and Wong 2001; http://www.DChip.org) software. The scanned images were examined for any visible defects. Satisfactory image files were analyzed to generate raw data files saved as CEL files using the default settings of GCOS 1.2 from Affymetrix. We used a default target intensity value (TGT) setting of 500. The normalization factor value was set to 1 to extract data without normalization. Default parameter settings for the Rice Genome array were used. The scaling factor for the arrays ranged from 3.1 to 8.5. The detection calls (present, absent or marginal) for the probe sets were made by GCOS. Further analysis was done using DChip which incorporates a statistical model for expression array data at the probe level. The DChip program was set to import GCOS signals. The normalization of all arrays was performed using an invariant set approach. For calculating the expression index of probe sets we used the PM model and opted for truncating the low expression values to 10th percentile of the probe set signals called absent. The expression values were log2 transformed after calculating the expression index. DChip was used for comparative analysis of samples from salt stress and control treated plants for both genotypes. We identified the differentially expressed genes using the empirical criterion of more than 10-fold change (E-B > 1 or B-E > 1) and significant t-tests of P < 0.05 based on three biological replicates. The baseline array, which is a control sample, is denoted by B and salinity-treated experimental sample is denoted by E. The P-values assigned to differentially expressed genes were used as a ranking criterion for ordering of up-regulated and down-regulated genes. We used the DChip feature to assess the false discovery rate (FDR) in our data set using 250 permutations in most comparisons. Empirical FDR was below 15% when identifying differentially expressed genes. All microarray data from this work is available from NCBI GEO (http://www.ncbi.nlm.nih.gov/geo) under the series entry GSE4438. Expression validation by semi-quantitative RT-PCR Several key expression profiles obtained from chip hybridizations were further validated by semi-quantitative RT-PCR using first-strand cDNA. The japonica sequence of each gene was obtained from the TIGR rice database. Exonic sequences were used for the design of primers using Primer Express (Perkin-Elmer Applied Biosystems, Foster City, CA). The primer sequences for 18S rRNA (F-ATGATAACTCGACGGATCGC, R-CTTGGATGTGGTAGCCGTTT) were used as an internal control. The primer sequences for SKC1 were obtained from Walia et al. (2005). A cDNA first strand was synthesized using Taq-Man Reverse Transcription Reagents (Applied Biosystems, Forster City, CA; Ref: N808-0234) following the manufacturer’s instructions. Two micrograms of total RNA was converted into cDNA. Each cDNA was diluted 40 times and 5 μL of cDNA was used for three-step PCR. The number of PCR cycles was optimized (37 cycles) for the 5 genes selected for expression validated and 34 cycles for SKC1. Results A moderate level of salinity stress was imposed gradually nine or ten days before PI stage and plants were sampled for phenotypic characterization and expression profiling five days after PI (Fig. 1). All four genotypes used in this experiment were known to reach PI at different time points based on prior characterization of these genotypes over several growing seasons under comparable growing conditions and time of the year. One confounding factor in experiment involving salinity stress is that salinity is known to delay PI in all genotypes. With this in mind we designed our experiment to harvest the control and salinity stressed tissue of each of the genotype at similar developmental stage i.e. 4–5 days after the plants reached PI. This was done because developmental equivalence when comparing control plants with stressed plants within each genotype is more relevant when the focus is on a specific sensitive stage such as PI. Shoot ion uptake of salt-tolerant and sensitive lines Shoot sodium ion analysis when assayed under salinity stress clearly distinguished tolerant from sensitive genotypes. Salinity stress increased the Na+ concentration in all genotypes significantly (Table 1). The increase was nearly 10-fold in all genotypes except IR63731. Sodium ion accumulation in the tolerant genotypes, IR63731 and Agami, was much lower than in the sensitive genotypes. Agami, the salt-tolerant japonica, maintained a lower Na+ concentration even under control conditions and had the lowest Na+ concentration under stressed conditions. Shoot K+ concentration decreased in Agami, while the concentration in other three genotypes did not change significantly under salinity stress. Since K+/Na+ ratio under salinity is considered a more important index of salinity tolerance than ion concentrations separately, we ranked the four genotypes by K+/Na+ ratio in descending order: Agami (4.9) ≈ IR63731 (4.7) > M103 (3.7) > IR29 (2.9). Under control conditions, indicas had a lower K+/Na+ ratio than japonica varieties. In addition to measuring Na+ and K+ levels in the shoot tissue, we also analyzed the chloride levels. Shoot chloride concentrations increased in all genotypes under salinity stress. Sensitive japonica, M103 however, maintained a much lower Cl− level (433 mmol kg−1 compared to 535 mmol kg−1 in Agami) than other genotypes. Table 1Physiological responses of salt-sensitive and salt-tolerant genotypesTreatmentsNa+K+Cl−TrM103 control23.1 ± 1.1872 ± 52.1355 ± 30.412.1 ± 1.6M103 salt236 ± 16.5881 ± 40.2433 ± 32.58.9 ± 0.7IR29 control24 ± 3.0799 ± 23.2351 ± 24.310.8 ± 1.3IR29 salt279 ± 33.2808 ± 49.5525 ± 12.67.6 ± 0.3Agami control16.5 ± .96772 ± 23350 ± 30.57.0 ± 0.6Agami salt140 ± 20.1688 ± 25.8535 ± 12.14.2 ± 0.6IR63731 control25.7 ± 2.9740 ± 30.5279 ± 1111.0 ± 0.8IR63731 salt166 ± 42.7791 ± 33.2545 ± 35.57.2 ± 0.8The shoot Na+, K+ and Cl− concentrations were measured in mmol kg−1 dry weight. Transpiration rate (Tr) is expressed as m−2 s−1. Ion concentration values were obtained from six replicates. Tr values are mean ± 1 S.E. from 6 to 7 replicate measurements To further characterize the physiological status of the four genotypes, we measured the transpiration rate of the youngest fully expanded leaf of the main shoot (Table 1). The transpiration rate decreased in all genotypes under stress. A noteworthy observation was the significantly lower rate of transpiration maintained by tolerant japonica Agami under control and stressed conditions relative to all other genotypes. Tolerance in Agami therefore seems to depend on an avoidance mechanism. We also observed the plants for visual damage symptoms of salinity stress. The sensitive genotypes, especially IR29, had necrosis on some leaves about 1/3 the length of the leaf from the tip roughly one week after the initiation of stress. Expression responses of tolerant and sensitive genotypes to salinity stress We used two japonica and indica rice genotypes with contrasting salt tolerance for global gene expression under salinity stress. The developmental stage targeted in this study was PI. Salinity stress was applied nine or 10 days before PI and the tissue (crown and growing point) was harvested 5 days after the genotypes reached PI stage (Fig. 1; see “Materials and methods”). To define differentially expressed genes we used the following criteria: (1) t-test P-value < 0.05, (2) ≥ 2-fold change, and (3) FDR below 15% (see Materials and methods). Using these criteria we identified 292 and 346 probe sets which were up- and down-regulated in salt-sensitive japonica M103 respectively (Fig. 2). The same selection criteria identified 54 up- and 54 down-regulated probe sets in the salt-tolerant japonica, Agami. A comparable number of probe sets (55) were differentially up-regulated in salt-tolerant indica IR63731 under salinity stress. A total of 35 probe sets were down-regulated in IR63731. The expression response of sensitive indica, IR29 was characterized by up-regulation of 589 probe sets. In contrast, only 57 probe sets were significantly down-regulated in response to salinity stress. Complete lists of probe sets identified as differentially expressed in each of the genotypes are provided as supplemental data (Datasheet 9–11). Five probe sets identified as differentially expressed from the microarray datasets were checked using semi-quantitative RT-PCR (Supplemental Fig. 1). The results agreed for four of the genes examined. The primers for the fifth gene, Os02g52390, did not amplify from the indica subspecies, since they were derived from japonica exonic sequence. Nineteen probe sets were significantly up-regulated in all genotypes in response to salinity stress. A selected list of probe sets which were commonly induced is presented in Table 2. Only one probe set, Os.17057.1.S1_at, with no hits to the database was down-regulated by salinity stress in all genotypes. Fig. 2Overlap of up-regulated probe sets between different genotypes. (A) Overlap between the sensitive genotypes M103 and IR29. (B) Overlap between the tolerant genotypes. (C) Overlap among the japonica subspecies. (D) overlap among the indica subspecies. The up-regulated probe sets were identified using the combined criteria of a 2 or more fold change, P-value threshold of 0.05 and false discovery rate below 15%Table 2Selected genes commonly induced in all four genotypes and their P-values during the salinity stress imposed during reproductive stagePutative functionProbe setRice locusP-valueIR29M103IR63731AgamiWSI76 protein––riceOs.2677.1.S1_atOs07g488300.0020.010.0180.038Low-temperature-induced protein LTI30Os.12633.1.S1_s_atOs11g267900.0010.0090.0020.0005Myb-like DNA-binding domainOs.10333.1.S1_atOs07g028000.00010.0010.0010.001RNase S-like protein (drought induced)Os.12922.1.S1_atOs09g367000.0020.00040.000060.004MtN3Os.16044.1.S1_atOs02g309100.0360.0260.00030.006Low affinity nitrate transporterOs.32686.1.S1_atOs01g651100.00030.00090.01680.007Putative Myb-like DNA-binding ProteinOs.25453.1.A1_atOs03g555900.00020.00270.00360.015Bowman–Birk serine protease inhibitorOs.7612.1.S1_atOs03g608400.0140.0120.0240.008Nitrate transporter (NTP2)Os.45923.1.S1_atOs01g375900.00050.00030.00430.0013Anthocyanin 5-O-glucosyltransferaseOsAffx.4277.1.S1_atOs05g087500.0070.0230.0120.006Putative function is based on BLAST hits against TIGR rice, TIGR Arabidopsis and NCBI nr database We made several comparisons to identify genes which characterize the response of japonica and indica to salinity as well as genes which are expressed in tolerant and sensitive genotypes (Fig. 2). We identified 42 probe sets which were induced commonly between the two japonicas, Agami and M103. Similar comparison between the indica genotypes identified 41 probes sets as commonly induced. In an effort to search for transcripts which may be associated with salt tolerance of Agami, we identified 12 probe sets which were induced in Agami but not in M103. Two of these probe sets (Os.11975.2.A1_at and OsAffx.25546.1.S1_s_at) were induced in Agami but down-regulated in M103. Probe set Os.11975.2.A1_at had no sequence match to known genes. Probe set OsAffx.25546.1.S1_s_at represented a germin-like gene (GPL4), Os03g44880.. We also identified several probe sets which were induced in IR63731 but not in sensitive indica, IR29. These included RD22, a jacalin lectin protein and nitrate reductase (NR1). The expression of RD22 is known to be induced in response to salt stress and dehydration (Yamaguchi-Shinozaki and Shinozaki 1993). Besides focusing on the genes which were differentially expressed in response to salinity stress within each of the genotypes, we also performed genotype–genotype comparisons under control conditions. This analysis was performed to identify genes associated with salt tolerance that may be inherently transcribed at a higher level in one genotype than in another. Interpretation of such genotypic comparisons using plants profiled under control conditions is done in light of the fact that the compared genotypes have similar developmental stage but not same chronological age. We found 539 probe sets which were significantly up-regulated in Agami relative to M103 under control conditions (Supplemental Data Sheet 13). A similar comparison yielded 686 probe sets expressed at a higher level in IR63731 than IR29 (Supplemental Data Sheet 14). We BLAST searched the target sequence of all the probe sets against the rice genomic sequence assembled at TIGR (http://www.tigr.org) to obtain the genomic position of the best BLAST hit gene model. We found several probe sets which were elevated in Agami relative to M103 to be tightly clustered. Clusters on chromosome 1, 5, 10 and 12 were particularly large (Fig. 3A). A dense 19-probe set cluster occurred on the long arm of chromosome 10. This cluster localized to a segment of 1.1 Mb and represented 15 rice gene loci. Another 11-probe set cluster on chromosome 1 extends from 11.07 Mb to 12.61 Mb (1.54 Mb in length). This region is of particular interest because two QTL for salt tolerance have been mapped to this region (Bonilla et al. 2002; Lin et al. 2004). The number of gene clusters obtained from the comparison of indica genotypes was relatively small and most were found on chromosome 1 (Fig. 3B). Two of the clusters on chromosome 1 were located in the region of a salt tolerance QTL Saltol. The first was an 8-probe set (six gene loci) cluster (9.03–9.91 Mb) and the second was a 14-probe set (eight gene loci) cluster extending from 11.19 Mb to 12.62 Mb. The positions of these two clusters may coincide with two separate QTL in the Saltol region reported by Lin et al. (2004) and Bonilla et al. (2002). Additionally, a dense 16-probe set, 537 kb long cluster was located on the distal end of chromosome 1. Fig. 3Rice chromosomes with gene clusters based on genotypic comparisons. (A) Rice gene model clusters on chromosomes 1, 5, 10 and 12 obtained from genotypic comparison of M103 with Agami under control conditions. (B) Rice gene model clusters on chromosome 1, obtained by genotypic comparison of IR29 with IR63731 using unstressed samples. The chromosome displays only up-regulated rice genes in the tolerant lines under control conditions. The gene clusters with shaded area co-localize to previously identified as a salt tolerance QTL on chromosome 1 Genes involved in ion homeostasis Maintenance of low Na+ and a favorable K+/Na+ ratio are desirable traits for rice growing under saline conditions and ultimately manifest as salt tolerance. Considering the importance of ion transporters in exclusion, redistribution, and compartmentalization of ions we searched for and identified several cation transport-related genes which significantly responded to the salinity treatment (Table 3). One such gene was a sodium transport protein (Os02g07830) induced (P < 0.02) in IR29 only. In contrast a sodium/calcium exchanger was down-regulated in M103. Another gene (Os12g25200), encoding for CLC-d chloride channel was induced in M103 and IR29 by 17-fold (P < 0.0001) and 7.3-fold (p < 0.00008), respectively. Several genes associated with transport of K+ were differentially regulated in response to salinity stress. Three of these genes belong to the KT-HAK-KUP family of transporters associated with high affinity K+ uptake from the roots (Santa-Maria et al. 1997). In M103, a transporter (Os08g36340) similar to HAK4 was up-regulated by stress, while HAK5 (Os07g01220) was down-regulated. Another cation channel gene (Os01g52070), with sequence similarity to KAT3 was down-regulated in M103. A potassium transporter, KUP1 (Os09g27580), was induced in IR29 by more than 5-fold (P < 0.002) under salinity stress. We did not detect any cation transporters significantly responding to salinity stress in the tolerant genotypes. Table 3Genes from selected categories which were differentially expressed under saline conditions in one or more rice genotypesPutative functionProbe setLocusM103IR29IR63731AgamiIon homeostasisK+ transporter (HAK4)Os.2358.1.S1_atOs08g363401.20.63––CLC-d chloride channelOs.27207.1.S1_atOs12g252004.162.880.78–Cation channel protein (KAT3)OsAffx.21536.1.S1Os01g52070−1.05–––Na+/Ca2+ exchanger proteinOs.20354.1.S1_atOs03g45370−1.11–––K+ transporter (HAK5)Os.27454.1.S1_xOs07g01220−1.54–––Cation (Na+) transport proteinOs.50590.2.A1_atOs02g07830–1.0––K+ transporter (KUP1)Os.6037.2.S1_x_atOs09g27580–2.51––Flowering and inflorescenceCentroradialis protein (CEN)Os.57548.1.S1_atAt2g275501.551.44––Zinc finger protein (ID1)Os.4766.1.S1_at1.470.62––Constan-like protein (CO6)Os.7920.1.S1_atOs06g444501.391.00.68CCT motifOs.16422.1.S1_s_atOs08g150501.04–––CLAVATA1 receptor kinaseOs.7123.2.S1_x_at0.821.05––AP1-like MADS-box proteinOs.2348.1.S1_a_at−0.91−1.09––Gigantea-like proteinOs.7987.1.S1_atOs01g08700−1.04––−1.74Floral homeotic protein (AP1)Os.12750.2.S1_s_atAt1g69120−1.12–––Flowering locus TOsAffx.15765.1.S1Os06g35940−1.131.7––Homeodomain (KNAT7)Os.4164.2.S1_a_atAt1g62990−1.66–––Sex determination tasselseed 2Os.15281.1.S1_at–3.93––No apical meristem (NAM)Os.37548.1.S1_at–1.81––AuxinCYP83B1Os.11110.1.S1_atAt4g31500−1.71–––Auxin-repressed protein ARP1Os.12735.1.S1_at1.21–––Auxin-responsive proteinOs.20151.1.S1_atAt3g25290–1.27––Auxin transporter PIN1Os.37330.1.S1_at−0.94––−1.1Axi1Os.52961.1.S1_at−1.0–––Phototropic-responsive NPH3Os.56849.1.S1_atAt1g030101.03–––NPK1-related protein kinaseOs.5940.1.S1_at–1.46––Auxin response factor 10Os.8374.1.S1_atOs04g43910−1.17–––Anthocyanin pathwayChalcone, stilbene synthasesOs.11154.1.S1_aOs03g067001.111.460.921.3Phenylalanine ammonia-lyaseOs.25687.1.S1_xOs02g41680–1.25––Isoflavone reductaseOs.32454.1.S1_atAt4g39230−1.96–––4-coumaroyl-CoA synthase 3Os.4377.1.S1_atAt1g65060−1.4–––Flavonoid 3′-hydroxylase Os.46551.1.S1_atAt5g07990–−1.09––Dihydroflavonol 4-reductaseOs.48545.1.S1_atOs04g53810–1.94––Ferulate-5-hydroxylaseOs.9727.1.S1_atAt4g36220−1.64–––Chalcone isomerase (CHI)Os.9929.1.S1_atAt3g55120−1.31–––5-hydroxyferulic acidOsAffx.13783.1.S1At5g54160–1.02––Fold change in each genotype is expressed as log2 (log21.2 is equal to 2.2-fold change numerically). The P-value threshold of 0.05 was used for significant differential expression. A (–) indicates no change detected Since the shoot ion analysis indicated a strong Na+ exclusion phenotype in the tolerant lines, we further explored our data set for any association of cation transporters. On comparing the tolerant Agami with M103 under control and salinity stress conditions, we found a cation transporter (Os01g20160) to be expressing at a higher level in Agami by 2.6-fold (P < 0.01) and 2.8-fold (P < 0.003) compared to M103 under control and salt stressed conditions, respectively. On extending the genotypic comparison to indica genotypes, we found Os01g20160 transcript to be present in IR63731 at a higher level than in IR29 (1.98-fold, P < 0.01) under control conditions. The cation transporter gene 01g20160 is located in the vicinity of the Saltol region (Bonilla et al. 2002) and has a perfect sequence match to the SKC1 gene reported to be the tolerance conferring gene in the salt tolerance QTL studied by Ren et al. (2005). We pursued the expression of this gene using semi-quantitative RT-PCR in roots and shoot tissue of several rice genotypes which were part of the salinity stress experiment conducted during the vegetative stage study (Walia et al. 2005). The expression of 01g20160 in genotypes IR63731, Pokkali, FL478 and IR29 was analyzed in the root and shoot tissues for vegetative stage (V), and shoot tissue of IR63731 and IR29 at the PI stage (R) under control and stressed conditions (Fig. 4). The expression of Os01g20160 increased in response to salinity in the roots of tolerant lines IR63731, Pokkali and FL478 but not in salt-sensitive IR29. The mRNA accumulation of Os01g20160 in shoot tissue increased in response to salinity in all four genotypes analyzed during the vegetative stage. Fig. 4Expression of SKC1 in different indica genotypes. Semi-quantitative RT-PCR was performed to study the expression of SCK1 in the root and shoot tissue of IR63731 and IR29 under control and stressed conditions during vegetative (V) and reproductive (R) stages. Expression was analyzed for roots and shoot tissue of tolerant lines Pokkali and FL478 during vegetative stage of growth. The curly bracket encloses a control and the corresponding salt stressed sample Differential expression of auxin associated genes in rice under salinity stress Phytohormones including auxin indole 3-acetic acid (IAA) play a central role in plant growth, and development. Auxin is also known to mediate responses to light and gravity. In our expression analysis, several auxin-related genes were differentially regulated by salinity stress, indicating that salinity stress is associated with auxin response in rice. Almost all of these auxin related genes were differentially expressed in sensitive genotypes M103 and IR29. Most striking was the response of M103 to salinity in this context. The expression of cytochrome P450 83B1 (CYP83B1) was significantly (P < 0.04) down-regulated under salt stress. The expression level of CYP83B1 is known to determine auxin levels in plants (Barlier et al. 2000). Other auxin related genes down-regulated in response to salinity include a putative axi1 and two auxin response factors. In contrast, a mitogen activated protein kinase cascade gene, NPK1 which is known to suppress early auxin response gene transcription is induced in response to salinity in IR29. An increased level of nonphototropic hypocotyl 3 (NPH3) transcript was observed in M103 under salt stress. A rice ortholog of NPH3 is involved in phototropic response of coleoptiles through lateral auxin translocation (Haga et al. 2004). Another gene involved in auxin transport, PIN1 was found to be down-regulated in Agami. Several of the auxin related genes and their response to salinity are listed in Table 3. Effect of salinity stress on floral transition and development in rice The transition from vegetative to reproductive stage, termed floral transition, is a dramatic phase change in plant development. The transition controlled by a complex genetic network that integrates environmental signals such as day length and temperature (Simpson et al. 1999). We found several rice genes which are associated with floral transition and morphological development to be responsive to salinity stress imposed at the PI stage. Centroradialis (CEN), a gene which plays a role in phase transition and panicle morphology, was induced in both the sensitive genotypes, M103 and IR29 in response to salinity. A zinc-finger protein indeterminant1 (ID1) linked to phase transition signaling in maize, was induced at a higher level (2.7-fold, P < 0.0001) in M103 compared to IR29 (1.5-fold, P < 0.02). In contrast, the flowering locus T (FT) (Os06g35940) gene was differentially regulated in M103 (down-regulated) compared to IR29 (up-regulated) under salinity stress. FT is known to be downstream of gigantea, which represses flowering in rice. A gigantea-like gene (Os01g08700) was down-regulated in both japonica genotypes M103 and Agami. Several other genes involved in phase transition and organ identity and size determination were differentially regulated in M103, IR29 and Agami but not in IR63731. On comparing the expression patterns of IR63731 with IR29, we found several genes including gigantea, FT, terminal flower 1 (TFL1), and embryonic flower 2 (EMF2) to be expressing at a higher level in IR63731 under control conditions. Some of these differentially regulated genes and their expression levels in different genotypes are detailed in Table 3. Discussion In this study, we applied global gene expression profiling analysis to study the transcriptional responses of four rice genotypes comprised of tolerant and sensitive indicas and japonicas. For these genotypes, each with well-characterized salinity tolerance from prior studies, we measured gene expression during PI, the most salt-sensitive stage of development, considered to be the major factor in yield loss in rice. The results from this study, which used a moderate, agronomically relevant salinity stress, demonstrate striking differences in expression responses between tolerant and sensitive genotypes. The genotypes were also characterized for their physiological responses to obtain insights which enabled us to correlate some of the phenotypic observations with expression results. Under salinity stress two key features, one physiological and the other within the transcriptomes, clearly distinguished the tolerant genotypes from the sensitive genotypes. First is the ability of the tolerant lines IR63731 and Agami to exclude Na+ ion from the shoot tissue compared to sensitive lines. The second feature, observed at the gene expression level, is the strikingly small number of genes which responded to salinity stress in the tolerant genotypes. This is likely a manifestation of the low shoot Na+ trait in tolerant lines which experience significantly lower toxicity at the cellular level and therefore requires fewer adjustments of cell metabolism and architecture. Among the genes which were commonly induced by salinity in all genotypes were abiotic stress response related genes such as water stress induced (WSI76), low temperature induced protein (LTI30) and a drought-induced RNase S-like gene. The WSI76 encodes for galactinol synthase (Takahashi et al. 1994) and was responsive to low temperature and osmotic stress. Galactinol synthase is required for the committed step in raffinose family oligonucleotides (RFOs). It was also induced in the tolerant line FL478 when salinity was imposed at early vegetative stage, but not in IR29 (Walia et al. 2005). Besides the genes involved in response to dehydration, we also detected at least two genes involved in nitrate transport, which may be compensating for salinity induced nutritional deficiency. A nitrate transporter, NRT1 was induced in Mesembryanthemum crystallinum in leaf and root tissue under salt stress (Popova et al. 2003). A low-affinity nitrogen transporter and NRT2, both localizing to chromosome 1 of rice were induced in all rice genotypes in response to salinity stress (Table 2). Ion homeostasis in rice under salinity stress Our data indicated that the tolerant genotypes IR63731 and Agami maintained significantly lower Na+ concentration in the shoot tissue compared to sensitive lines under salinity stress. Sodium ion concentration in the plant is considered to be a good indicator of the ion toxicity levels to which the cells are exposed. The lower ion specific stress in tolerant genotypes is reflected in the relatively small number of genes which respond to salinity in tolerant lines when compared to sensitive IR29 and M103. We found several ion homeostasis related genes such as ion transporters which were responsive to salinity stress in the sensitive genotypes. One such gene encoding for Na+ transporter (Os02g07830) was induced by salinity stress in IR29. This gene has high sequence similarity to the ArabidopsisHKT1. The barley ortholog of Os02g07830 was up-regulated under salt stress in the shoot tissue of a barley genotype, Morex during early seedling stage (Walia et al. 2006, submitted). This barley genotype particularly and barley in general, is known to accumulate a high level of Na+ in the shoot tissue during salinity stress. Salt tolerance in barley is believed to be mainly achieved through efficient compartmentalization of toxic ions besides other mechanisms conferring tissue tolerance. High Na+ in the shoot tissue was also documented for rice genotype IR29 at vegetative and reproductive stage under salinity stress (Walia et al. 2005; Table 1). The expression of Os02g07830 did not change significantly in the other three genotypes examined. Based on these pieces of information, we can speculate that Os02g07830 may be involved in redistribution or compartmentalization of Na+ within the plant in genotypes which are relatively inefficient in Na+ exclusion from the plant. Another rice gene recently proposed to confer salt tolerance through recirculation of Na+ ions is SKC1 (Ren et al. 2005). SKC1 was identified as gene underlying a rice QTL for maintaining high shoot K+ in the shoot under salt stress (Lin et al. 2004). The SKC1 gene also encodes a HKT-type transporter and functions as a Na+-selective transporter in rice. Rice locus, Os01g20160 previously annotated as a cation transporter corresponds to SKC1 based on sequence similarity. This locus has been reported to be induced in shoot tissue by salt stress in both the sensitive IR29 and tolerant FL478 genotypes under salinity stress imposed during early vegetative stage (Walia et al. 2005). However, the induction in salt-tolerant FL478 was significantly higher than in IR29 under stress. In contrast to these prior results with vegetative stage stress, we did not detect a significant up-regulation of SKC1 in the shoot tissue of any of the four rice genotypes at PI profiled under salt stress in this study. However, on making genotype–genotype comparisons between Agami-M103 and IR63731-IR29, we found SKC1 to be expressing at a higher level in Agami compared to M103 under both control and saline conditions. Interestingly, the shoot Na+ ion analysis indicated that Agami maintains lower Na+ concentration than M103 under both control and saline conditions (Table 1). The expression of SKC1 was reported to be up-regulated by salt stress in root but not shoot tissue (Ren et al. 2005). In contrast, our results with indica genotypes at vegetative stage indicated an up-regulation of SKC1 by salt stress in the shoot (crown plus growing point tissues), an observational discrepancy that may be attributed to the very different methods of applying salt stress between the two studies. We pursued the expression analysis of SKC1 further by analyzing the shoot and root tissues of tolerant indicas Pokkali, IR63731 and FL478 and sensitive line IR29 (Fig. 4). Our results indicate that SKC1 is salt up-regulated in the shoots of all four indica genotypes under the conditions that we used. Under the more gradual and moderate stress conditions that we applied, SKC1 was up-regulated in the root tissues only in salt-tolerant lines, not in salt-sensitive IR29. Auxin and response of rice to salinity stress Salinity modulates the expression of several auxin-associated genes in the sensitive japonica genotype, M103. The expression level of the gene encoding CYP83B1 was down-regulated in M103 but not in other genotypes. A loss-of-function mutation in CYP83B1 resulted in accumulation of indole 3-actaldoxime (IAOx) which is redirected to auxin biosynthesis (Bak et al. 2001). Taken together, it is plausible that auxin homeostasis in M103 is disturbed in response to salinity by virtue of a reduction of CYP83b1 gene expression. Another gene involved in auxin signal transduction, axi1 was also down-regulated in M103 (Hayashi et al. 1992). Although the mechanisms of auxin signal transduction in plants are still largely obscure, there is considerable evidence linking auxin perception and various ion transporters located in plant cell membranes (Blatt and Thiel, 1993; Barbier-Brygoo, 1995). Anion channels have been indicated to be involved in auxin signaling (Thomine et al. 1997). More specifically, auxin stimulated Cl− efflux/influx has been of interest for several decades (Bentrup et al. 1973; Rubinstein, 1974; Babourina et al. 1998). Bentrup et al. (1973) reported that IAA significantly reduced the uptake of Cl− by cells of Petroselinum sativum, while Babourina et al. (1998) reported a stimulatory effect of auxin on Cl− uptake by oat coleoptiles. The modulation of auxin related genes in M103 during salt stress is attractive in context of the shoot Cl− concentration in M103. The sensitive genotype maintained a significantly lower Cl− concentration under salt stress compared to other genotypes (Table 1). Interestingly, a CLC-d chloride channel (Os12g25200) was induced in M103 by 17-fold (P < 0.0001) in response to salinity stress. Three of the genes induced in IR29 in response to salinity were associated with auxin. The chloride channel (Os12g25200) was induced in IR29 by 7-fold (P < 0.00008). Also induced was a gene for an NPK1 kinase-like protein under salinity stress. Constitutive expression of a tobacco MAPKK NPK1 is reported to dampen auxin responsiveness of a reporter gene (Kovtun et al. 1998) and confer salt tolerance (Kovtun et al. 2000). Additionally, expression profiling of the salt-overly sensitive (sos2-1) mutant in comparison to wild type Arabidopsis revealed differential induction of auxin-related genes in the mutant under salt stress (Kamei et al. 2005). These reports and our data indicate a significant role of auxin in response of some rice genotypes to salinity stress. Salinity stress and floral transition In the current work we focused on shoot meristem tissue under salinity stress imposed during PI. We detected several genes associated with developmental transitions and flowering to be modulated by salinity stress. Salinity delays PI in rice (Zeng et al. 2002), and in our study this delay was generally greater for sensitive than tolerant genotypes. In this context we found the up-regulation of rice genes with sequence match to terminal flower 1 (TFL1)/centroradialis (CEN) in the sensitive genotypes to be interesting. Over expression of TFL1/CEN homologs of rice delayed phase transition and altered panicle morphology (Nakagawa et al. 2002). The delay in initiation of panicle in sensitive genotypes, especially M103 under salinity stress, was longer than tolerant genotypes. The expression of TFL1/CEN was not significantly changed in the tolerant lines under stress. Both TFL1 and CEN belong to the family of phosphatidyl-ethanolamine-binding proteins (PEBPs) (Bradley et al. 1996). Another gene which is involved in flowering transition is FT, which is also a member of the PEBP family. The role of FT is however antagonistic to TFL1, as FT induces floral transition. The expression level of FT was down-regulated in M103 as expected due to a delayed floral transition under salinity stress. Surprisingly, FT was up-regulated in IR29. Up-regulation of FT in IR29 was also observed during the vegetative stage (Walia et al. 2005). The expression of FT was not modulated by salinity stress in tolerant lines IR63731 and Agami during PI and in FL478 during vegetative stage. An up-stream member of the floral induction pathway to which FT belongs is gigantea (GI). Suppressed expression of the rice ortholog of gigantea (OsGI) through RNA-mediated interference (RNAi) resulted in a late flowering phenotype under short day conditions relative to wild-type plants (Hayama et al. 2003). The expression level of gigantea-like gene was significantly reduced in M103 and Agami under salinity stress (Table 3). None of these genes were differentially modulated in tolerant indica IR63731. In brief, the gene expression analysis identified genes which explain the observed delay in PI of M103 under salinity stress. High salt tolerance has often been linked to genotypes which take relatively long to reach maturity. The genotypes selected for this experiment also differ in their time to reach maturity. Therefore it can be argued that the late maturing genotypes IR63731 and Agami are likely to be more salt tolerant merely as a consequence of higher biomass translating into higher yield. However, Zeng et al. (2002) reported a lack of correlation between the duration of reproductive growth period and salt tolerance based on seed yield. This study also included three of the four genotypes used in the current study, namely IR63731, Agami and M103 and employed a comparable level of salinity stress. Since difference in time to maturity for different genotypes can potentially lead to confounding effects on gene expression measured by microarrays, we decided to focus on the developmental stage just after PI was observed for each treatment regardless of the growth conditions (control or stressed) and genotype. This involved staggered harvesting of control and stressed plants for each of the genotype, hence normalizing for delay in PI resulting from salt stress. The effect of difference in development among genotypes used for transcriptome analysis can be avoided by using nearly isogenic lines which differ in salinity tolerance at PI but have similar developmental behavior. However such optimal lines focusing on tolerance at PI are presently not available. Commonalties in the salinity stress response of IR29 at early vegetative versus reproductive stage Sensitive genotype IR29 responded to salinity stress during PI by inducing a large number of genes. Up-regulation of large number of genes was also reported for IR29 under salinity stress during early vegetative stage relative to tolerant genotype FL478 (Walia et al. 2005). We explored the extent of commonalities in salinity induced responses of IR29 by comparing the induced genes during vegetative and reproductive stage. One of the most striking stress responses of IR29 during the vegetative stage experiment was the induction of genes involved in the flavonoid pathway. In the present study, we again observed that several of the flavonoid pathway related genes were induced in IR29 under salt stress, in this case at PI. Among the genes induced include those encoding for phenylalanine ammonia lyase, chalcone synthase, dihydroflavonol 4-reductase, and flavonone 3-hydroxylase. Therefore, up-regulation of the flavonoid pathway as a response to salinity stress appears to be a general characteristic of IR29. Interestingly, some of the genes involved in the flavonoid pathway were down-regulated in the sensitive japonica, M103. So the induction of flavonoid pathway in response to salinity stress is not a universal response of sensitive rice genotypes. We further compared the IR29 data from the vegetative stage study with the current IR29 PI data. In general, we found that the genes responding to salinity stress at vegetative stage were strikingly different from those which responded when the stress was imposed during PI. A total of 524 probe sets were differentially expressed in response to salinity at reproductive stage but not during vegetative stage. We were especially interested in genes which mapped to regions of the rice genome associated with salt tolerance based on QTL studies. One such gene (Os01g24710) encoding for SALT protein was induced during the reproductive stage. This gene has a jacalin-like lectin domain and physically maps to the Saltol region on chromosome 1 (13.88 Mb). Interestingly, another gene (Os01g25280) also encoding a SALT-related protein was induced during the vegetative stage in IR29 (Walia et al. 2005). This gene also maps to the Saltol region (14.26 Mb) on chromosome 1. The salT gene in rice is responsive to salt stress and the phytohormones abscisic acid and jasmonic acid (JA) (Claes et al. 1990; Moons et al. 1997). We found some genes involved in JA biosynthesis, and JA responsive genes, to be induced by salinity stress in IR29 during reproductive stage but not during vegetative stage. These include two lipoxygenases (Os03g52860 and Os12g37260), a gene encoding for 12-oxo-phytodienoic acid reductase (OPR2) (Os06g11240) and a jasmonate induced protein (JIP). The induction of JA biosynthesis and responsive genes during salinity stress is interesting in context of the report that post-treatment of salt-stressed rice with JA dramatically improves the physiological activities otherwise inhibited by stress (Kang et al. 2005). Results from our laboratory have also indicated an ameliorating role of JA pretreatment in salt-stressed barley on shoot Na+ concentrations and photosynthetic performance (Walia et al. 2007). Besides salT, we found several other key transcripts related to salinity stress response as differentially expressed during reproductive stage. A potassium uptake protein (Os09g27580) induced by salinity stress during the reproductive stage in IR29 was one such transcript. This gene localizes to a region on chromosome 9 (16.7 Mb) which is associated with a major QTL for K+ uptake in rice identified by Koyama et al. (2001). This region on chromosome 9 was more recently linked to a QTL controlling root Na+ concentration (Lin et al. 2004). Another gene Os12g39400 (Os.16453.1.S1_at) encoding for a salt-tolerance zinc finger protein (STZ/ZAT10) was up-regulated during reproductive stage under salinity stress. The STZ/ZAT10 is a transcriptional repressor which is believed to regulate the COR/RD gene expression (Lee et al. 2002). The expression of STZ/ZAT10 was not up-regulated in any other genotype under salt stress. Based on this comparison, we can state that very few genes are commonly induced in response to salinity stress in a given genotype when sampled at two different developmental stages. Electronic supplementary materials Below are the links to the electronic supplementary materials. ESM 1 (XLS 64 kb) ESM 2 (XLS 74 kb) ESM 3 (XLS 21 kb) ESM 4 (XLS 21 kb) ESM 5 (XLS 111 kb) ESM 6 (XLS 23 kb) ESM 7 (XLS 21 kb) ESM 8 (XLS 18 kb) ESM 9 (XLS 126 kb) ESM 10 (XLS 32 kb) ESM 11 (XLS 128 kb) ESM 12 (XLS 29 kb) ESM 13 (XLS 16 kb)

Skeletal_Radiol-4-1-2335296

Plantar fasciitis and calcaneal spur formation are associated with abductor digiti minimi atrophy on MRI of the foot

Objective To determine the association of atrophy of the abductor digiti minimi muscle (ADMA), an MRI manifestation of chronic compression of the inferior calcaneal nerve suggesting the clinical diagnosis of Baxter’s neuropathy, with MRI markers of potential etiologies, including calcaneal spur formation, plantar fasciitis, calcaneal edema, Achilles tendinosis and posterior tibial tendon dysfunction (PTTD). Introduction Entrapment of the first branch of the lateral plantar nerve (inferior calcaneal nerve), or Baxter’s neuropathy, produces medial heel pain which may be indistinguishable from heel pain due to plantar fasciitis [1–3]. Baxter’s neuropathy accounts for up to 20% of cases of heel pain [4] but is frequently overlooked as a potential cause of pain, even when pain persists after surgical intervention for other potential causes. On MRI, the presence of atrophy of the abductor digiti minimi muscle (ADMA) reflects chronic compression of the inferior calcaneal nerve and suggests the clinical diagnosis of Baxter’s neuropathy [1, 5]. Treatment for Baxter’s neuropathy includes conservative therapy and, in cases in which non-interventional management is ineffective, surgical release [6]. Entrapment of the inferior calcaneal nerve may result from altered biomechanics, reflected by posterior tibial tendon dysfunction (PTTD) or Achilles tendinosis, or may result from direct mechanical compression of the nerve due to plantar fasciitis and/or plantar calcaneal enthesophytes [1, 7]. Our study was performed to assess the frequency of MRI manifestations of these potential etiologic factors in patients with and without ADMA to test the hypothesis that MRI findings reflecting altered biomechanics or direct compression of the inferior calcaneal nerve occur more frequently in patients with ADMA. Materials and methods Patient population This retrospective study included 200 individuals who were referred for MRI of the hindfoot due to the presence of foot pain. These were individuals seen in a subspecialty teleradiology group practice with participant referrals from 40 states. The study was conducted with data from August 2006 to January 2007 and included patients of all ages. The study group was composed of 100 consecutive individuals with abductor digiti minimi atrophy and 100 consecutive individuals without abductor digiti minimi atrophy. No eligible participants were excluded. The subspecialty teleradiology group practice does not have an Institutional Review Board (IRB), and the study was conducted in compliance with the Declaration of Helsinki principles and the Health Insurance Portability and Accountability Act. Owing to the retrospective nature of the study, informed consent was not obtained. Participants’ ages ranged from 10–92 years [mean = 49.0 years, standard deviation (SD) = 16.9 years] and included 139 women and 61 men. MRI technique Patients were scanned in magnets ranging from 0.2 T to 1.5 T. Long- and short-axis fat and fluid sensitive sequences were obtained without the administration of gadolinium. The technical parameters for all magnet strengths were as follows: sagittal T1-weighted, sagittal short-tau inversion recovery (STIR), oblique axial STIR, coronal T1-weighted, coronal STIR, plantar flexed axial proton density, axial T2-weighted, and axial T2-weighted with fat saturation or STIR. Data acquisition In our retrospective analysis, potential patients were identified through review of reports of consecutive MRI studies obtained over a 6-month period. One hundred patients with ADMA and 100 patients without ADMA were selected for inclusion through review of reports, with subsequent confirmation of the presence or absence of ADMA through review of images. Patients were identified by review of MRI reports, with subsequent review of MR images. We reviewed 1,780 MRI reports in order to identify 100 patients with ADMA. Two radiologists (U.C., and A.L.), each with 4.5 years of experience of MRI interpretation and who were blinded to the initial interpretation, by consensus interpreted the MR images of both groups. Each MR image was evaluated for presence and absence of plantar fasciitis, calcaneal spur formation, calcaneal edema, PTTD, and Achilles tendon dysfunction. Atrophy was defined by the presence of any amount of fatty replacement of muscle mass on T1-weighted images (Figs. 1 and 2). Plantar fasciitis was defined by thickening, partial or full thickness tear of the plantar fascia. Calcaneal spur formation was defined by the presence of a bony outgrowth at the medial calcaneal tuberosity. Calcaneal edema was defined by increased signal intensity within the calcaneus adjacent to the insertion of the plantar fascia on fluid-sensitive sequences. PTTD was defined by tendon thickening, tenosynovitis or partial or full thickness tear. Achilles tendon dysfunction was defined by tendon thickening, peritendonitis or partial or full thickness tear. Fig. 1Coronal T1-weighted image demonstrating normal abductor digiti minimi (ADM) muscleFig. 2Coronal T1-weighted image demonstrating fatty replacement of the abductor digiti minimi muscle (ADM), indicating atrophy Statistical analysis Age and gender distribution, as well as the frequency of plantar fasciitis, calcaneal spur formation, calcaneal edema, PTTD, and Achilles tendon dysfunction, were compared in patients with and without ADMA. Age was compared with analysis of variance (ANOVA). For the categorical variables we used, as appropriate, Pearson chi-square analyses or Fisher’s exact test to compare the proportion of those with atrophy to those without. We used logistic regression to determine odds ratios and 95% confidence intervals for each of the variables to atrophy. First, we included each variable in a univariate model to determine its relationship to atrophy. We then simultaneously included all the variables in a multivariate analysis to determine the relationship of the variable to atrophy, while simultaneously adjusting for all the other variables. Alpha level was at 0.05, and all tests were two-sided. SPSS version 15.0 (2006) (SPSS, Chicago, USA) was used for the analyses. Results Overall, of 200 patients in the study (100 with ADMA and 100 without ADMA), 25 (12.5%) had Achilles tendinosis, 18 (9%) had calcaneal edema (n = 18), 55 (27.5%) had calcaneal spurs, 41 had other tendon abnormalities, 63 (31.5%) had plantar fasciitis, and 43 (21.5%) had PTTD. Fourteen (7%) patients had isolated ADMA in the absence of any other identified pathological condition. Comparison of the frequency of the above findings in patients with and without ADMA is shown in Table 1. Those with ADMA were significantly older than those without atrophy (P < 0.001). There were no gender differences. Table 1Comparisons of demographic and clinical variables for those with and without atrophy of the abductor digiti minimi muscleVariableNo atrophy (n =100) % (no.)Atrophy (n=100) % (no.)PAge (years)a<0.001[Mean (SD)]40.8 (15.8)57.2 (13.7)Gender0.28Women66.0% (66)73.0% (73)Men34.0% (34)27.0% (27)Achilles tendinosis<0.001No97.0% (97)78.0% (78)Yes3.0% (3)22.0% (22)Calcaneal edema0.005No97.0% (97)85.0% (85)Yes3.0% (3)15.0% (15)Calcaneal spur<0.001No93.0% (93)52.0% (52)Yes7.0% (7)48.0% (48)Other tendon abnormality0.60No81.0% (81)78.0% (78)Yes19.0% (19)22.0% (22)Plantar fasciitisb<0.001No89.0% (89)47.5% (47)Yes11.0% (11)52.5% (52)PTTDNo89.0% (89)68.0% (68)Yes11.0% (11)32.0% (32)<0.001aMean and SDs compared with ANOVA, since continuous variable. All other analyses with Pearson’s chi-square, except for Achilles tendinosis and calcaneal edema, which were with Fisher’s exact test due to small subgroupsbData missing for one individual with atrophy Patients with atrophy compared to those without atrophy had significantly greater proportions of Achilles tendinosis, calcaneal edema, calcaneal spur, plantar fasciitis, and PTT dysfunction. Table 2 shows logistic regression analyses for the variables. In univariate analyses, increased age, presence of Achilles tendinosis, calcaneal edema, calcaneal spur, plantar fasciitis, and PTT dysfunction were all significantly associated with increased odds ratios for atrophy. In the multivariate analysis, only increased age, presence of calcaneal spur, and presence of plantar fasciitis were significantly associated with atrophy. Presence of Achilles tendinosis only had a P value of 0.09, and calcaneal edema and PTT dysfunction had non-significant P values. The multivariate analysis was also repeated, using data only from the individuals older than 21 years, in case inclusion of data from patients younger than 21 years (n = 15) had affected the results; the odds ratios, P values and significance levels were almost exactly the same. Table 2Logistic regression analyses for atrophy of the abductor digiti minimi muscle (OR odds ratio, CI confidence interval)VariableUnivariate OR (95% CI)PMultivariate OR (95% CI) PAge (years)1.08 (1.05, 1.11)< 0.0011.06 (1.03, 1.09)< 0.001Gender0.280.51 Women1.001.00 Men0.72 (0.39, 1.32)0.75 (0.33, 1.73)Achilles tendinosis< 0.0010.09 No1.001.00 Yes9.12 (2.63, 31.60)3.69 (0.81, 16.88)Calcaneal edema0.0070.47 No1.001.00 Yes5.71 (1.60, 20.39)1.89 (0.35, 10.19)Calcaneal spur< 0.0010.02 No1.001.00 Yes12.26 (5.18, 29.10)3.60 (1.28, 10.17)Other tendon abnormality0.600.71 No1.001.00 Yes1.20 (0.60, 2.39)1.19 (0.48, 2.97)Plantar fasciitis< 0.0010.01 No1.001.00 Yes8.95 (4.27, 18.77)3.35 (1.31, 8.56)PTTD0.0010.31 No1.001.00 Yes3.81 (1.79, 8.10)1.72 (0.60, 4.88) Of the 200 patients in the study, the consensus reviewers (U.C. and A.L) disagreed with findings in the initial report of eight , three with ADMA and five without ADMA. In two of the cases of ADMA, the presence of a calcaneal spur was not mentioned in the initial report. In the third case of ADMA, the presence of posterior tibial tendinosis was not mentioned in the initial report. For three of the five patients without ADMA, the presence of a calcaneal spur was not mentioned in the initial report. For one of the patients without ADMA, the presence of Achilles tendinosis was not mentioned in the initial report. For one patient without ADMA, the presence of plantar fasciitis was not mentioned. Discussion Patients with ADMA had a significantly greater frequency of Achilles tendinosis, calcaneal edema, calcaneal spur, plantar fasciitis, and posterior tibialis tendon dysfunction than those without ADMA. After logistic regression analysis, only increased age, calcaneal spur, and plantar fasciitis remained significantly associated with ADMA. The inferior calcaneal nerve supplies motor branches to the flexor digitorum brevis, quadratus plantae and abductor digiti minimi muscles and sensory fibers to the long plantar ligament and calcaneal periosteum [1]. The inferior calcaneal nerve originates from the lateral plantar nerve near the trifurcation of the posterior tibial nerve at the level of the medial malleolus and courses between the abductor hallucis and quadratus plantae muscles [8] (Figs. 3 and 4). When it reaches the inferior border of the abductor hallucis muscle, it turns laterally, passing anterior to the medial calcaneal tuberosity (or spur, if present), between the quadratus and underlying flexor digitorum brevis muscles until it reaches the abductor digiti minimi muscle [7, 9–11] (Fig. 3). The orthopedics and podiatrics literature suggests two possible sites of entrapment of the inferior calcaneal nerve which might result in Baxter’s neuropathy: first, in patients with altered biomechanics, such as excessive pronation (flatfoot deformity), the nerve may be compressed as it turns laterally between the quadratus plantae and abductor hallucis muscles [12], and second, the nerve may be compressed as it passes anterior to the medial calcaneal tuberosity, or, if present, plantar calcaneal spur [7, 8, 13]. Our study evaluated potential markers of inferior calcaneal nerve compression due to either etiology. PTTD and Achilles tendinopathy were assessed as markers of altered biomechanics which might predispose the individual to inferior calcaneal nerve compression. Calcaneal spur formation and plantar fasciitis were assessed as potential causes of inferior calcaneal nerve compression as the nerve passes anterior to the medial calcaneal tuberosity. The persistence of an association between calcaneal spur formation and plantar fasciitis after logistic regression analysis supports the importance of these factors in the etiology of Baxter’s neuropathy. The presence of 14 patients with isolated ADMA without other abnormalities suggests that Baxter’s neuropathy may also occur in the absence of mechanical entrapment secondary to altered biomechanics. Fig. 3An anatomic drawing of the sole of the foot after removal of the layers containing the skin, plantar fascia, and part of the flexor digitorum brevis muscle and abductor hallucis muscle cut away. Posterior tibial nerve trifurcation. PTN posterior tibial nerve, ICN inferior calcaneal nerve, LPN lateral plantar nerve, MPN medial plantar nerveFig. 4An anatomic drawing of a coronal section through the right talocrural and talocalcaneal joints (ADM abductor digiti minimi muscle, FDB flexor digitorum brevis muscle, AH abductor hallucis muscle) Baxter’s neuropathy can produce chronic heel pain. Clinically, the symptoms of Baxter’s neuropathy may be indistinguishable from those of plantar fasciitis, and the two conditions are often superimposed [5, 7, 14, 15]. Initial treatment for either condition is the same (taping to prevent abnormal motion, Achilles/plantar fascia stretching exercises, and medications such as steroids or non-steroidal anti-inflammatory agents [7]. Subsequent treatment includes injections and physical therapy, with surgical intervention reserved for patients who fail to respond to conservative measures. For patients in whom heel pain persists despite conservative therapy, MRI evaluation is indicated prior to any surgical intervention. When a plantar fasciotomy is performed for clinically presumed plantar fasciitis in patients with occult Baxter’s neuropathy, symptoms often persist, because at least a portion of the underlying cause of pain has not been addressed [14]. When pre-operative MRI demonstrates abductor digiti minimi atrophy, the surgeon is afforded the opportunity to advise the patient of the underlying abnormality and its prognosis, and, when indicated, to alter the surgical approach to include surgical release of the inferior calcaneal nerve. The radiology literature rarely mentions ADMA as an MRI manifestation of Baxter’s neuropathy. One study identified ten subjects with MRI evidence of ADMA out of 476 subjects who underwent MRI evaluation of the foot or ankle and found that ADMA was associated with obesity and the presence of a heel spur [5]. Another study evaluated patterns of muscle atrophy on MRI of the foot and found that, of 29 patients with focal denervation atrophy involving the muscles of the foot, 19 had ADMA. Of patients with denervation in the lateral plantar nerve distribution, 18 had atrophy involving the ADM and one had atrophy of the adductor hallucis muscle [16]. The largest study to date was a prospective evaluation of the prevalence of ADMA on MRI, which found ADMA in 6.3% of 602 consecutive patients who underwent MRI of the foot and ankle [17]. The prevalence of ADMA in that study is corroborated by the findings in our study, in which it was necessary for us to review 1,780 MRI studies in order to identify 100 patients with ADMA, corresponding to a prevalence of 5.6%. The study also found that the most common associated findings in patients with ADMA were plantar enthesophytes and plantar fasciitis, but it did not include a statistical analysis of the prevalence of associated findings in patients with and without ADMA. Our study includes the largest number of patients with ADMA on MRI and is the first study to perform logistic regression analysis to determine independent variables associated with ADMA. Limitations of the study include retrospective data collection, which may lead to selection bias; however, we attempted to minimize selection bias by including 100 consecutive patients in each study group. In addition, the inclusion of 15 pediatric patients might have affected the results, as younger patients are less likely to have degenerative conditions of the foot or ankle and could potentially have falsely decreased the frequency of findings in the group of patients without abductor digiti minimi atrophy. However, a repeat statistical analysis, from which the pediatric patients had been excluded, demonstrated no significant difference in results. In summary, our study showed a significant association between atrophy of the abductor digit minimi muscle, an MRI manifestation of Baxter’s neuropathy, with age, plantar calcaneal spur formation, and plantar fasciitis. These findings support the notion of an etiologic role for compression of the inferior calcaneal nerve as it passes anterior to the medial calcaneal tuberosity in the development of Baxter’s neuropathy.

Eur_Radiol-2-2-1705470

New coil concept for endoluminal MR imaging

Our aim was to conduct a prospective study to evaluate staging accuracy of a new coil concept for endoluminal magnetic resonance imaging (MRI) on ex vivo gastric carcinomas. Twenty-eight consecutive patients referred to surgery with a clinically proven primary gastric malignancy were included. Surgical specimens were examined with a foldable and self-expanding loop coil (8-cm diameter) at 1.5 Tesla immediately after total gastrectomy. T1- and T2-weighted and opposed-phase sequences (axial, frontal sections; 3- to 4-mm slice thickness) were acquired. Investigators blinded to any patient information analyzed signal intensity of normal gastric wall, gastric tumor, and lymph nodes. Findings were compared with histopathological staging. On surgical specimens, 2–5 gastric wall layers could be visualized. All gastric tumors (26 carcinomas, two lymphomas) were identified on endoluminal MR data (100%). Overall accuracy for T staging was 75% (18/24); sensitivity to detect serosal involvement was 80% and specificity 89%. N staging correlated in 58% (14/24) with histopathology (N+ versus N−). The endoluminal coil concept is feasible and applicable for an ex vivo setting. Endoluminal MR data provided sufficient detail for gastric wall layer differentiation, and therefore, identification of T stages in gastric carcinoma is possible. Further investigations in in vivo settings should explore the potential of our coil concept for endoluminal MR imaging. Introduction Despite the fact that the incidence of gastric cancer is declining in most Western countries, it remains the second leading cause of cancer mortality worldwide [1, 2]. Simultaneously, the incidence and prevalence of cancer arising from the gastric cardia has been increasing since the 1970s [3]. Adenocarcinoma of the esophagogastric junction can be classified according to Siewert et al. based on morphology and anatomical location [4]. Adequate surgical resection (R0) is the only potentially curative therapy for eligible patients with gastric carcinoma [4]. In the West, gastric cancer is often diagnosed at an advanced stage of disease not eligible for surgery. Fewer than 50% of patients undergo R0 resection [5, 6]. Recently, new concepts of multimodal treatment strategies for locally advanced gastric carcinoma have been investigated [7, 8]. Local treatment by endoscopic mucosal resection for early gastric carcinomas is currently being evaluated [9, 10]. Several phase II and III clinical trials for neoadjuvant chemotherapy in gastric carcinoma showed its feasibility and safety [11, 12]. Its purpose is to eliminate or delay systemic metastasis and reduce micrometastatic spread of disease. Another benefit is potential reduction of tumor volume in initially unresectable advanced tumor stages (downstaging), therefore increasing resectability rate. An adequate treatment strategy, especially in regard to the concept of neoadjuvant chemotherapy, requires precise clinical staging to depict relevant prognostic factors and identify resectable tumor stages. This also accounts for endoscopically treated early gastric cancers as well as for adenocarcinomas of the gastroesophageal (GE) junction since each type of tumor requires a different surgical approach. Thus, the exact knowledge of tumor morphology and tumor invasion into gastric wall is a crucial information for clinical staging. Up to now, computed tomography (CT) and magnetic resonance imaging (MRI) have not provided the spatial resolution to fulfill this requirement with sufficient accuracy [1, 13–16]. Results for staging accuracy of endoscopic ultrasound (EUS) also seem to be improvable [17–19]. The main aspect of gastric wall differentiation is spatial resolution. By using endoluminal radiofrequency (RF) coils for MRI, image quality and spatial resolution can be enhanced. MRI provides superior soft tissue contrast, which makes it useful for tumor invasion detection; furthermore, information not obtainable by other imaging modalities is acquired. Various authors reported the use of endoluminal RF coils as a diagnostic and staging tool in gastrointestinal diseases and rectal and prostate cancer [20–27]. A major problem with endoluminal RF coils is placement close to the region of interest and depth of visualization. Our approach uses a foldable and afterward self-expanding loop coil design (8-cm diameter), which enhances spatial resolution and depth of visualization The aim of our study was to assess normal gastric wall architecture and signal intensity (SI) on endoluminal MRI as well as signal intensity and appearance of gastric carcinomas and their related lymph nodes. Subsequently, we used endoluminal MRI as staging modality on ex vivo gastric carcinomas. Findings were compared with histopathological staging. Materials and methods Patients A prospective study was conducted on patients with gastric malignancies referred for total gastrectomy to the surgery department of our university hospital. They were included in the study if the lesion was identified by EUS or CT suggesting a gastric carcinoma or if histological workup of biopsies confirmed the diagnosis. The study protocol was approved by the institutional review board, and informed consent was acquired from each patient prior to surgery. We investigated 28 consecutive patients, 11 women and 17 men (age range 46–87, median 67 years). Tumors were located at the cardiac region in 11 cases, the fundal area in two, the corpus in six, and the antropyloric area in nine. Immediately after surgery, gastrectomy specimens were taken to the MR suite to conduct the examination within a time frame of 2–3 h (Fig. 1). Specimens where then sent to histopathology unchanged from the procedure. Fig. 1Flowchart of study design Endoluminal RF coil The endoluminal RF coil (patent issued 10/2005, German Patent and Trade Mark Office, No: 10127850.0–35, Grenacher et al.), developed in collaboration with Fraunhofer Institute for Biomedical Engineering (IBMT, St. Ingbert, Germany) consists of a foldable and self-expanding receiver loop (8-cm diameter). It is coated with a biocompatible material (silicone) to prevent direct contact of the wire with stomach tissue [29]. The flexible characteristics of the shape memory metal (nitinol) used allow passage through the instrument channel (13-mm diameter) of a specially designed MR-compatible endoscope. The MR-compatible endoscope was not used in this study and is described elsewhere, but it is mentioned in this context because RF coil design is reflected by the specifications of the endoscope. A nonferromagnetic tuning box connects the RF coil via a standard interface with a 1.5-Tesla MRI scanner (Symphony, Siemens, Erlangen, Germany), which was used in this study. Gastrectomy specimens were placed in a container with 3–4 l of 0.9% sodium chloride solution to ensure good contrast with gastric tissue and adequate expansion of gastric folds. The loop of the RF coil was placed on the mucosa side of the tumor. After the tumor was located with a scout sequence, images where acquired in axial and frontal sections of 3- to 4-mm thickness. Scanning was performed with T1- and T2-weighted and opposed-phase sequences (Table 1). Table 1Imaging sequence parametersSequenceRepetition time (TR)Echo time (TE)AcquisitionsSlice thickness (mm)MatrixField of view (FOV) (mm)Pixel size (mm)T1-weighted TSE5511424512x5122000.39x0.39T2-weighted TSE3,52070231024x10242200.21x0.21T2-weighted GRE1,0503513512x5122200.43x0.43T1-weighted GRE opposed phase187743256x2652200.86x0.86TSE turbo spin echo, GRE gradient recalled echo Image evaluation Image data were read on certified diagnostic work stations independently by two board-certified radiologists with 10 and 12 years of experience, respectively, and one board-certified surgeon with 8 years of experience in endoscopic ultrasound on gastric carcinomas. Investigators were blinded to any clinical information as well as patient identification. After separate reading, a discussion was held about the quality of data and unique features of the case. Additionally, final agreement was achieved if disagreement on tumor stage had initially occurred. With respect to the unique imaging technique, the resolution and detail of image data a consensus staging was reasonable. Radiological and histopathological classification of the tumor was done according to the Tumor Node Metastisis (TNM) staging system of the Union Internationale Contre le Cancer (UICC) [30]. Determination of tumor stage was done as follows: T1: tumor invasion of the lamina propria of mucosa or invasion of submucosa, wall thickening, and signal intensity changes confined to these layers.T2: tumor invasion extended to the muscularis propria, and additional thickening through gastric layers with signal intensity changes showing either a homogeneous or inhomogeneous lesion without serosal abnormalities. If invasion extended beyond muscularis propria into an adjacent fat tissue plane without serosal infiltration, it was considered as T2.T3: tumor invasion of mucosa, submucosa, and muscularis propria, with infiltration of the serosa or changes in signal intensity presenting with micronodular strands as growth into extraserosal fat tissue.T4: tumor invasion into adjacent organs or structures clearly presenting as contiguous tumor extension or a mass with similar signal intensity as the gastric tumor. N staging was done counting the lymph nodes detectable on MRI, regarding them as pathologic due to their signal intensity. Since gastrectomy samples varied in lymph node content due to surgical technique and extension of the gastrectomy not comparable to an in vivo situation, N factor was graded into N+ for positive findings of lymph nodes and into N− for absence of lymph nodes. In each case, normal gastric wall features were assessed. Signal intensities of the mucosa, submucosa, muscularis propria, and serosa or subserosa were recorded, as well as the amount of layers in which the normal gastric wall could be differentiated into. A score system from 1 to 5 (1=poor; 5=very good) was used for several study items to weigh the power of their findings. Each sequence, for example, was rated for image quality on a range from 1 to 5, representing insufficient to very good. Signal intensity and morphology of the gastric tumors and demarcation to normal gastric wall were used to describe characteristics of tumor presentation. The lesion was identified by irregular architecture of gastric wall, abnormal thickening, and change in signal intensity of gastric wall layers, which usually present with sharp and clear demarcation into three to five different layers with distinct signal intensities and thickness. Tumor size was not measured, as it does not influence T staging or provide further information. Histopathology was considered the gold standard, and radiological data where compared with its results. All data are presented as absolute numbers and relatively as percentages. Concordance of histopathological results and radiological findings are reported as overall accuracy. Sensitivity, specificity, and accuracy for T factor, N+/− factor are provided. Detection rate of serous membrane invasion or the differentiation between T2 and T3 stages was assessed, as it defines advanced stages of disease. Ninety-five percent confidence intervals (CI) are given if statistically applicable. Results The setup of our study proved to be feasible, as shown elsewhere [29]. Two cases were not included because of insufficient image quality. Histopathology found 26 carcinomas (15 adenocarcinomas, nine signet cell carcinomas, two mixed type carcinomas) and two gastric lymphomas. Four tumors were classified as pT1, 15 as pT2, 3 as pT3, and 2 as pT4. Gastric lymphomas were excluded from the study although they where detected as such in both cases (100%). Gastric tumor staging In 24 (100%) cases, the tumor was identified. Initially 46% (11/24) of T stages were staged correctly; 1/4 of pT1 tumors, 7/15 of pT2 tumors, 2/3 of pT3 tumors, and 1/2 of pT4 tumors were classified correctly (Table 2). Interpretation of image data was done without knowledge of tumor location in respect to the gastric region, such as cardia, fundus, corpus, or antrum, as it would be possible on an in vivo setting such as endoscopy or CT. In a second reading session, the interpreters were informed about the gastric region of the primary tumor for each case. In knowledge of gastric region, six, first as T3 classified carcinomas of the cardia, were correctly staged as pT2 and one, first as T3 classified tumor (cardia), was correctly identified as pT4 (Table 3). This gives an overall accuracy of 75% (18/24)± –17.32 (CI) (Table 4). The mean score for T-factor staging was 3.54 (range 2–5). Sensitivity, specificity, and accuracy for detecting serosal involvement was 80%±16.00, 89%±12.27, and 88%±13.23, with a mean score of 4.5 (range 3–5; Table 5). Overstaging occurred in 4/24, three pT1 tumors were classified as T2 or T3 stages, and one pT2 tumor was identified as T3 stage (Table 6). In this case, the tumor replaced all gastric wall layers and invaded perigastric fat tissue but did not infiltrate the serous membrane. Insufficient separation of gastric wall layers and the presence of abnormal signal intensities most likely desmoplastic reactions or peritumor inflammation were identified retrospectively as reasons for overstaging of pT1 tumors. Two tumors (pT2, pT3) were understaged as T1 and T2 stage. One case was a diffuse type pT2 adenocarcinoma with disseminated infiltration into the muscularis propria, which is not visualized and thus presented as T1 stage. Table 2Comparison of histologic with endoluminal findings without knowledge of tumor region Endoluminal magnetic resonance imaging (MRI) findingsHistologic findingsT1T2T3T4Total (n=24)T1121–4T2177 –15T3–12 –3T4––11 2Table 3Comparison of histologic findings with endoluminal findings acquired with knowledge of tumor region Endoluminal magnetic resonance imaging (MRI) findingsHistologic findingsT1T2T3T4Total (n=24)T1121 —4T21131—15T3–12 –3T4–––22Table 4Sensitivity, specificity, and accuracy for each T stage; overstaging and understaging rates T1T2T3T4OverallSensitivity1/413/15 2/3 2/2 –Specificity19/20 6/9 19/21 22/22 –Accuracy20/24 19/24 21/ 24 24/24 18/24 Overstaging3/4 1/15 ––4/24 Understaging–1/15 1/3 –2/24 Table 5Results for serosal invasion detectionEvaluation of serosal invasionSensitivity4/5Specificity17/19Accuracy21/24False positive2Table 6Signal intensities of gastric wall layers on T1-weighted, T2-weighted, and opposed-phase imaging from inside to outside T1 imaging (n=21)aT2 imaging (n=27)aOpposed phase imaging (n=3)aTwo layers (n=7)bHigh––IntermediateThree layers (n=33)bHighIntermediateHighLowHighLowIntermediateIntermediateHighFour layers (n=10)bHighIntermediate–LowLowIntermediateIntermediateLowLowfive layers (n=1)*–Intermediate–HighIntermediateLowIntermediateaNumber of sequences analyzedbNumber of sequences with n layers Nodal involvement Lymph node detection resulted in N+ for detected lymph nodes in 17 cases with N− in 7 cases. Sensitivity for detecting N+ was 71%±18.22, specificity 29%±18.07, and accuracy 58%±19.72. This reflects the limitation of the study design. Ex vivo gastrectomy specimens do not provide the same quantity of lymph nodes as pathological workup. The mean score rating N factor staging was 4.4 (range 2–5). Image analysis Separation of gastric wall layers ranged from a minimum of two visible layers to five distinct visible layers (mean 3.10) in 51 analyzed sequences: 21 T1-weighted and 27 T2-weighted sequences and three opposed-phase sequences (Table 6). T2-weighted sequences depicted an average of 3.37 gastric wall layers; T1-weighted sequences 2.76. Mean score for image quality of the sequences used was 3.88 (range 2–5), with 4.11 for T2- and 3.62 for T1-weighted sequences. Image analysis revealed that in four out of six incorrectly staged T factors, tumors were described as inhomogeneous or diffuse concerning demarcation to normal gastric wall. This may be one reason for misinterpretation of T factor. Signal intensity of the mucosa for T1-weighted sequences was mostly classified as hyperintense; for T2-weighted, it was intermediate. Submucosa, muscularis propria, and serosa or subserosa presented with hypointense, intermediate, hypointense SI for T1-weighted; hyperintense, intermediate, hypointense for T2-weighted sequences (Table 6). Lymph node and tumor appearance correlated in 86%±9.9 (44/51 sequences), with intermediate SI for T1- and T2-weighted sequences. Out of eight sequences without correlation in lymph node and tumor SI, two sequences showed a homogeneous tumor in T1-weighted and an inhomogeneous lesion in T2-weighted images (Fig. 3). In three opposed-phase sequences, lymph nodes (hypointense) showed different signal intensity than the primary tumor (intermediate). T2-weighted sequences (Table 4) visualized more gastric wall layers and provided the interpreter with a better demarcation of tumor volume from normal gastric wall than did T1. T1-weighted sequences showed a better delineation and contrast of lymph nodes in adjacent structures (Fig. 2). Fig. 2a, bTwo consecutive T2-weighted images (a, b) of a T1 early gastric carcinoma (white arrowheads), well differentiated (intestinal type of Lauren classification), located at the subcardial region. Lymph nodes in adjacent fat tissue with a high signal intensity (white arrows) are visualized. Morphology of normal gastric wall is pointed out by open arrows. (*) marks the position of the receiver coilFig. 3a, bT2 signet cell tumor of the cardia region (diffuse type of Lauren classification, white arrowhead) on T1-weighted (a) and T2-weighted imaging (b). The extent of tumor mass and the diffuse infiltration into gastric wall is better visualized on T2-weighted images due to the mucinous character of the tumor (black arrowheads). The tumor appears more homogeneous on T1-weigthed images (a). (*) marks the position of the receiver coil Discussion Endoluminal MRI has been used by various groups to asses its ability for locoregional imaging of esophageal and gastrointestinal disease. The first group [20, 31] to introduce an MR endoscope reported promising results for the diagnosis of gastrointestinal diseases and pointed out limitations of their design [20, 31]. Image quality of endoluminal MRI was not as clear as those acquired by EUS, and their success rate of obtaining clear images of the stomach was low (58%) [20]. Although their study achieved 89% for T-staging accuracy, its patient collective consisted mainly of advanced gastric carcinomas (T3, T4). Other authors reported the use of an MR endoscope with a similar design [21, 28, 32], sharing the incorporation of the RF coil with 10×30 and 10×47 mm into the tip of the MR endoscope. This is a useful design for imaging at the esophagus, where both groups showed good results in staging esophageal cancer [20, 28]. The small cylindrical RF coil design implemented in the tip of the MR endoscope may be a reason for problems in obtaining sufficient images in the stomach. Our approach using an insertable and self-expandable RF coil (8-cm diameter), independent from the MR endoscope, seems to be more suitable for imaging of the stomach by providing enhanced depth of visualization. In 93%, sufficient image data was obtained. Detection of resectable lesions can only be accomplished by exact knowledge of tumor invasion in relation to gastric wall layers. New multimodal treatment strategies require the information of tumor stage, as it defines enrollment into neoadjuvant therapy approach. Clinical staging based on imaging studies can provide important information for adequate therapy decision if they fulfill this requirement. Gastric wall differentiation and identification is essential for accurate staging of gastric malignancies. Our approach can depict between two to five gastric wall layers and therefore deliver sufficient detail as basis for tumor invasion detection. Signal intensities of gastric layers assessed in our study correspond mainly to those reported previously by other authors [20, 33, 34]. Visualization of the serosa layer remains a problem in all published studies with MRI or endoluminal MRI on gastric wall. Some authors investigated ex vivo gastrectomy specimen with a 1.0-Tesla MRI device and were able to differentiate up to five different gastric layers, but visualization of the serosa or subserosa was not achieved [35]. One reported study used a 2.4-Tesla MRI system and failed, as did another study with a 4.7-Tesla MRI system, to visualize serosa or subserosa as a distinct layer [34, 36]. One work group reported possible identification of serosa or subserosa on gastrectomy specimens after fixation with formalin [33]. With the use of a 1.5-Tesla MRI device and a 4-cm-diameter loop coil, muscularis propria and serosa or subserosa showed isointense signal intensities but were outlined by subserosal fat tissue with contrasting signal intensity. The results were confirmed by another work group using a similar setup [37]. Identification of the serosa is necessary to securely differentiate between T2- and T3-stage gastric carcinoma. Although a detection rate of 88% for serosal involvement was achieved in this study, serosa as a single gastric wall layer was not clearly visualized. In seven cases with four to five visible gastric layers, a thin outer layer on T2-weighted sequences was detected and could possibly be identified as serosa or subserosa with subserosal fat tissue. This finding was inconsistent and varied in different T-stage carcinomas. This may be because of isointense signal intensity of muscularis mucosae and serosa or subserosa or the fact that the thickness of the serosa layer is not sufficient for secure identification [38]. It is not certain whether tumor invasion into or close to the serosa or subserosa layer changes its signal intensity presentation. Indirect detection by subserosal fat tissue seems possible, but in our experience, there is some variance in the thickness of subserosal fat tissue in gastric carcinoma specimens. One reason for this could be tumor cachexia. An interesting learning point is reflected by the results of T staging. Knowledge of the gastric region where the tumor is located is important since T staging of the stomach depends on the gastric region. Tumors of the cardia and part of the fundus region where a serosal layer is missing are considered T2 whereas they would be T3 in other gastric regions. The study setup was intended to be as close as possible to a future in vivo setting with an MR endoscope, but tumor location and gastric region cannot be read on image data from gastrectomy specimens. An in vivo setting would enable investigators to acquire this information. After providing the tumors' gastric region, T staging accuracy improved to 75%. Thus, the gastric region of tumor location is an important fact for adequate staging of tumor invasion. While this study used an ex vivo model of gastric carcinomas and the amount of patients included could be higher, the entire range of T factors for gastric carcinomas is covered. Another limitation of this study is the missing congruence of lymph nodes present in the gastrectomy specimens due to surgical technique to those available for histopathology. This is reflected by N-factor staging results. An in vivo setting should be more accurate but poses new problems, such as motion artefacts due to patient or endoscope movement. Modern rapid MRI sequences and spasmolytic agents may be able to overcome these obstacles in an in vivo setting, but this is a goal for future studies. It is important to point out pitfalls of this imaging technique, as overstaging is a problem in many imaging modalities. Peritumor inflammation, micronodular affection of the serosal membrane, and diffuse-type gastric carcinoma were retrospectively identified as reasons for misinterpretation of tumor invasion. Overstaging occurred in 17%, with three T1 tumors not correctly classified. This is encouraging since all T1 tumors were detected as tumors but in two cases were classified as T2 tumors and one as T3. Only one incorrectly classified T2 tumor contributed to overstaging. This indicates that overstaging was not a problem of differentiation between T2 and T3 stages of disease. It is necessary to focus on T1 carcinomas, as experience with T1 carcinomas in imaging modalities in Western countries is low due to the fact that most gastric cancers are diagnosed at an advanced stage. It is important to learn more about signal intensity and morphologic presentation of early gastric carcinomas on endoluminal MRI to improve detection rate. Endoluminal MR imaging provides enough detail to visualize a T1 tumor (Fig. 2). Endoscopic mucosal resection as one treatment option for early gastric carcinoma requires an imaging method that is able to visualize and identify these stages securely and provide information that is observer independent. Further studies should include a larger quantity of early stages of gastric carcinomas to analyze accuracy potential of endoluminal MRI. MR images provide information not obtainable by other imaging modalities. A gastric tumor can present with different appearances on T1-weighted imaging than on a T2-weighted imaging (Fig. 3). Depth of visualization of a method is essential for staging a gastric tumor, assessing its depth of invasion into gastric wall and into adjacent organs and structures, as well as depicting involved lymph nodes. Our image data can provide detailed information about gastric wall invasion as well as organ invasion (Figs. 2 and 4). Fig. 4a, bT4 gastric tumor (intestinal type of Lauren classification, open arrows) on T2-weighted images of the same plane. Transition of normal gastric wall into tumor mass (white arrows). Pancreas (white arrowheads) adjacent to the gastric wall demonstrates depth of visualization (a). Tumor invasion into pancreas (black arrowheads) (b) In this study, we have shown that our concept is functional and feasible for MRI of the stomach, and staging results comparable with other imaging modalities were achieved. Endoluminal MRI in combination with conventional MRI could assess local and regional imaging in one session, giving accurate information about local staging and metastatic spread of disease. Secondary contrast media should be evaluated for its use in improving gastric wall layer discrimination and identification as well as tumor demarcation. Additional studies are needed to assess the clinical feasibility of this coil concept in vivo for imaging of the upper gastrointestinal tract. High-resolution imaging of the pancreas could be another possible application. Today, most tumors of the stomach and esophagus are staged by conventional methods, such as endoscopy, EUS, or CT. Future prospects of endoscopic MRI may be limited to certain tumor entities, such as early or advanced carcinomas, but it could be able to answer additional specific clinical questions for relevant therapy decisions. Summary Endoluminal MRI with the coil concept is feasible and applicable. T staging of ex vivo gastric carcinomas is possible. Results for T-factor staging are preliminary but promising, as they reflect the first experience with this technique. Image quality and resolution obtained with the endoluminal RF coil are convincing for future use in gastrointestinal imaging. Limitations of this study include missing congruence in quantity of lymph nodes for N staging, a limited amount of some T-stage gastric carcinomas, and the ex vivo setting. Taking the limitations into account, the coil design should be investigated in an in vivo setting after assessing safety aspects to evaluate the prospects given by the ex vivo results. The clinical potential may be limited by providing a staging tool for certain tumor entities. With more experience in endoluminal MRI and interpretation of endoluminal image data, its use could be extended toward other fields in gastrointestinal imaging.

Clin_Oral_Investig-4-1-2238791

How valid are current diagnostic criteria for dental erosion?

In principle, there is agreement about the clinical diagnostic criteria for dental erosion, basically defined as cupping and grooving of the occlusal/incisal surfaces, shallow defects on smooth surfaces located coronal from the enamel–cementum junction with an intact cervical enamel rim and restorations rising above the adjacent tooth surface. This lesion characteristic was established from clinical experience and from observations in a small group of subjects with known exposure to acids rather than from systematic research. Their prevalence is higher in risk groups for dental erosion compared to subjects not particularly exposed to acids, but analytical epidemiological studies on random or cluster samples often fail to find a relation between occurrence or severity of lesions and any aetiological factor. Besides other aspects, this finding might be due to lack of validity with respect to diagnostic criteria. In particular, cupping and grooving might be an effect of abrasion as well as of erosion and their value for the specific diagnosis of erosion must be doubted. Knowledge about the validity of current diagnostic criteria of different forms of tooth wear is incomplete, therefore further research is needed. The process of diagnosis and current criteria for dental erosion From the chemical view, the aetiology of dental erosion can be defined as the chronic exposure of the teeth to extrinsic or intrinsic acids under the condition that the oral fluids are undersaturated with respect to tooth mineral [23, 26]. Under in vitro conditions without physical impact, teeth demineralise centripetally (Fig. 1), a feature of substance loss which is normally not observed in the mouth. In fact, the multitudes of physical and chemical assaults occurring during a lifetime result in a more or less characteristic pattern of tooth wear. The classification of wear is therefore made from morphological features which are frequently seen clinically. The tooth morphology as apparent after eruption is the idealised status, deviations of which, if not caries or trauma, are diagnosed as (erosive) tooth wear. Various forms of wear including dental erosion are listed in the International Classification of Diseases [52] thus defining them as a disease (for critical discussion of this notion see [12]). Fig. 1Effect of the continuous exposure of a human third molar to 10% citric acid. The amorphous, centripetal tissue loss is obvious (a unaffected tooth, b tissue loss after 4, c 8, and d 12 h immersion time) Looking at what “diagnosis” is, one will find mostly definitions like “identification of disease from signs or symptoms”, implicating that the physician or scientist “reads off” from the patient thus detecting the disease. Health and disease, however, is not a given condition, but is constituted by the theoretical concepts and the discursive practice the physician or scientist is subjected to [12]. That is to say, rather than reading off the disease from the patient, a pattern of criteria is projected onto the diagnosed subject which determines the diagnostic procedure and outcome. Following this approach, the diagnostic process, in a first step, is a theoretical concept ordering signs and symptoms to diseases which takes place within the configurative power of the discourses. In whatever discourse the physician acts, the act of clinical diagnosis, in a second step, is classification. The conclusion reached through this process is called “a diagnosis”. From this background, it becomes obvious that the diagnostic process must be accompanied by a recurrent reconsideration of its theoretical concepts and the reevaluation of the criteria used. The latter is what the paper aims to achieve. The diagnosis of tooth wear in general and erosion in particular is made from its lesion characteristics, from the results of nutritional, medical and occupational analysis, and from dietary records. The diagnostic process can be more differentiated with the individual patient, whereas in field trials it is restricted to the classification of lesion shape. Throughout the literature, there is more or less consensus about the aetiology of the various forms of tooth wear, the clinical criteria for dental erosion and its differential diagnosis [12, 14]. The early signs of erosive tooth wear appear as changes of the optical properties of enamel resulting in a smooth silky–shining glazed surface. When the tissue loss continues, changes in the original morphology occur. On smooth surfaces, convex areas flatten or concavities develop, the width of which clearly exceeds the depth. Lesions are located coronal from the enamel–cementum junction (CEJ) with an intact enamel rim along the gingival margin. On occlusal and incisal surfaces, rounding and cupping of the cusps and grooving of the incisal edges occur, and restorations may rise above the level of the adjacent tooth surfaces. In advanced cases, the whole occlusal morphology disappears. The validity of these clinical criteria for erosion, however, has not been under critical consideration. Validity, in general terms, means the degree, to which a measurement measures what it purports to measure. For the issue addressed here, criterion validity is the relevant term, meaning the extent to which the measurement correlates with an external criterion under study. A further aspect of validity is concurrent validity. Concurrent validity means to which extent the measurement and the criterion refer to the same time [24]. In the case of erosion this would address the question if (a) the diagnostic criteria reflect lesions being an effect of an exposure to acids and (b) if the presence of characteristic signs is concurrent with an acid exposure. The implementation of current diagnostic criteria Considering how the diagnostic criteria for erosion were established, the literature reveals that it was individual clinical experience and case reports rather than systematic research. One of the first publications on the characteristics of acid induced tissue loss was in 1946 from Robinson [43] and in 1947 from Stafne and Lovestedt [47]. Robinson, whilst being in doubt about the aetiology of dental erosion, described the lesions as located on smooth surfaces and as wedge shaped, cup shaped, disk shaped, irregular in form, L- or U-shaped, or simply as small round depressions or larger surface lesions. It is not clear how these descriptions were established, but Robinson referred to a work from McClure and Ruzicka [31] describing the morphology of lesions of rats teeth after being fed with lactate and citrate drinking fluid. Stafne and Lovestedt presented their observations in subjects with known acid exposure, amongst them 50 patients with frequent consumption of lemon juice. They did not give a concrete description of lesion shape, but attributed hypersensitivity, absence of stain and defects with rounded margins as effect of the action of the acids. Their most important sign of diagnostic value was the presence of fillings projecting above the surface of the tooth. In their publication, a number of clinical images were included, presenting lesions clearly matching current erosion criteria. More than twenty years later, in 1970, it was Pindborg [40] who gave the often cited definition of erosion as being superficial loss of dental hard tissue by a chemical process which does not involve bacteria. He described the clinical signs of chemically-induced tissue loss as usually located to the gingival third of the facial surfaces, possibly also located at proximal surfaces, lesions to appear shallow, disc-shaped, smooth, polished, or scooped out. In contrast to abrasion, he attributed erosive lesions to be located evenly on the left and right side. It appears noteworthy, that Pindborg, as well as Robinson, ascribed cupping of the cusps, loss of the occlusal morphology or loss of crown height, and incisal grooving to attrition which in their publications was defined as result of mastication. It is probably Eccles and Jenkins [8, 10] who were the first to give a detailed and systematic description of the lesion characteristics and also suggested a system for classification. The basis of the development of the clinical criteria was a sample of 72 patients seen in the dental hospital over a period of 9 years [8]. All cases were thoroughly documented with respect to their medical and dietary history, that is, it was a group of subjects with known exposure to intrinsic or dietary acids. The findings derived from this group of subjects (Table 1) have been retained nearly unchanged until now. The main aspects are in general terms loss of surface contour, shallow concavities on smooth surfaces, cupping and grooving on occlusal/incisal surfaces and restorations rising above the level of the adjacent tooth surface. Table 1Diagnostic criteria for dental erosion as outlined by Eccles and Jenkins [10] and Eccles [8]Diagnostic criteria for dental erosionInitialAbsence of developmental ridges of the enamel, smooth glazed surfaceAdvanced Facial/oral surfacesConcavities whose breadth greatly exceeds their depthLesion ovoid or crescentic in outline, concave in cross section orLesion entirely in the crown, irregular in outline, punched out appearance Occlusal/incisal surfacesSurfaces appear flattened, depression of the cusps (cupping) and on the incisal edges (grooving), edges of restorations raising above the level of the adjacent tooth surface Conclusions from epidemiological studies using current diagnostic criteria Current criteria, derived from a relatively small sample, have been applied in case reports [8, 9, 16, 32, 39, 50] and studies with risk groups [15, 17, 21, 25, 34, 38, 44, 48, 51]. Indeed, when used in subjects with known or strongly assumed exposure to acids, most studies revealed a higher prevalence of lesions in the exposed groups compared to the control groups (Table 2). Table 2Prevalence of lesions in risk groups deriving from the use of current diagnostic criteria for dental erosion IndexGroup sizePrevalence risk groupPrevalence control groupIntrinsic acid exposureMeurman et al. [34] reflux diseaseEccles and Jenkins indexn = 11728/117 = 24%No control groupRytömaa et al. [44] eating disordersEccles and Jenkins indexn = 14022/35 = 63%12/105 = 11%Öhrn et al. [38] eating disordersLussi indexn = 13379/81 = 98%Minor, less severeIncisal/occlusalIncisal/occlusalGrade 1: 93%Grade 1: 73%Grade 2: 52%Grade 2: 23%BuccalBuccalGrade 1: 30%Grade 1: 19%Grade 2: 9%Grade 2: 6%PalatalPalatalGrade 1: 21%Grade 1: 10%Grade 2: 5%Grade 2: 0%Extrinsic acid exposureLinkosalo and Markkanen [25] vegetariansEccles and Jenkins indexn = 5216/26 = 60%0/26 = 0%Wiktorsson et al. [51] wine tastersEccles and Jenkins indexn = 1914/19 = 74%No control groupGanss et al. [15] raw food dietLussi indexn = 206127/130 = 98%66/76 = 87% These observations support the finding that subjects with continuous exposure to acids have a higher rate of lesions with a specific characterisation. This is, however, not enough support for the assumption that, vice versa, subjects presenting with such defects must be exposed to acids. Analytical epidemiological studies on random or cluster samples attempted to relate the occurrence of lesions with any of the known aetiological factors for erosion, but only few studies of this type have been published. Most of them include children or adolescents, but there is lack of studies on older population groups. Furthermore, off the few studies, some used the Tooth Wear Index (TWI [46]) which is not designed to assess dental erosion specifically. The overall findings (Table 3) are controversial since some authors found no or only a partial relation between aetiological factors and the occurrence or severity of lesions [2, 3, 5, 7, 20, 30, 35, 37, 49], whereas others revealed strong relationships [22]. In addition, a relation to the intake of yoghurt or other foodstuff which certainly has no erosive potential was mentioned [1, 27, 35]. Table 3Analytical epidemiological studies attempting to relate the occurrence of (erosive) wear to aetiological factors Index, group size, age and prevalenceConclusionJärvinen et al. [22]Eccles & Jenkins indexCitrus fruits: odds ratio (OR) 2Case-control, n = 100 eachSoft drinks: OR 413–83-year-oldsLussi et al. [27]Lussi indexSignificant relation to the consumption of fruit, acidic drinks, yoghurt, vomitingn = 41726–30- and 46–50-year-olds at least 36 and 43% resp. with any erosionBartlett et al. [5]TWI (Smith and Knight)No significant relation to drinks or other acidic foodn = 210Significant relation to heart burn11–14-year-olds57% had wear in enamel on more than 10 teethJaeggi et al. [20]Lussi indexNo relation to any aetiological factorn = 41719–25-year-olds at least 82% with erosionAl-Dlaigan et al. [1]TWI (Smith and Knight)Significant relation to drinks and fruit, but also to milk, yoghurt and beern = 41814-year-olds48% low, 51% moderate 1% severe lesionsAl-Majed at al. [2]TWI (Smith and Knight) modified for erosionNo association to erosive drinks for the total samplen = 862Significant association to frequency of drinks at night and duration of drinks retained in the mouth only in advanced cases (n = 95)12–14-year-olds95% with erosionMathew et al. [30]Lussi indexNo relation to the intake of sport drinksn = 30418–28-year-olds37% with erosionVan Rijkom et al. [49]Modified Lussi indexNo relation to acidic drinks and fruitsn = 40015–16-year-olds30% with visible smooth wearArnadottir et al. [3]Modified Lussi indexNo significant association to risk factorsn = 27815-year-olds72% grade 124% grade 25% grade 3Nunn et al. [37]TWI (Smith and Knight) modified for erosionNo significant association with dietary factorsn = 1726Significant relationship with gastro-oesophageal symptoms4–18-year-olds36, 56 and 34% with any erosion on buccal and palatal surfaces of the incisors, and first permanent molars resp.Dugmore and Rock [7]TWI (Smith and Knight) modified for erosionDrinking fizzy pop: odds ratio 1.59–2.52 depending on amount and frequencyn = 114912-year-oldsNo relation to eating apples, citrus fruit56% with erosionMilosevic et al. [35]TWI (Smith and Knight) on labial and lingual surfaces in front teeth, occlusal surfaces of first molarsNo association to apples, fresh orangesn = 2385Weak association (OR 1–1.4) to yoghurt, grapefruit, salad dressing, vinegar, fruit juice, fizzy drinks14-year-oldsStrong association to herbal/lemon tea (OR 3.97)53.5% with exposed dentine There are many points of discussion for explanation. At first, (erosive) wear is the effect of various concurrent or past chemical exposures from different sources, the variety of which can hardly be covered by simple questionnaires, and especially in younger people the dietary intake and lifestyle might be open to variations. Furthermore, questionnaires may reflect only in part the dietary habits due to interpretation by the responder or seasonal changes in daily life habits. When the young age of the groups studied is considered, the determinants for dental erosion may not have acted for long enough, and, in addition, there might be considerable differences in the individual susceptibility to erosive demineralisation [15]. It must, however, also be taken into consideration that current diagnostic criteria might not be valid enough to really reflect the effect of a chronic acid exposure. Taking into account that tooth wear is the result of various factors, the occlusal and incisal surfaces of teeth are not only exposed to acids but are particularly prone to physical impacts from mastication. From this background, the finding that many studies using erosion indices reveal that occlusal lesions are the most prevalent and that it is the first lower molar which is most affected [19], is of particular importance. Occlusal surfaces might be the only location exhibiting a relation between lesion severity and prevalence, and age [28] thus indicating a significant contribution of abrasion from mastication. Furthermore, particularly the teeth from ancient remains exhibit defects being strikingly of the same shape as those attributed to erosion today (Fig. 2). Fig. 2a Occlusal aspect of a subject living on a raw food diet with multiple acid impacts, and a medieval subject b with an assumed abrasive diet (images a, b, and c samples from [13, 15]). Occlusal/incisal defects in a subject with chronic vomiting d and in a medieval subject c. The shape of lesions from predominantly erosive and predominantly abrasive aetiology is strikingly similar Comparative studies on lesion characteristics of wear Attempts have been made to characterise the shape of wear resulting from abrasion and erosion [6, 13]. A comparison of subjects with known exposure to a non-acidic but coarse diet and subjects with an acidic, but refined diet using silicon impressions revealed that in the former, the scooped out dentine was significantly shallower than in the latter. Furthermore, the deepest lesion region in teeth from subjects with abrasive diet was located at the functional cusps, whereas in teeth from subjects with an erosive diet, the deepest region of the defects was located more centrally. The authors concluded that in cases when the dentine is severely scooped out, the causative agent is most likely of erosive origin whereas scooping tends to be shallower when wear is caused by abrasion. The conclusion of the study was to suggest a quantitative diagnostic procedure by calculating the depth/breadth ratio for clinical differential diagnosis [6]. Another approach was to compare the shape of defects occurring in subjects with substantially different nutrition patterns [13]. The study included three groups of age matched individuals — randomly selected contemporary subjects, medieval remains and a group of subjects living on raw food diet. From the latter group, extensive information was available on their intake of acid food qualifying their nutrition as rather erosive. For the medieval group, an abrasive nutrition was assumed from knowledge about the general nature of the early medieval diet. As to the smooth surfaces, the most interesting finding was that in the medieval group, no lesions were observed (Table 4). Even in cases with severe wear, the lingual and buccal aspects of the teeth appeared undisturbed and, in many cases, even developmental ridges were present. The lack of cervical defects is also reported by Aubry et al. [4] and Kaidonis (in this issue) and also inherently supported by the fact that the overwhelming number of anthropological studies on tooth wear only deal with occlusal and proximal wear and that the indices used there do not include criteria for cervical wear [11, 36]. There are, however, few studies reporting wear on lingual surfaces in archaeological remains [18, 41] and there is one publication discussing regurgitation as a possible aetiological factor [42]. In this study, 151 adult pre-Conquest British skulls were examined using the TWI. Indeed, buccal and lingual wear was observed to the same degree as in a contemporary comparison group, but cervical wear did not develop beyond criterion 0 in most individuals. In 20% of the skulls, however, a cervical wear to TWI score of 2–3 occurred. No information was given about the shape of the lesions, and from the images included, it appears that their characteristics would not match the current criteria for dental erosion. Table 4Prevalence of lesions of defined shape in three groups (n = 100 each) of subjects with substantially different nutrition patterns [13] Abrasive diet (medieval group)Acidic diet (raw food group)Average western diet Incisal/occlusal surfacesIncisors/caninesGrooving93%96%90%n.s. Molars/premolarsShallow cupping (<0.5 mm)87%59%47%p < 0.001Deep cupping (>0.5 mm)78%45%4%p < 0.0001 Smooth surfaces (all teeth)Concavity coronal to the CEJ0%63%8%p < 0.0001V-shaped defects0%38%10%p < 0.0001 In contrast to the findings in the medieval group, shallow defects located coronal from the CEJ were considerably prevalent in the raw food group (Table 4, Fig. 3b), whereas in the western diet group, the prevalence of this kind of lesions was much lower and corresponds to its general prevalence in Germany [45]. Fig. 3a Buccal aspect of teeth 44–47 with significant loss of crown height, but without any lesion in a medieval remain [13] with severe generalised occlusal wear c. b Occlusal defects in a subject living on a raw food diet with a high intake of acidic food [15]. The shape of the occlusal lesions is similar to c, but combined with shallow lesions with intact cervical rim lesions. d Same subject with Fig. 4b with an initial buccal lesion As to the occlusal/incisal surfaces, grooving and cupping was common in all groups even though most often in the medieval group, followed by the raw food group, and the western diet group. The conclusion from this study was that shallow defects on smooth surfaces might be a valid criterion for dental erosion, whereas cupping, and especially incisal grooving, was common in all groups and therefore not valid for a differential diagnosis. The similarity of occlusal/incisal defects in all groups might be explained from a tribological view [29]. Wear, as a result of abrasion, occurs as three body wear that means the intervention of an abrasive slurry or bolus. During mastication, the abrasive tends to hollow out softer surface regions. Cupping or grooving can occur when deeper enamel regions with lower microhardness [33] are exposed or when the dentine is reached. In the case of erosion, tissue loss is on one hand caused by direct dissolution of mineral but also due to an increased susceptibility of acid softened surfaces to physical wear. In these cases, also a non- or less abrasive bolus could cause similar defects when acting on acid weakened surfaces. Hence, the occlusal/incisal substance loss observed in individuals prone to dietary acids may be explained as pronounced abrasion/demastication of acid softened surfaces. Therefore, it is questionable if the occlusal morphological criteria used for the diagnosis of occlusal erosion per se are valid. Even though there are often striking similarities between lesions of predominantly abrasive or predominantly erosive origin, there is, on the other hand, often also considerable variation in lesion shape in cases of erosion. The morphology of occlusal lesions in subjects with known exposure to acids varies from deep hollowing out of the cusps to totally amorphous loss of the occlusal structure even in subjects with a similar dietary history (Fig. 4). Fig. 4Occlusal tissue loss from erosive aetiology can also be of strikingly different shape either presenting as deeply hollowed out lesions (a subject with raw food diet [15], b subject with excessive consumption of orange juice) or as amorphous generalised tissue loss affecting the entire surface (c, d subjects with excessive consumption of erosive drinks). An interesting feature is seen in an adolescent with a history of severe anterior open bite with only the molars being in function e Substance loss occurred from excessive consumption of a cola type drink. In the premolars, dentine is proud of the surface. f Hollowing out the entire occlusal surface with enamel remnants in the centre, aetiology is the excessive consumption of sport drinks Conclusion The validity of current diagnostic criteria for dental erosion has not been systematically studied, even though there is consensus about their definition. The following conclusions can be drawn: Grooving of incisal surfaces is a common phenomenon and possibly the effect of any physical or chemical impact. It should be considered to abandon grooving of anterior teeth and canines as a clinical criterion for dental erosion.Shallow defects located coronal from the CEJ may predominantly occur as effect of chronic acid exposure and might be pathogonomic for dental erosion. This assumption is supported by the finding that these types of lesions are not present in ancient remains even in cases of severe wear.Cupping of cusps is the most uncertain criterion because it can be an effect of abrasion as well as of erosion. In industrialised countries, abrasion is not expected to be a significant factor in young people. Cupping occurring at younger ages can therefore be an effect of erosion. At older ages, however, physical and chemical impacts add up increasingly and cupping will therefore be of little diagnostic value in adults. These conclusions are drawn from very few studies; therefore systematic research on this issue is needed. Nevertheless, there is enough support for a criticism of current diagnostic criteria particularly in the light of the development of a new index.

J_Comput_Aided_Mol_Des-4-1-2270923

What do we know and when do we know it?

Two essential aspects of virtual screening are considered: experimental design and performance metrics. In the design of any retrospective virtual screen, choices have to be made as to the purpose of the exercise. Is the goal to compare methods? Is the interest in a particular type of target or all targets? Are we simulating a ‘real-world’ setting, or teasing out distinguishing features of a method? What are the confidence limits for the results? What should be reported in a publication? In particular, what criteria should be used to decide between different performance metrics? Comparing the field of molecular modeling to other endeavors, such as medical statistics, criminology, or computer hardware evaluation indicates some clear directions. Taken together these suggest the modeling field has a long way to go to provide effective assessment of its approaches, either to itself or to a broader audience, but that there are no technical reasons why progress cannot be made. Introduction Virtual screening in the pharmaceutical industry is an essential part of molecular modeling’s contribution to lead discovery and, to a lesser extent, lead optimization. This has led to considerable research into what method or approach works best, typically by means of ‘retrospective’ evaluations, i.e. attempting to predict future, i.e. prospective, behavior by appraising techniques on known systems. Despite this there is no agreed upon theory as to how to conduct a retrospective evaluation. As a consequence, it is very difficult for an outsider to assess if methods are getting better, have stayed the same, or even worsened over time. In a practical enterprise, such as drug discovery, the proposed benefits of virtual screening, i.e. avoiding the cost and time of a real screen, have to be weighed against one simple question: does it actually work? Without proper metrics of success, i.e. ones that go beyond the anecdotal, molecular modeling is not guaranteed a vibrant future. Observed as a general exercise, there are four elements that ought to be standard for any prediction study, whether of a virtual screen, or any general pattern recognition method. The first is whether the study is well designed. The second is what metrics are used to evaluate the outcome. The third is a consideration of significance, i.e. error analysis. And the fourth is a reliable assessment of whether the results are particular or general. All four aspects are important and yet it is rare for a study in any field to meet all these criteria. Even in the most critical part of drug discovery, i.e. clinical trials, there is considerable room for improvement, as several recent retrospective studies of the medical literature have demonstrated [1, 2]. In reports on virtual screening, in fact in molecular modeling in general, it is rare to find an adequate consideration of any of these issues. Why is this? Why is the modeling field so poor at the most basic elements of evaluation? A charitable view would be that, as with communication skills, most modelers receive little appropriate formal training. Certainly there is no central resource, whether scholastic review, book or paper. A slightly less charitable view is that journals have not developed standards for publication and as such there is little Darwinian pressure to improve what the community sees as acceptable. It is to be hoped that this is a learning curve, i.e. that editors will eventually appreciate what is required in a study. An extreme view is that we are poor at evaluations because we simply do not matter very much. If large fortunes were won or lost on the results from computational techniques there would be immense debate as to how to analyze and compare methods, on what we know and exactly when we know it. There would be double blind, prospective and rigorously reviewed studies of a scale and depth unknown in our field but common in, for instance, clinical trials. In short, there would be standards. In the hope that virtual screening is, in fact, worthwhile we provide comment, suggestions and research on two important aspects, namely experimental design and performance metrics. Although the two are intimately linked, i.e. an experiment should be designed with a mind to what is being measured, there are distinguishable aspects. On experimental design, extensive properties, such as number of targets, actives and inactives, need to reflect a statistical understanding of the current unreliability, or high variance, of methods [3–5]. So dominant is this variance that it almost renders moot any discussion of other matters, such as decoy design. However, ultimately all aspects of design are important. On decoy selection we suggest the necessity of clarifying design intent and suggest four broad categorizations. In analyzing results, the issue of correlation is considered. This often arises in the context of the 2D similarity of actives from congeneric series, but the general issue also concerns decoys, targets and methods. Research is proposed that would clarify essential and poorly understood issues, such as the transference of predictability between closely related systems. On evaluation metrics we examine the AUC (Area Under the Curve) of ROC (Receiver Operator Characteristic) curves [6–9]. Consideration of why the AUC is a popular measure in many disciplines suggests standards by which virtual screening metrics ought to be judged. Finally, by evaluating average properties of large numbers of systems, and by considering simple cost/benefit examples, we bring into question the validity and utility of metrics proposed to capture ‘early’ behavior. Experimental design In what follows we consider the importance of both intensive and extensive properties of an experiment. An intensive property is something intrinsic to a design, whereas extensive properties change with the size of the system. For example, the type of decoys used in a retrospective study is an intensive property; the number of such is an extensive property. We believe the most overlooked intensive characteristic is the design goal, i.e. what is trying to be proved. This typically falls into a few discrete classes and appropriate labeling would help combine lessons from different studies. For extensive quantities we consider how common statistical approaches can aid the choice of numbers of actives, decoys and targets. Finally, actives, decoys, targets or methods are not always independent and this has to be quantified even in as simple a matter as comparing two programs. Techniques for accounting for correlation within an experimental design are known but rarely applied. Intensive properties One of the most basic issues in designing a retrospective screen is how to chose decoys. Typically there are a certain number of active compounds and one wishes to see if a method can distinguish these from a second set, presumed inactive. This is the most basic of classification problems. Is X of type A or type B? The legal system often has the same dilemma, e.g. was X at the scene of a crime or not? A police line-up has all the components of a virtual screen. Usually the number of actives (suspects) is small, usually one. The number of decoys (called ‘fillers’) has to be sufficient that random selection does not compete with real recognition; a minimum of four is usual. But it cannot be so large that guilt is hidden within the statistical variance of the innocent. The fillers need to be convincing, i.e. not outlandishly dissimilar to the guilty party, but not too similar or even potentially also at the scene (i.e. false false positives). As courtroom verdicts can depend on the appropriateness of a line-up, standard procedures are well known. We make the argument for four types of virtual screening experiments; each with its own intent. Each of the four designs ultimately consists of a random selection of decoys but after the application of different filters. Universal. Any compound available to be physically screened, typically either from vendors or corporate collections.Drug-like. Available and drug-like, typically by applying simple chemical filters.Mimetics. Available, drug-like and matched to known ligands by simple physical properties.Modeled. Available, drug-like and derived using 3D modeling on known ligands or the intended targets.Although no classification scheme could be perfect, fair comparison of studies requires an alignment of intent. In general, decoys get ‘harder’ from A to D, although this is not necessarily true on a case-by-case basis and is itself an interesting area of research. The first, and perhaps least in favor, is the universal selection of decoys. A catalogue of compounds from a vendor or set of vendors is treated as a general population from which to draw. An example of a virtual study with universal decoys can be found in Rognan et al. [10]. Although this method is now uncommon, it has an interesting intent. Faced with all compounds available for testing, does a method distinguish known actives without using prior knowledge of what makes a compound active? Drug discovery has a long and successful history of grinding up exotic plants and animals and screening for activity and so this is a reasonable, if old-fashioned, approach. In the Rognan set, for example, we find I3, not likely to be a drug but none-the-less an interesting molecule. The problem with universal decoys is two fold. First, is it random enough? The space of all possible chemistry is exceedingly vast [11, 12]. The concept that a few thousand compounds, in particular from a vendor database, could act as a thorough sampling is implausible. In fact, there is now evidence suggesting known chemistry is very restrictive [12]. Because of ‘inductive bias’, a concept frequently highlighted by Jain [13, 14], we tend to make what we know might work, instead of sampling of what can be made. Second, is it possible a universal decoy such as I3 might stand out pretty much the same way a shady character would stand out against a selection of school children, shop clerks and nuns? Paul Hawkins has described this as the ‘dog’ test [15], i.e. if your dog could tell the difference between the actives and inactives what have you really proved? Actually, potentially a lot but only if the rest of the experiment is designed with this choice of decoys in mind. The problem here is one of dynamic range. If it is too easy to distinguish an active then the only way to distinguish between the methods is to test many, many times, i.e. to improve the statistical power necessary to rank one method above another. As is well known, and discussed below, the error in any metric depends on both the number of actives and the number of inactives. While it is trivial to increase the number of (presumed) inactives almost without bound, the number of actives is normally very finite. Only in some of the more impressive published studies does the number of actives exceed a hundred [3, 4] and it is this limitation that really hinders random decoys being an effective experimental design. We note that this is only a presumed inadequacy of universal decoys; in fact such decoys may prove difficult for some computation methods, the Hawkins dog test not with-standing. The point is that a presumed limitation can be overcome by applying basic error analysis. A more typical selection procedure is to choose from a decoy set that is ‘drug-like’. Of course, there is no rigorous definition of ‘drug-like’, but this does not stop it being widely used. The intent is to mimic modern physical screens and not test everything but instead be guided by current dogma as to what a drug might look like. The most prevalent of these descriptions is the famous Lipinski Rule-of-Five, but there are many variants [16]. This is not dissimilar to how police line-ups are actually constructed; ‘fillers’ are normally acquired from local jails. In theory, this should also be a harder test of methods because there are less easily discernable inactives, although this is not proven. Examples of this approach are the studies of McGaughey et al. and Warren et al. One potential advantage of this approach is that because decoys are derived from characterized collections they are more likely to be known to be inactives. This is typically only an assumption for universal decoy sets. It is debatable as to how big a problem false decoys are, but clearly they do not help. There are also issues with drug-like decoys. Some companies’ collections are heavily biased towards certain targets that may or may not be related to the retrospective study at hand. The study by McGaughey et al. reported significant differences in the efficiency of decoys chosen from the MMDR, a kind of ‘consensus’ drug-like collection, compared to ones from their internal Merck database. This trade of generality for local applicability is a characteristic of many aspects of evaluations. For instance, should targets be chosen to represent all possible systems, a subset of pharmaceutical interest or a class within that subset? What is gained in local applicability is often lost to generality and prospective predictability. The third approach is to find mimetic or modeled decoys. These are meant to stress-test methods and should be used to compare approaches, rather than necessarily evaluate real-world performance. Mimetic decoys are constructed so that ‘simple’ methods cannot tell known ligands from decoys. The rationale is utilitarian; why should one chose to use sophisticated methods when simple, ligand-based, ones can do just as well? Approaches include matching physical properties, for instance size, number of hydrogen bond donors and acceptors, lipophilicity, charge or flexibility [17]. An example of this approach is the DUD dataset [18] of Irwin et al. Here, for each target thirty-six decoys are found for each active by matching physical properties, forming a mimetic set referred to as DUD-self. The combined set, i.e. across all targets is drug-like and is referred to as DUD-all. Mimetic decoys can sometimes be depressingly effective, as illustrated by Irwin et al. However, in not all cases were DUD-self decoys harder to distinguish than decoys from DUD-all. This at least suggests physical property mimetics are not guaranteed to provide a reality check for methods claiming to capture the physics of drug-target interaction. Modeled decoys go one step further than mimetics by eschewing the concept of comparison to simple, practical, methods and instead designing directly against the method under study. As an example, suppose decoys for a docking study were chosen such that every decoy had good shape complementarity with some part of the active site, i.e. it fit well. It is widely known that basic shape complementarity is usually necessary for binding and forms a major component of most scoring functions. Such a set of decoys would them make for a challenging test of scoring functions. But is this a good test of docking? If essentially random performance is seen an observer might decide docking is without merit, whereas an appropriate conclusion would be that complex scoring functions are useless. The inevitable desire of methods to be seen to be useful often prevents modeled decoys being chosen, even though they potentially address the most interesting scientific questions. Irwin et al. had intended their mimetic decoys to act more as modeled decoys, i.e. the aim was to make things harder for docking programs, but, as mentioned, this was not always achieved. Although mimetic and modeled decoy selections have virtues, they can also hamper comparison between different studies. In the case of the mimetic approach, the definition of ‘simple’ evolves over time, especially as property calculations improve or change. To arrive at a set of modeled decoys the procedure applied must be scrupulously described, e.g. how is the protein prepared, how is the ligand protonated etc, and complete and accurate descriptions of published virtual screening procedures are rare. However, there seems no reason a consensus could not be reached by interested parties. Standard protocols could be developed, shared and used to verify results. The problems are more of will than ingenuity. Given the above discussion, what is the appropriate decoy set to use? A universal set with sufficient actives to enable discrimination between methods? A drug-like set built from one group’s definition of a corporate collection, but perhaps without general applicability? A mimetic set to produce physically similar decoys? Or a modeled set defined so as to tease out specific differences between methods, even at perception of poor performance? A suggestion by Geoff Skillman [19] provides a novel framework. Given the speed of modern computers and the cost of storage and transmittal of information, there seems no reason a retrospective study could not contain all decoy types, with careful labeling of individual intent. The authors can make of their data what they will, for instance by reporting performance against a subset of decoys. However, if a broader set is included in the supplementary material, others can make use of the data for potentially different purposes. One of the proposals of this paper is for modeling to move beyond the anecdotal towards the systematic. Full reporting of data is essential but a further step would be to include alternate data so that others can construct purposes beyond the original intent. Extensive properties In addition to intensive properties, there are the extensive properties such as how many actives, decoys and targets are used. Once again the important consideration is knowing what we want to know. If the purpose is to evaluate a single method on a single target the necessary extensive properties are quite different than for a broad study on the efficacy of several methods on many targets. We illustrate this with some basic error analysis. The Central Limit Theory (CLT) states that the average of M measurements tends towards the true mean with an error proportional to √(V/M), where V is the average squared difference of a quantity from its estimated mean. Thus, the error is an intrinsic quantity, the square root of V, divided by an extrinsic quantity, the square root of the number of measurements. Famously, we have to take four times the number of measurements to reduce the error by a factor of two. What does this say as to the number of actives, decoys or targets that should be used to accurately measure the performance of a method? If the performance is similar no matter what actives, decoys or targets are used then the variance is small and M can be small. However, this is not the situation for modeling techniques applied to real systems. Instead, it is the ruling zeitgeist that ‘performance may vary’ [3–5]. Just how variable are virtual screening methods? Figure 1a and b illustrate the extent of the problem by presenting a reanalysis of the Warren et al. study from GSK, with eight different docking methods and our own work on the DUD dataset (DUD-self decoys) using four different virtual screening techniques. The performance metric is the AUC averaged over each dataset. The number of targets for Warren et al. is eight and for DUD forty, i.e. a five fold difference. As expected, the error bars, i.e. the confidence limits at 95%, are slightly more than twice as big for a method assessed against the GSK set than against DUD. In addition it is clear that although methods in the GSK test could be broadly classified as better or worse, this is subject to considerable statistical dispute. The resolving power of DUD begins to be apparent in Fig. 1b, where one can begin to put some significance to the generally held belief that ligand-based methods perform better than docking and about as equivalently as 2D methods [4, 21]. The average AUC and error bars for GOLD from the Warren study are included in for comparison purposes only. A more quantitative analysis of this data will be presented below in the section on correlation between methods. Fig. 1(a) Average AUC values across docking programs in the Warren study, with 95% confidence intervals. Where programs were run in multiple modes the best average AUC was retained. (b) Average AUC values plus error bars across the DUD dataset for four in-house methods. Docking: FRED, Ligand-based: ROCS, 2D: Lingos and MACCS Keys [20]. Also included for comparison purposes is the average AUC for GOLD against the Warren set with associated error bars What is the source of so much variation such that even forty targets are barely able to provide statistically supportable conclusions? In general, given a property measurement that has N independent sources of error, the expected error is formed from the root mean square of the individual sources of error, thus: For our case we can write: The variances are intrinsic properties to ‘targets’, ‘actives’ and ‘inactives’. How do we know what these variances are? One way is to boot-strap, i.e. leave out a randomly chosen fraction of the targets, or subset of actives or inactives, and measure changes in performance. Repeating this procedure many times gives a statistical sampling of the sensitivity to outliers and the number of measurements. Alternatively, in some cases the variance can be established more precisely. In the case of AUC, for example, it can be shown that for a particular target the variance for both actives and inactives can be approximated by: where the sums are over all actives or inactives, pi is the probability this active i has a higher score than an inactive, qj is the probability an inactive j has a higher score than an active and 〈〉 represents the average of a quantity [7]. Typically, the variances of both actives and inactives are roughly equal. This leads to useful insights as to the required ratio of decoys to actives. When this ratio is 100:1, the net error is only larger by 0.5% than if we were to use an infinite number of decoys. A ratio of 40:1, roughly that of the DUD-self set, yields an impact about 1%. At 10:1, this impact is about 5% and at 4:1 about 11%. Note that these effects on the error estimates, not on the actual average. What does this look like in practice? Figure 2 shows the AUC values for FRED applied to DUD (self-decoys), along with associated 95% confidence intervals for each system. Given these AUCs and contributions to variance from actives and inactives, we can directly address whether the source of the variance across targets is due to insufficient sampling of actives and decoys, or an intrinsic property of methods. By the CLT,Fig. 2AUC values ordered from left to right by number of actives for each target in the DUD set. Program used: FRED with Chemscore as the posing and scoring function. Error bars are 95% confidence intervals for each virtual screen where Therefore Table 1 shows contributions to the square of the average error in the mean AUC across DUD for our four methods calculated in this manner. First, as expected the contribution from the inactives is about forty times less than that of the actives (because the intrinsic variances are similar and there are thirty-six times more decoys than actives in DUD-self). Secondly, it is clear that the errors due to target variability is roughly ten times higher than that due to actives. As independent errors add as squares, this implies only about 5% of the observed confidence limit on the target-averaged AUC is due to the finite number of actives. This leaves 95% of the 95% confidence limit due entirely to the considerable variation from target to target. Comparing DUD-self to the careful evaluation of McGaughey et al. from Merck, and Warren et al. from GSK, the latter have roughly four times more actives and four times less systems, i.e. they are more careful studies of particular systems (error bars are 50% smaller per system) but substantially less useful for general conclusions (error bars are twice as big per method). Remarkably, even if the number of actives in each DUD-self set were reduced by a factor of ten, causing a threefold increase in the error estimation per target, the net error of averages over all systems is only increased by about 30%.Table 1The contribution to observed variance from actives, decoys and targets over the DUD dataset (DUD-self decoys)Method〈Err2〉 − Decoys〈Err2〉 − Actives〈Err2〉 − ObservedEst. 〈Err2〉 − TargetsFRED0.0000480.00200.0230.021ROCS0.0000250.00220.0410.039MACCS 0.000040.00170.0300.028LINGOS0.0000390.00170.0350.033The estimated error (squared) from the variation between targets is estimated from the observed variance and corresponds to that which would be obtained if the number of actives and inactives were infinite The conclusions from this analysis of extensive properties are straightforward. When calculating the properties of a single system the number of actives is fairly important, but the number of inactives does not have to be substantially larger. A ratio of decoys to actives of 4:1 only has an error 11% higher than the limiting value from an infinite number of inactives. It would be more useful to include sets of inactives designed for different purposes than to attempt to ‘overwhelm’ the actives with decoys.If the purpose is to test a method against other methods with 95% confidence then the number of systems required is very large, much larger than even DUD. In our analysis the contributions to the variance from a limited numbers of actives is almost insignificant compared to the target-to-target variation. For example, it would take over 100 test systems to tease apart the difference between the ligand-based method ROCS and the docking program FRED with 95% confidence. (See below.)The variance between systems is such that the number of actives per target does not need to be very large, perhaps even as low as ten. As such, suggestions to only include representatives of chemical classes, e.g. see Good and Oprea [22], may be statistically quite valid. Correlations A key assumption underlying much statistical analysis is the independence of samples, for instance that any two measurements are uncorrelated. This is often a good assumption but it is not hard to find counterexamples. Consider the case where a decoy is included twice. We have gained no new information. Yet Ni, the number of decoys, has increased and so the error goes down. Clearly the error has not really been reduced. Instead, the decoys are no longer independent. In the line-up analogy, this would correspond to including identical twins as fillers. While this is an unlikely mistake, what about two individuals that look very similar? How independent are two molecules and what does this even mean? The temptation is to reach for the familiar chemical definition, i.e. 2D similarity. Even though there is no rigorous definition of chemical similarity, it is a major concern in selecting active populations from chemically related (congeneric) series. Methods that either rely on chemical similarity, or are heavily influenced by it, may not be making independent assessments. Clearly 2D methods fall into this category, and sometimes ligand-based 3D methods. Ideally, methods that use protein structure, such as docking, ought to be less affected, but this is far from proven. Suggestions as to how to improve matters include reducing the set of active to a smaller set of representative structures [22], or giving more weight to the first compound discovered in a series [23]. (Application of similar protocols to decoys is seldom discussed, perhaps because they are less likely to be congeneric). These are practical suggestions derived from knowing the nature of drug discovery. There is also a general approach that eschews the particulars. Two compounds are considered operationally dependent if their rankings under different tests are correlated. For instance, a method that had a size-bias would tend to rank a pair of molecules of comparable extent similarly, no matter what the target. Even without 2D similarity, this implies a less than perfect independence. Imagine a method where all the decoys are of one size and all the actives another. No matter what the actual number of actives and inactives, there are essentially only two molecules, an active and an inactive, and our ability to extract meaningful statistics is severely compromised. Note that the operational part of this definition depends on the nature of the method, i.e. dependence is conditional on the nature of the procedure investigated. Similar situations occur in assessments of genetic linkage. The degree of dependence amongst a set of markers is evaluated by constructing a matrix where the entries are the correlation of phenotypic scores between any two markers. The eigenvalue spectrum of this matrix is then used to assess the actual number of degrees of freedom [24]. Crucially, though, correlation can only be estimated by knowing the behavior of a pair of samples/compounds over many tests/targets. At first glance this suggests that the same set of decoys should be used across all targets in a study. If decoy selection is universal or drug-like this is typically a part of the design, i.e. the decoy set is reused. However, mimetic and modeled decoys rely on the nature of the actives, which will vary from target to target. This might seem a dilemma, i.e. we want to reuse decoys so we know if they are correlated but we cannot reuse them because one set of decoys may be completely inappropriate for another target. Here Skillman’s suggestion is again useful, i.e. there is nothing to stop us including the decoy set for one target in the virtual screen of a second target not as decoys but rather to gain information on operational independence. To distinguish the role of decoys from one system applied to a second system for purposes other than assessing performance, we suggest the term ‘latent’ since these secondary decoys should be hidden from the calculation of performance metrics. In addition to the concept of latent decoys, latent actives can be used to measure operational independence independent of 2D similarity. Another possibility is to use the actives from one system as explicit decoys in other systems. For instance, in the Warren et al. study each set of actives also formed the decoy set for the other targets. The intent was just to produce a set of drug-like decoys, but it fortuitously provides the most compact form for a rigorous estimation of decoy/active independence. This work will be presented elsewhere, along with an elaboration of the techniques for assessing operational correlation. We turn now to the question of target independence. As we have seen, and is widely appreciated, the variation of performance of methods from target to target is considerable. But do certain targets, or classes of targets, behave similarly for certain methods? For instance, one would expect that a docking program parameterized against certain binding motifs would perform similarly across all targets with this motif, if only because of inductive bias. Or one might assume that isoforms of a target are sufficiently akin that docking methods would perform similarly on each. Fortuitously, the Warren study provides one such example in the inclusion of PDFE and PDFS. Figure 3 shows the difference in performance of methods used on isoforms versus the average difference between all other pairs of targets. It is clear there is less variance between the isoforms than unrelated targets. If this were a generalizable result it would have two consequences. On the positive it would mean that methods could be quantified for certain types of problems without requiring large numbers of targets, i.e. because the variance is smaller. On the negative, it would mean that just as considerations need to be made for the true statistical power of closely related actives, or inactives, similar considerations need to be made for targets, increasing the number of targets required to either discern general differences between methods or to reliably gauge progress of a single approach. And, as we have seen, to measure global performance on independent systems already requires sampling beyond common practice.Fig. 3Docking performance against the two isoforms in the Warren study (PDFS and PDFE), compared to the averaged difference over all other pairs of targets The PDFE/S example is a single data point. It is entirely possible that the variation between targets of similar class, or highly conserved isoforms, or even different forms of the same protein structure is not small and that calculating mean properties is still formidable. One might imagine this is a well-researched area, but this appears not to be the case. Retrospective experiments are designed, however poorly, to give an estimate of to what to expect for the next new target and so targets are chosen to be diverse. Software would seldom be used in default, out-of-the-box, mode when there is considerable domain knowledge, i.e. within a set of closely related targets. Hence, the question of method variance over similar systems appears to have been over-looked. The final aspect of independence is correlation between methods. Suppose we have method A and method B, each tested on the same set of targets with the same set of actives and decoys and the results show A is consistently slightly better than B. How can we prove this difference is statistically significant? At first glance this would seem difficult. As illustrated in Fig. 1a and b, the variance of any one computational method over a set of targets in invariably large. As such, the error bars on an average property, such as an AUC or enrichment, are big. So although the average behavior of method A is slightly better than B, this difference would appear statistically insignificant. However, if the test systems are indeed identical this is not the correct assessment. Instead, the CLT is applied to the set of measured differences between methods, e.g. for an AUC example the variance becomes: The formula for Vardiff can be rewritten as: Here Corr(A,B) is a measure of the correlation between methods A and B and is related to the Pearson correlation coefficient, thus: If the tests of methods A and B are independent then the correlation is typically assumed zero and the variance is just the sum of the variances of both methods. If variances were roughly equal, this would typically be a joint error bar √2 larger than the individual error bars. (This also means the common practice of evaluating whether two methods are statistically different by whether their individual error bars overlap is generally incorrect.) However, if the tests applied to methods A and B are identical, correlation needs to be explicitly included. In the case of A always slightly better than B, we need to assess whether the mean difference is larger than the joint confidence limits generated from the variance of the difference between A and B. In fact, if A is better than B by a constant amount we are guaranteed statistical significance because the variance of the difference is zero. In general, methods tend to be positively correlated so that the joint confidence limits are lower than from independent measurements. Confidence limits are convenient because they give a visual estimate of the possible range of true values, typically at a 95% level of confidence. However, joint confidence limits are less graphical as they pertain to pairs of methods. In addition, there is considerable concern in other fields as to the arbitrariness of the 95% value. The origin of this number is R. A. Fisher, whose work in the 1920s still dominates much of the field of practical statistics. Fisher, primarily an agriculturalist, observed a 10% increase in cabbage yield when manure was used. He also observed that only one in twenty plots without manure showed a yield greater than 10%, and so the 95% cut-off was born! From this, and because of the general utility of Fisher’s work on experimental design, a p-value of 0.05 dominates many fields, in particular clinical trials. Essentially, a p-value is the probability a null hypothesis can be rejected. In our case the null hypothesis would be that method B is in fact better than method A, despite average values suggesting the reverse. Under the assumption the difference in performance between two methods is as predicted by the CLT (i.e. Gaussian), we can assign a (p-)value to the probability one method is better only because of random chance. We do this by calculating the area under the normal form for which one appears better than the other. The mathematics of this is shown below: where 〈A–B〉 is the average difference between the methods, the other variables are as defined as above and erf is the inverse cumulative Gaussian, or error, function. If Nt is small, i.e. less than twenty, then a slightly different functional form is more accurate (i.e. from the Student t-test) because the CLT only applies in the limit of large N. For practical purposes the difference can be ignored. The smaller p, the stronger the case for A being better than B. Note, we are not proving how much better A is than B. The best estimate of A’s superiority is still the mean difference of whatever property we are measuring. Rather the p-value refers to the dichotomous question, is A better than B? Table 2 illustrates the above concepts for the four methods listed in Fig. 1b applied to DUD (DUD-self decoys). The diagonal entries are the mean values of the AUC of each method, followed by the associated 95% confidence limits. The upper triangle of the table contains the naïve joint confidence limits, i.e. by summing individual confidence limits, the joint confidence limits assuming independent tests and the joint confidence limits properly calculated with correlation. The lower diagonal contains the Pearson correlation coefficient for each pair of methods, followed by the p-value for the hypothesis that the better method is so only by chance. So, for example, the probability that the 2D and ligand based methods are better than the docking program FRED by chance are around the 10% level, whereas the differences between these three methods themselves is close to 50:50, i.e. in this example we can distinguish ligand-based methods from a docking protocol but not one ligand-based method from another. This is most likely because DUD is not designed to test ligand-based retrieval containing, as it does, many 2D similar actives. Several others have made this point, including the curators themselves [14, 18, 22]. Examples from other fields of how to apply these procedures to differences in AUC can be found in Hanley and McNeil [9].Table 2Statistical measures necessary to accurately assess the relative performance of methods, here applied to the DUD data set (DUD-self decoys)MethodFREDROCSMACCSLINGOSFRED0.684/0.0430.11/0.08/0.070.1/0.07/0.060.1/0.07/0.065ROCS0.17/0.090.732/0.0650.12/0.085/0.050.125/0.09/0.05MACCS 0.03/0.050.70/0.470.734/0.0550.115/0.08/0.055LINGOS0.19/0.140.65/0.360.54/0.310.72/0.061Diagonal terms: average AUC/95% confidence limits. Upper triangle terms: naïve joint confidence limits/joint confidence limits assuming different tests/joint confidence limits assuming same tests and accounting for correlation. Lower triangle terms: Pearson correlation coefficients/p-values that a method has a higher mean AUC by random chance One of the advantages of p-values is the statistical machinery, again developed by Fisher [25], for combining values from different studies. A classic example is the effects of tobacco. It was not one study that convinced the medical profession, but a series of studies and the facility to combine the results that lead to the overwhelming conclusions as to the health risks of smoking. In statistics this is referred to as “Meta-Analysis”. Despite this, and the wide application of p-values in other fields, they are largely absent from modeling, with a few exceptions [5]. In conclusion, correlation is important in all aspects of virtual screening, but perhaps most important and most easily corrected for in the comparison of pairs of methods. Neglecting the effects of correlation between tests is a frequent problem even in clinical studies [2] and to our knowledge has not been properly applied to comparing methods in virtual screening. Metrics Given an experimental design, what quantities should be measured to assess performance? The question suggests a sequential process, i.e. design the experiment and then measure something, whereas good design takes into account what is going to be measured, in particular to what accuracy. However, assuming a given design, how do we extract useful information? In this section we consider what should be measured and why. This is not a quandary specific to virtual screening, in fact is it universal to all prediction exercises. This very commonality can help suggest worthwhile approaches. It also suggests that measures constructed specifically, even uniquely, for chemical virtual screening should be held to a similar standard to those prevalent in the wider world. Is virtual screening really so different from, say, Internet page ranking? In particular, we will consider the issue of ‘early’ behavior, i.e. measures that reward ranking some active compounds near the top of a list. By considering real-world financial parameters we ask whether ‘early’ behavior is even necessarily to be prized. By looking at a large number of virtual screens, we will ask whether such ‘early’ measures are necessary and whether they can be predicted from more fundamental and well-understood properties. Finally, the application of accurate error bounds will be shown to suggest at least one way of quantifying the advantage an expert brings to well-studied systems. Properties of virtual screening metrics A long list of metrics has been applied to virtual screening. What makes for a good metric? The unfortunate answer with some papers is “any metric that will make my method look good”. And if no known metric will suffice, then simply make a one up. This is a typical indicator of an under-regulated and under-developed field. Computer manufacturers used to habitually make up their own measures for the latest processor or operating system, leading to much confusion and annoyance. As a consequence, in 1988 SPEC (Standard Performance Evaluation Corporation) was formed and SPEC Marks became the standard benchmark of anything worth measuring. SPEC had a simple philosophy: “The key realization was that an ounce of honest data was worth more than a pound of marketing hype” [26]. SPEC Marks have evolved over time to now cover CPU, graphics, Java, mail servers, file servers, parallel performance, high performance computing and other aspects. In other words, SPEC is not a single measure because not all users want the same thing, but this does not mean manufacturers can create their own metrics. Rather SPEC is an umbrella organization for a set of open and diverse groups that consider, ratify and develop benchmarks. In this spirit, this section will concentrate on what ought to be general characteristics of a good metric rather than all prevalent quantities. In a somewhat circular manner, one of the first characteristics of a good measure is that everyone uses it. Clearly one of the problems with a field with diverse measures is incomparability, the “apples and oranges” problem. The most straightforward solution is not imposition of a particular standard but full disclosure of all data. The authors of a study may want to present enrichment at 5%, but if the data is freely available others may calculate the enrichment at 1% or 13% or whatever they wish. This would inevitably lead to standardization as independent parties harvest data from many sources, publishing larger and larger studies on the advantages and disadvantages of different methods and measures. This would provide another example of meta-analysis described above. Sometimes a valid excuse against disclosure is that compounds or targets are proprietary. However, just providing lists of actives and inactives in rank order with unique, but not necessarily identifying, tags is enough to calculate most of the metrics for a particular virtual screen. Currently the field of modeling lacks even an agreed upon format for the exchange of such rarely available information. However, if we are going to report a statistic what properties should it have? From considering measures that have become standard in other fields, what characteristics define a good measure? We suggest the following short list:Independence to extensive variablesRobustnessStraightforward assessment of error boundsNo free parametersEasily understood and interpretable Take for example the very popular “enrichment” measure. Everyone understands the concept of enrichment: swirl a pan of water and gravel from the Klondike river in 1896 in just the right way and you ended up with mostly gold. In virtual screening you look at the top few percent and see whether there are more actives than you would expect by chance. As a mathematical formula this is typically presented as: The problem with this measure is that the enrichment becomes smaller if fewer inactives are initially present. Imagine panning for gold with all the sand removed. There would still be the same gold in the pan, along with maybe some pebbles and small rocks, but the eventual relative improvement after ‘panning’, i.e. ‘enrichment’, is reduced. The problem is that the (Fraction of Actives Found) contradicts requirement (i), i.e. is a function of extensive quantities, the number of actives and inactives. This means that enrichment is not actually a measure of a method; it is a measure of a method and a particular experiment. If the ratio of inactives to actives becomes very large it is assumed this problem disappears, i.e. that the limiting behavior obeys (i). This is not true if the enrichment at a given percent is large, i.e. at precisely the points of most interest. Also, enrichment does not meet requirements (ii). At a small enough percentage the enrichment becomes an unstable function of the exact positions of actives in a list. There is also no agreed upon percentage, making this an adjustable parameter (often abused). Finally, other than by bootstrapping, the author knows of no simple assessment of error bounds. However, it is an intuitive measure, easily understood, passing rule (v), and so almost uniquely to this field is the most common metric reported. Some have been aware of the lack of robustness of enrichment and proposed metrics that average over all percentages with weighting schemes. Before we consider these measures we point out that a simple fix to the common variant of enrichment is to make the enrichment refer to the fraction of inactives, not to the fraction of all compounds. This simple change makes the enrichment independent of extensive quantities, more robust, accessible to analytic error approximation [27] and yet suffers only a slight reduction in interpretability. Only a few, such as Jain [14] have used this alternate form. Perhaps the only important failing of this measure is that it lacks a specific name. For the purpose of this paper it will be referred to as the ROC enrichment to distinguish it from the widely abused variety. ROC enrichment has better properties because is related to an even better metric, the AUC, defined as the area under a ROC curve. A ROC curve is simply a plot of the discovered active fraction versus the discovered inactive fraction. (Each point on the ROC curve can be translated to a ROC enrichment by dividing by the fraction of inactives). The AUC is the average of this property over all inactive fractions. Many excellent treatises can be found [6–9] and it has become a standard for classification performance in many disciplines (medical diagnostics, radiology, clinical testing, criminology, machine learning, data mining, psychology and economics to name a few). It satisfies all of the criteria listed above as a metric, including (v), ease of interpretation. The AUC is simply the probability that a randomly chosen active has a higher score than a randomly chosen inactive. The main complaint against the AUC is that is does not directly answer the questions some want posed, i.e. the performance of a method in the top few percent. This is akin to complaining that SPEC Marks do not do a good job of evaluating mobile phone processors; a fair complaint perhaps but hardly justifying creating a new benchmark without the strengths of existing standards. The AUC ought to at least be held as such a standard against which new measures are judged. Early performance in virtual screening Figure 4 illustrates the supposed limitations of the AUC as a measure of performance. The graph shows two ROC curves, each with an AUC of exactly 0.5. Overall this means that an active is as equally likely to out-rank an inactive than the other way around. However, clearly in the case of the solid line a certain fraction of actives is being scored significantly higher than most inactives, while another fraction is being scored worse, i.e. it is only the average behavior that appears even-handed. Similarly, the dashed curve illustrates the case where the actives are all scored better then a certain fraction of the inactives but worse than another fraction. The solid and dashed curves are instances of bimodal score distributions for the actives and inactives respectively. Since the goal of a virtual screen is to save us the trouble of actually screening all the compounds it is entirely reasonable to prefer good ‘early’ behavior. And yet the AUC does not distinguish between such curves and so, it is claimed, is not appropriate.Fig. 4Example ROC plots for “early” and “late” methods It is against this backdrop that metrics such as RIE [28] and BedROC [29] were developed. In both cases the essential idea is to give early rankings of actives more weight than late rankings. In RIE/BedROC actives are given a weight depending on their position in the list using an exponential function running from 1.0 for the top ranked compound to a number typically close to zero for the lowest ranked. The exponential factor, beta, determines how fast this exponential dies away from the top rank and controls how much the RIE/BedROC parameter focuses on the top of the list. The larger beta the greater the early focus. In RIE the sum of active weights is normalized by the RIE of a random distribution of actives. In BedROC normalization is by the maximum dynamic range, i.e. the score with all the actives ranked at the top minus the score with all the actives ranked at the bottom. In addition, by first subtracting the score of the worst-case scenario, BedROC has the elegant property of running from 0.0 to 1.0. The rational behind using these approaches is to give precedence to actives ranking early but not to fall into the trap of choosing a single enrichment value, i.e. be more robust to perturbations in the rank ordering. Impressively, Truchon and Bayly also derive analytic estimations of the error bounds for BedROC and give some suggested values for the beta parameter. Some incorrect statements regarding the AUC mar their work, for instance that random scores do not give an AUC of 0.5, and that the AUC is dependent on the number of actives and inactives, but overall the work is an interesting attempt to answer a perceived need. Applied to the examples in Fig. 4, the ratio of the BedROC score of the solid line to the dashed is about two for a beta of ten and about ten for a beta of twenty. So does BedROC or RIE qualify as a good metric for virtual screening? Comparing against the five criteria listed above, both are more robust than enrichment, and the error protocols for BedROC satisfies criteria (iii). RIE suffers from having an ill-defined numerical interpretation (i.e. how good is an RIE of 5.34?). BedROC attempts to overcome this by scaling between 0.0 and 1.0, but does this qualify as being understandable? There is no absolute, interpretable meaning to a BedROC (or RIE) number, only a relative meaning when ranking methods. Unfortunately, neither BedROC nor RIE satisfy criteria (i) or (iv), i.e. both are dependent on extrinsic variables and have an adjustable parameter, the exponential factor beta. The former, as we have seen, means that scores can only be compared in the limiting case of an excess of inactives and, as in the case of enrichment, this excess has to persist even when the enrichment of actives is very high, i.e. it is exactly when the actives are predominantly at the top of the list that both BedROC and RIE (and enrichment) are most sensitive to the total number of inactives. Interestingly, it would be possible to reformulate both metrics to avoid this problem. Just as ROC enrichment is a better metric than enrichment, an exponential weighting across the ROC curve, rather than to the individual rankings of actives amongst inactives, would remove the sensitivity of these measures to extensive properties. However, there would still remains the issue of the arbitrariness of the exponential factor beta. Just as with enrichment thresholds there is nothing intrinsically wrong with the freedom to select a threshold that is of interest to the particular research group. Some companies might have the facility to physically screen ten percent of their database, another only one percent. However, as a characteristic of a method, or a class of methods, it is a disadvantage. A proponent of a method has a free parameter with which to make their method look favorable or even just less unfavorable. In the example in Fig. 4, a factor of two in BedROC between the methods (beta = 5.0) does not sound anywhere near as bad as a factor of ten (beta = 20.0). Cost structures of virtual screening There is no fundamental meaning to BedROC or RIE. Neither gets to the real heart of why the solid curve in Fig. 4 represents a better method than the dashed curve. In what follows we will argue that this can only be stated with respect to a set of assigned costs, assumed but never stated. We start by noting that the current focus on early enrichment is actually a change in values for the industry. This author recalls conversations in the mid-1990s wherein the concept of missing any potential lead compound was deemed unacceptable. By contrast, a preference for early behavior implies it is acceptable to miss a significant fraction of potential actives in favor of finding a few good leads. There is merit in this approach. Often a chemistry team can only follow up on a small number of leads. High throughput screens can take several months to design and bring on-line, time chemists could use to explore initial leads from a smaller focused set designed by a virtual screen [30]. What are not made explicit in this shift are the costs of the four components of any virtual screen: true positives (TP), false negatives (FN), false positives (FP) and true negatives (TN). Not wanting to miss anything is equivalent to assigning an infinite cost to a false negative. This was never sensible, but reflected a ‘lottery’ mentality prevalent at the time. The reality is that virtual screening never finds drugs; at best it can find things that might, after considerable effort, become drugs. In addition, the attrition rate at many stages in the drug design process means any lead-like compound is at best a bet that will often fail, costing many millions of dollars. A lottery ticket is potentially worth millions; the expected value, i.e. averaged over all contingencies, is usually less than the cost of the ticket. The assumption behind virtual screening is that the value of a true positive similarly averaged is worth the cost of computers and modelers. This is an unproven conjecture. The assignment of a cost structure to the components of a screen is common in the field of medical diagnostics. Here the costs can be estimated with some reliability. A true positive represents the successful diagnosis of a condition that will save money when treated. A false positive means further, costly, tests will need to be performed. A false negative might cost a lot if a more severe condition develops. Finally, a true negative can be set to the cost of the test or a small saving if compared to a more expensive test. If these values are assigned to each “truth table” component (TP, FP, TN, FN), a ROC curve can be transformed into a cost curve. A small caveat is that the ROC curve deals with true and false positive rates and so to transform to real costs the expected number of actives and inactives is required, or at least the ratio of the two. Suppose we apply a cost structure to Fig. 4 as follows:TP = 8.0FN = −2.0FP = −0.16TN = 0.02 Positive numbers are favorable, for instance the cost assigned to a true negative is the saving from not physically screening a compound. At any point in the curve the cost of progressing with all compounds higher than a given threshold t depends on the False Positive Rate (FPR) and True Positive Rare (TPR): Let us assume Na/Ni = 1/100, then: This is a simple linear scaling of the graphs in Fig. 4, as shown in Fig. 5a. As expected, the best approach is to take the method with early performance over the later performance. Notice that the late performing method is never cost effective and even the early method is only cost effective for a narrow range of rankings.Fig. 5(a) Cost weighted versions of the curves in Figure 4 as per the first description in the text. (b) Cost weighted versions of the curves in Fig. 4 as per the second description in the text Now consider a slightly difference weighting:TP = 8.0FN = −2.0FP = −0.04TN = 0.03 Figure 5b illustrates the effect of these new weightings. By reducing the cost of a false positive by 75%, i.e. to around the savings of a true negative, both methods are always cost effective. Furthermore, although the early method has a clear maximum at around 20% of the database, it is actually worth physically screening about 75% of the database. These examples are obviously only illustrative, but the point they make is real. Early enrichment is important only because of an assumed cost structure. Clearly much more complicated models could be constructed, possibly with real data, as with medical tests. However, to the author’s knowledge this has never been published, presented or even discussed within the industry. It is an assumption that early enrichment is better. Likewise, it is also an assumption that virtual screening itself is a productive exercise compared to physical screening. Averaged properties of virtual screening Suppose the cost structure of virtual screening does favors early enrichment. Can we at least say metrics such as RIE and BedROC, perhaps reformulated to be independent of extensive variables, are superior to AUC? If the early behavior shown in Fig. 4 were indeed repeated from system to system then clearly this would be the case. In Fig. 6 we show data from Warren et al. for twenty docking procedures averaged over all eight targets in the study. Examination of these curves reveals nothing that resembles the biphasic nature anticipated from Fig. 4. Individual curves might occasionally suggest biphasic behavior but there is little evidence for this in targetaveraged ROC curves. Figure 7 shows similar curves for the four methods in Fig. 1b averaged across the DUD set. The curves in Fig. 7 are smoother because the averaging across forty targets in DUD is more extensive than the eight from GSK and show even less evidence of biphasic behavior. There are two possibilities for these observations. Either the individual curves are not biphasic or the averaging dilutes this characteristic. It is possible to imagine a technique that would rank one type of actives well, perhaps hydrophobic moieties, but ranks others badly, e.g. hydrophilic ones, but that the proportions of each set differ target to target such that the total behavior appears monophasic. To see if this might be the case we examined two hundred and seventy virtual screens from the Warren study, looking for a divergence between BedROC, with exponential parameter 5.0, and AUC, i.e. an abnormally low AUC and a high BedROC, although possibly the reverse. The results are shown in Fig. 8. Clearly there is a strong correlation between BedROC and AUC. Similar correlations were also seen when higher exponential factors were employed and suggest no evidence for biphasic behavior. A better AUC naturally leads to a better BedROC value and does so with surprisingly little variation. It might still be possible that one of the methods has average biphasic behavior, for instance if all the BedROC/AUC points for a method trended higher. Figure 9 shows this is not the case. Here, the average AUC per method is compared to the average BedROC value for that method. In addition, the correlation is stronger the better the BedROC value, so that methods that have a good AUC will also have a (strongly correlated) good BedROC score. Fig. 6Averaged ROC curves for twenty methods in the Warren study for which scores for all eight targets where available. Programs and scoring functions listed to the right of the graphFig. 7Average ROC curves for FRED, ROCS, MACCS keys and LINGOS over DUD, with DUD-self decoys. FRED was run with the ChemGauss3 scoring functionFig. 8BedROC scores with an exponential factor of 5.0 versus the AUC for 270 virtual screens from the Warren studyFig. 9The average AUC for each method run against all eight targets in the Warren study versus the averaged BedROC score for each such method There is one exception to monophasicity, shown circled in Fig. 9. This point lies outside the 95% confidence limits of both AUC and BedROC. Its BedROC score is significantly higher than expected and its AUC is around 0.5, i.e. random ranking. The target protein represented by this point is PPAR-δ and the method is MVP, a program developed in-house at GSK by Mill Lambert. In conversation Lambert freely admitted that not only did he have extensive knowledge of this target, he used all of this information to tune MVP. Unfortunately, because of certain aspects of the target he could only select one of three chemical classes for this ‘hands-on’ treatment at a cost to the two other classes. Hence, MVP had to be biphasic. It seems interesting that out of two hundred and seventy virtual screens the only outlier from the BedROC-AUC correspondence is an example of expert intervention. An unintended consequence of this study might be a method to spot and quantify expert contributions to virtual screening, i.e. by comparing early behavior, either with BedROC or other metrics, to that predicted from the fundamental measure of AUC. Conclusions In this study we have considered several aspects of experimental design and performance metrics for virtual screening. There is clearly interest in doing things the right way, not least because of a popular, if unproven, belief that virtual screening saves the pharmaceutical industry money. As with many relatively young endeavors, molecular modeling has been long on promises and short on standards, and it is standards that ultimately deliver the proof that our field is useful. For many years the computer industry suffered from similar growing pains. Not only were there few, if any, reliable comparison metrics for different processors, operating systems, compilers and so forth, the proposed benefits of computers were more assumed than quantified. These days no one doubts the impact of the computing revolution. It is to be hoped that a similar statement can one day be made for molecule modeling. It is with this in mind that the following observations and recommendations are made. On the issue of experimental design we propose:Decoy selection needs to be properly labeled as to intent to facilitate inter-study comparison. We have suggested four classifications, universal, drug-like, mimetic and modeled based on examples from the literature and on typical use-case analysis.Providing access to primary data would allow the field to gain cumulative knowledge. The field of modeling has almost no “meta-analysis”, i.e. research combining the results from studies, largely because of a lack of standards as to procedures and measures, but also due to the lack of primary data. A comprehensive format for virtual screening information would be useful.The inclusion of multiple decoy sets of different design and intent for each target in an evaluation would, in combination with (i) and (ii) above, greatly increase the cumulative value of published studies.The number of targets, actives and inactives need to be carefully considered with respect to the purpose of the experiment and the required accuracy of the results. These can be derived from simple statistical methods that are almost never applied.The effects of correlation between actives or inactives can be generally defined as an operational quantity. This could be investigated if actives and inactives for one target were included as explicit or latent decoys for all other targets. Warren et al. provides an example of the first, i.e. decoy sets were made from the actives of other targets. The second is an extension of point (iii), i.e. include multiple sets of decoys in a study but for different purposes. In conjunction with (ii) above, this would provide material for a rigorous analysis of operational correlation in virtual screening.Correlation between targets needs further research, in particular the question of the variance of computational methods on closely related systems.Differences between methods, especially within a single study over multiple targets, should only be reported if the effects of correlation are included. Editors of journals should never publish papers that suggest one method is better than another if these basic statistics are not employed. At a minimum it is recommended that the method variances along with correlation-corrected joint confidence limits be reported. This would allow the estimation of p-values for any assessment of method superiority.On the issue of performance metrics we propose:Deciding on the metrics to be reported should be a community effort, although access to primary data to encourage “meta-analysis” would aid the autonomous adoption of metrics.There are good reasons metrics such as the AUC are popular in other fields and any new or additional measures for virtual screening need to be assessed against the characteristics that have made such metrics successful. Five characteristics required for a metric to be of similar heft to the AUC are proposed: independence to extensive variables, robustness, error bounds, no adjustable parameters and ease of interpretation. As an illustration, an improvement to the common enrichment measure is described. We propose the term “ROC enrichment” for this new measure. Similar improvements to early measures are proposed.Currently, it would seem that providing AUCs and a few ROC enrichment values for the early part of a screen, e.g. 1% and 2%, would capture most average behavior of interest.The assumption that ‘early’ behavior is necessarily a benefit is based on an assumed cost structure that may or may not hold. Similar statements are true for virtual screening in general. A rigorous attempt to assign real-word costs would be of use to the field.We have found very little evidence that suggests average behaviors cannot be accurately predicted by AUC or obvious extensions there of. Those suggesting otherwise need to provide clear-cut, statistically valid, evidence.Divergence from (v) may be an indicator of local or domain knowledge, i.e. knowing the right answer and/or extensive knowledge of the system under study. A potential future area of research is whether this is also an indicator of over-parameterization, posterior system preparation or other reliance on retrospective knowledge. Interestingly, 2D methods applied to DUD, showed no evidence of such a divergence.In conclusion, there is no reason it is not possible to establish standards in the field of molecular modeling necessary to enhance the quality of publications and allow a reliable assessment of methods and progress. However, there are also powerful incentives not to be rigorous. As one invested scientist was heard to pronounce, “livelihoods are at stake”. This is true; we suggest the livelihood of the entire field. Whether the modeling community has the will to enact such measures may well determine whether future generations of scientists look back and see a field that became essential to drug discovery or one that became a mere footnote.

Ann_Hematol-2-2-1705486

Lenalidomide in the context of complex karyotype or interrupted treatment: case reviews of del(5q)MDS patients with unexpected responses

Lenalidomide has particular activity in patients with transfusion-dependent del(5q) myelodysplastic syndromes (MDS), but mechanistic information is limited regarding the relationship between erythroid and cytogenetic responses. We reviewed medical records from three distinct subgroups of del(5q) MDS patients who had unexpected effects with lenalidomide treatment: 1. two patients with complex karyotypes who achieved both cytogenetic remissions and transfusion independence; 2. two patients with 5q- syndrome who took lenalidomide for less than 12 weeks but remained transfusion independent for 15+ months still displaying del(5q) metaphases after 6 and 12 months; and 3. one patient who was a non-responder on lenalidomide during treatment but became transfusion independent for 13+ months after discontinuation. All but the latter patient in this series had reduction of affected metaphases, suggesting that erythroid responses might be mediated by result from partial or complete suppression of the malignant clone, either directly or indirectly through modulation of the bone marrow microenvironment. These clinical observations illustrate the heterogeneity of del(5q)MDS pathogenesis and the diversity of lenalidomide responses within this patient subset. Introduction Myelodysplastic syndromes (MDS) with chromosomal deletion of 5q [del(5q) MDS] are heterogeneous diseases [3, 5]. Apart from the 5q- syndrome [7, 16], patients with del(5q) MDS may present with one additional chromosomal abnormality, with multiple additional chromosomal abnormalities leading to a complex karyotype, or with an increase of bone marrow and/or peripheral blasts irrespective of chromosomal complexity. These distinct disease subgroups have dramatically different prognostic features [3, 5, 6, 12]. Lenalidomide is a thalidomide analog with a distinct clinical profile that has demonstrated erythroid responses leading to RBC transfusion independence, particularly in del(5q) MDS [10, 11] and in some patients with Low or Int-1 MDS lacking the del(5q) chromosomal abnormality [14]. Although the exact mechanism of action has not been defined, lenalidomide is known to have multiple biological activities including anti-angiogenesis, immunomodulation, anti-cytokine, and direct toxic effects on malignant bone marrow cells [1, 4]. We experienced unexpected effects of lenalidomide in del(5q) MDS patients that are suggestive of the mode of action in this patient subgroup and may have implications for future use of the substance in this patient population. To our knowledge, these types of responses to lenalidomide have not been previously described. Study design Between November 2003 and May 2006, 43 patients with del(5q) MDS with or without additional chromosomal abnormalities were treated with lenalidomide at St. Johannes Hospital, Medizinsche Klinik II, Duisburg, Germany. As of December 27, 2005, lenalidomide has been approved for clinical use in the United States, but not in Europe. These patients received lenalidomide as participants of clinical trials or through an expanded access program. Of the cases reported in this paper, only one patient took part in a lenalidomide clinical trial, and that patient was a non-responder. Patients were informed of inclusion of their information in the present report and consent was given. Results and discussion Patients with complex karyotypes Patient 1 A 59-year-old man diagnosed in October 2005 with International Prognostic Scoring System (IPSS) [9] Int-2 risk MDS, >5% bone marrow blasts (RAEB-I) [2, 17], one cytopenia, and complex karyotype: 44, XY, del (5)(q13q33), -7, -15, -18, -19, -19, +mar (15)/46, XY(7) was transfusion-dependent and received two units of packed RBC every 2 to 3 weeks since diagnosis. He had a previous history of two myocardial infarctions, insulin-dependent diabetes mellitus, arterial hypertension, and deep venous thrombosis. Treatment with lenalidomide (10 mg p.o., once daily) commenced in December 2005 and this patient has remained on treatment as of May 2006. He experienced grade 1 neutropenia and grade 2 thrombocytopenia, which required no treatment. The time to transfusion independence was immediate; no further transfusions were required after the initiation of lenalidomide treatment. Hemoglobin was >13 g/dl after 8 weeks, although the time to normalization of hemoglobin (normal level at local laboratory=14 g/dl) has not been reached. Repeat cytogenetic analyses conducted in January and March 2006 showed consistent reductions in the number of abnormal metaphases (Table 1). Table 1Sequential bone marrow cytogenetic analysesPatient numberAnalysis dateKaryotype110/200544, XY, del(5)(q13q33), −7, −15, −18, −19, −19, +3mar [15]/46, XY [7]1/200644, XY, del(5)(q13q33), −7, −15, add(18)(p11), −19, add(19)(q12)[8]/46, XY [7]3/200645, XY, del(5)(q13q33), −7, der(12;14)(q10;10), −14, −15, add(18)(p11), −19, add(19)(q12), +3mar [3]/46; XY [17]23/200446, XX, der(1)t(1;2)(p13;?), der(2)t(1;2)(p13;q31)del(2)(p23), del(5)(q15q31) [19]/ 46, XX [1]12/2004 (FISH)No evidence of 5q31 deletion6/2005 (FISH)No evidence of 5q31 deletion1/2006No evidence of 5q31 deletion32/200446, XX, del(5)(q13q33) [20]5/200546, XX, del(5)(q13q33) [12]/ 46, XX [13]45/200446, XY, del(5)(q13q33) [20]12/200546, XY, del(5)(q13q33) [13]/ 46, XY [8]51/199646, XX, del(5)(q13q33) [15]10/2002No change, non-mosaic pattern of del(5q)3/200312/20036/200411/200406/2006(FISH) indicates that additional fluorescence in situ hybridization was performed. Patient 2 A 78-year-old woman with IPSS Int-1 MDS [no blasts, one cytopenia, complex karyotype: 46, XX, der(1)t(1:2)(p13;?), der(2)t(1;2)(p13;q31)del(2)(p23), del(5)(q15q31)(19)/46, XX (1)], who was diagnosed in March 2004, had an RBC transfusion requirement of two packed RBC every 8 weeks. This lady’s case report has been published in an earlier issue of this journal, but with a shorter follow-up [6]. She had a previous medical history of grade 1 renal insufficiency. Treatment with lenalidomide (10 mg p.o. once daily) commenced in July 2004. The dosage was reduced to 5 mg p.o. once daily from November 2004 to November 2005, and was subsequently reduced to 5 mg p.o. every other day from November 2005 to present. The actual treatment duration was 22 months. Adverse events included grade 2 neutropenia, grade 2 thrombocytopenia, and grade 3 scalp itching. Whereas neutropenia and thrombocytopenia required no additional medication, scalp itching was treated with antihistamines and low doses of prednisone (10 mg) for 2 weeks. Transfusion independence was reached after 1 month and normalization of hemoglobin was reached after 4 months of treatment. Repeat cytogenetic analyses showed complete cytogenetic remission with no evidence of 5(q31) deletion in December 2004 (4 months after start of treatment), in June 2005 and in January 2006 (Table 1). This del(5q) patient with a complex karyotype achieved both durable complete cytogenetic remission and long-term RBC transfusion independence. Patients with transfusion independence despite long-term interruption of medication Patient 3 A 33-year-old woman was diagnosed in February 2004 with IPSS Low-risk MDS (no blasts, one cytopenia) and del(5)(q13q33) as the sole karyotypic abnormality, disease characteristics that are consistent with 5q- syndrome. She had a transfusion requirement of two packed RBC units every 3 to 4 weeks. Lenalidomide treatment was started in November 2004, but therapy was interrupted after 28 days for erythematous lesions of the skin and grade 4 neutropenia. Transfusion independence was reached after 8 weeks without further medication and normalization of hemoglobin was achieved after 20 weeks. Grade 1 neutropenia persisted. Repeat bone marrow cytogenetics in May 2005 showed a reduction in del(5q) metaphases (Table 1). Although treatment was interrupted due to toxicity, this patient continued to improve with respect to both erythroid response and the number of del(5q) metaphases. This patient remains transfusion independent at 21 months after commencement of lenalidomide therapy and 20 months after discontinuation of therapy. Patient 4 A 49-year-old man with IPSS Low-risk MDS typical of 5q- syndrome [no blasts, one cytopenia, isolated del(5)(q13q33)] was diagnosed May 2004. He was transfusion-dependent, requiring two packed RBC units every 2 weeks before the start of lenalidomide therapy in November 2004. However, treatment was stopped after 12 weeks (February 2005) because of grade 3 skin itching. Transfusion independence had been reached after 8 weeks of treatment. Normalization of hemoglobin was achieved after 16 weeks and was ongoing at last visit (21 months) despite receiving no additional treatment. Mild leukopenia has persisted. A repeat karyotype analysis in December 2005 (13 months after onset of therapy) showed a reduction in del(5q) metaphases and no evidence of cytogenetic evolution (Table 1). The lengthy duration of response in this patient was notable and unexpected considering the relatively short treatment period and the persistence of the del(5q) clone. Patient with no response during treatment but transfusion independence after discontinuation Patient 5 A 58-year-old woman with no relevant previous history was diagnosed January 1996 with IPSS Low-risk MDS (no blasts, one cytopenia) with del(5)(q13q33) as the sole karyotypic abnormality. She was transfusion-dependent, receiving two packed RBC units every 4 weeks. Low-dose cytarabine was given in December 1996, resulting in 10 months of transfusion independence, but was again given in January 1998 without effect. All-trans-retinoic acid was administered from July 2002 to January 2003 without improvement of transfusion burden. She subsequently received lenalidomide from January 2004 to November 2004 without reduction in transfusions. This prolonged treatment of more than 10 months was administered in the hope of induction of a late response to the drug. Her hemoglobin value at commencement of lenalidomide therapy was 7.1 g/dl, and she needed two red blood cell transfusions every 3 weeks to maintain these levels throughout drug intake. However, 16 weeks after lenalidomide discontinuation, transfusion independence was achieved in this patient (March 2005) and without additional measures. The response had been ongoing for 16 months, without normalization of hemoglobin (10.9 g/dl at last visit in June 2006). Six repeat karyotype analyses from Oct 2002 to June 2006 all showed the initial finding of a non-mosaic pattern of del(5q) (Table 1). Discussion Clinical trials of lenalidomide have demonstrated cytogenetic and erythroid responses in patients with del(5q) MDS [10, 11] that have not been achieved with other non-cytotoxic agents [8]. The activity of lenalidomide in patients with complex karyotypes including del(5q) represents a major advance, as those patients have an ominous prognosis and are expected to transform rapidly to acute myeloid leukemia [6]. Patient 1, who had Int-2 MDS with del(5q) and a complex karyotype, is of special interest because repetitive karyotyping showed the number of affected metaphases to be gradually reduced during lenalidomide treatment. Patient 2, who also had del(5q) with a complex karyotype, experienced both complete cytogenetic response and normalization of hemoglobin. Two other patients with complex cytogenetic abnormalities are being treated at our institution, but the duration of treatment is too short for final evaluation. Whereas cytogenetic remission with lenalidomide treatment correlates closely with erythroid response [10], complete suppression of the malignant clone was not required for transfusion independence in four of the five cases (patients 1, 3, 4, and 5) reported here. Hematologic remissions occurred in three patients despite treatment interruption and persistence of the del(5q) karyotype (patients 3, 4, and 5). Patient 5 had a late effect and reached transfusion independence after discontinuation of lenalidomide. The long-term hematologic responses seen with patient 3/4 patients may be an exception, as another patient at our department had worsening of anemia 3 months after stopping lenalidomide after a short period of administration (5 weeks). Two other patients at our institution had interrupted lenalidomide treatment after prolonged intake (at months 24 and 13, respectively) in complete hematologic and cytogenetic remission for reasons other than an adverse event. Both patients remain in complete hematologic remission, ongoing for more than 6 and 21 months. Thus, altogether six out of 43 patients at our site have interrupted lenalidomide treatment for non-toxicity reasons. Five experienced an ongoing erythroid response and one patient relapsed after 3 months. Although all of the patients reported here have achieved transfusion independence, lenalidomide treatment does not uniformly lead to transfusion independence in del(5q) MDS. Of note, the lenalidomide MDS 003 study in del(5q) MDS patients shows that recurrence of the cytogenetic aberration or cytogenetic evolution is not an uncommon feature in this patient population during lenalidomide treatment: New chromosomal abnormalities occurred in 24 out of 119 patients, and no single type was seen in more than one patient [10]. Interestingly, chromosome 7 abnormalities occurred in only one patient, in contrast with the higher incidence in the MDS 001 study [11]. List et al. suggested that lenalidomide may act by suppression of the del(5q) clone [11]. All but one patient in our series reported here had reduction of affected metaphases, which supports this hypothesis. However, this may not be a result of an immediate cytotoxic effect, as evidenced by the slow decrease in malignant metaphases as seen with patient 1. The other responding patients, particularly patient 5, showed a slow erythroid response more consistent with a sustained effect on the bone marrow microenvironment. Effective bone marrow modulation might, in some instances, lead to (partial) suppression of del(5q) clones. However, even if clonal suppression did not occur, normal hematopoiesis may be sufficiently strengthened to support a partial erythroid response in some patients. Examples of such erythroid response without alteration of the underlying cytogenetic abnormality have occurred in studies with other immunomodulating drugs like antithymocyte globulin and cyclosporine A [15], differentiation-inducing agents like all-trans-retinoic acid [8], and with cytokine treatment like darbepoietin [13]. Thus, among individual patients within the del(5q) MDS subset, lenalidomide may affect the del(5q) clone and the bone marrow microenvironment to different degrees to result in hematopoietic normalization. Further study is needed to better understand the relationship between erythroid response and cytogenetic remission with lenalidomide treatment. In conclusion, although the above case reports obviously represent only selected cases out of a larger number of lenalidomide treated del(5q) MDS patients, these unusual and unexpected cytogenetic and erythroid responses suggest that some patients with complex karyotypes including del(5q) or who discontinue therapy may still benefit from lenalidomide treatment.

Psychopharmacologia-3-1-2080347

Lack of evidence for reduced prefrontal cortical serotonin and dopamine efflux after acute tryptophan depletion

Rationale Acute tryptophan depletion (ATD) is a widely used method to study the role of serotonin (5-HT) in affect and cognition. ATD results in a strong but transient decrease in plasma tryptophan and central 5-HT synthesis and availability. Although its use is widespread, the evidence that the numerous functional effects of ATD are caused by actual changes in 5-HT neuronal release is not very strong. Thus far, decreases in 5-HT efflux (thought to reflect synaptic release) were only reported after chronic tryptophan depletion or when ATD was combined with blockade of 5-HT reuptake. Introduction A widely applied method for the study of the role of serotonin (5-HT) in cognition and affective processing is acute tryptophan depletion (ATD; Biggio et al. 1974; Fadda et al. 2000b; Fadda 2000). Through dietary depletion of the essential amino acid tryptophan, peripheral and central availability of this precursor of serotonin can be rapidly lowered (Biggio et al. 1974; Moja et al. 1989). This, in turn, leads to reduced synthesis of the monoamine 5-HT (Gessa et al. 1974) and reduced central availability of this neurotransmitter (Fadda et al. 2000b; Lieben et al. 2004a). Since the introduction of this method, ATD has been successfully applied in both human and rodent research and has implicated the serotonergic system in cognition (e.g. memory impairments, reduced attention and increased impulsivity) and affective processes (e.g. relapse of depressed patients, mild depressive state, aggression and irritability; Fadda 2000; Riedel et al. 2002). Various studies have, furthermore, reported ATD-induced effects on typical ‘prefrontal’ tasks, like reversal learning (Rogers et al. 1999; Finger et al. 2007) and intra- and extradimensional set shifting (Park et al. 1994), but also ATD-induced changes in prefrontal cortex (PFC) activity as measured with functional magnetic resonance imaging (Rubia et al. 2005; Allen et al. 2006; Evers et al. 2005). Recent literature, however, shows that the reported cognitive and behavioural effects of ATD are not always reproducible and suggests that ATD as a method for studying 5-HT depletion is particularly suitable for subpopulations that show a vulnerability to 5-HT dysregulation (Fadda 2000; Booij et al. 2003; Fusar-Poli et al. 2006; Jans et al. 2007). These subgroups include remitted depressed patients (Bremner et al. 1997; Young and Leyton 2002; Neumeister et al. 2004), females (Nishizawa et al. 1997; Schmitt et al. 2000; Booij et al. 2002; Ellenbogen et al. 1996) and subjects with certain genetic variations in the 5-HT transporter gene (Roiser et al. 2006; Finger et al. 2007). Fusar-Poli et al. (2006) furthermore suggest that the effects observed after ATD might not be mediated solely through a 5-HT-related mechanism, but rather indirectly via compensatory mechanisms of other transmitter systems (see also Praschak-Rieder et al. 2004). One candidate for this is dopamine (DA). Although direct evidence for ATD-induced effects on other transmitter systems, including DA, is lacking, there is compelling evidence that 5-HT can directly regulate dopaminergic activity and, consequently, function (for a current review, see Alex and Pehek 2007). On the other hand, evidence for the effect of ATD on the actual neuronal release of 5-HT in the brain is sparse and, in most cases, indirect. Human studies are limited in their measurement to imaging studies of central 5-HT synthesis (Nishizawa et al. 1997), 5-HT receptor occupation (Udo de Haes et al. 2002; Praschak-Rieder et al. 2004) and cerebrovascular fluid measurement of tryptophan and 5-HT metabolism (Carpenter et al. 1998). Although these studies partially confirm reduced synthesis of 5-HT and do report lowered levels of the metabolite of 5-HT in cerebrospinal fluid, evidence for reduced release of 5-HT is absent. A number of rodent studies, in general employing the same method of ATD, suggest that in addition to reduced tryptophan availability and reduced 5-HT synthesis, the actual synaptic release of 5-HT is lowered. Evidence for the assumption that ATD does in fact lower 5-HT release is, however, inconclusive. Microdialysis studies have been able to show reduced extracellular 5-HT after ATD only in the presence of a selective serotonin reuptake inhibitor (SSRI; Fadda et al. 2000b; Bel and Artigas 1996; Stancampiano et al. 1997), while there are no reports that show decreased or unaffected 5-HT efflux in animals that are not treated with SSRIs. This is an important observation, as non-physiological blockade of serotonin reuptake can interfere with normal transmitter release, making a clear-cut interpretation of the data difficult. Furthermore, Marco and Meek (1979) show that SSRI treatment in rats can reduce synthesis, which, in turn, could augment the depletion of 5-HT pools induced by ATD (see also Carlsson and Lindqvist 1978; Moret and Briley 1997; Fuxe et al. 1982 and Yamane et al. 2001). In line with the evidence for prefrontal mediated effects of ATD mentioned earlier, we examined the effect of ATD on extracellular concentrations of both 5-HT and DA in the PFC in spontaneously active, and behaviourally stimulated, animals. We aimed to investigate whether successful lowering of blood plasma tryptophan concentration with ATD leads to a reduction of medial prefrontal 5-HT efflux in the absence of a serotonergic reuptake blockade. Materials and methods All experiments were approved by the Animal Experimentation Committee of the Royal Netherlands Academy of Arts and Sciences and were carried out in agreement with Dutch Laws (Wet op de Dierproeven, 1996) and European regulations (Guideline 86/609/EEC). Subjects Subjects were 20 male outbred Wistar rats (Harlan/CPB, Horst, The Netherlands) weighing 175–200 g at arrival. Upon arrival, the animals were socially housed in groups of four in standard type IV macrolon cages where they were kept under a reversed day/night cycle (dimmed red light from 7 a.m. until 7 p.m., white light from 7 p.m. until 7 a.m.) for the duration of the experiment. After surgery and during experimental procedures, the animals were kept in individual cages (25 × 25 × 35 cm). Surgery Two weeks after arrival, the animals were subjected to surgery for placement of a jugular vein catheter and implantation of bilateral microdialysis probes. The weight of the animals at time of the surgery was 297–332 g. Rats were anesthetised with intramuscular Hypnorm (0.24 mg/kg fentanyl citrate and 7.5 mg/kg fluanisone, Janssen); subcutaneous Dormicum (0.75 mg/kg midazolam, Roche) was given for muscle relaxation. A silicon catheter was placed in the right jugular vein to allow for blood sampling, as described by Steffens (1969). For the subsequent implantation of the microdialysis probes, the animals were mounted in a stereotactic frame with the toothbar set at −2.5 mm. Bilateral probes were then placed in the medial PFC at an angle of 12° (A + 30 L ± 18; V − 55). Both microdialysis probes and jugular vein cannula were secured to the scull with dental cement and two cranial screws. Postoperative pain reduction was achieved with Temgesic (0.10 mg/kg buprenorphine, Schering-Plough), 0.03 ml/100 g b.w. s.c approximately 2 h after surgery. After surgery, the animals returned to individual cages for the remainder of the experiments. Experimental procedure Approximately 1 week after surgery, the animals were assigned to either the experimental or control condition, with the body weight of the animals matched between groups. Fourteen hours before the experiment, food was removed from the home cages. On the day of the measurement, the animals were transferred to the experimentation room where they were kept in cages that contained their own bedding material. For the duration of the experiment, the animals were connected to the microdialysis tubing and blood sampling lines, which were attached to a swivel and counterbalanced beam out of reach of the animals. In this way, the animals were unrestricted in their movement, and the collection of blood samples could be achieved without handling the animals. The collection of hourly blood samples started as soon as stable efflux of DA and 5-HT was observed (see below), which was approximately 2 h after connection. A first administration of the diet (see below) followed the initial blood sampling; a second administration was given 90 min after the first. Two and a half hours after the first diet administration, the rats were subjected to 20 min of ‘novelty’ stimulation. During this period, the animals were given access to an identically shaped cage, filled with new bedding material, by means of an opaque sliding door connecting the measurement cage with the ‘novelty’ cage. Optical movement detectors, connected to a PC running locally written software, were placed on top of the cages to detect possible changes in locomotor behaviour due to diet administration. All animals were tested individually on separate days. A radio played to mask background noise and minimise interference from experimental procedures. Treatment The experimental rats (TRP minus group) were treated orally with a protein–carbohydrate mixture lacking tryptophan (TRP; 4.0 g/kg Solugel P™ and 2.0 g/kg Maltodextrine) in a volume of 10 ml/kg (see Table 1). Two diet administrations were given, spaced 90 min apart. Control animals received the same mixture with added TRP (0.28% TRP of total protein). The exact composition of the mixture and the procedure for administration is identical to that described in Lieben et al. (2004a). To minimise stress, the rats were daily handled and habituated to oral injections with normal tap water (10 ml/kg). To minimise TRP intake through the normal rat chow, food was removed from the home cages 14 h before the start of the experiment. Water, however, remained freely available during this period. Table 1Composition of the nutritional mixture and determination of the amino acids content of the gelatin-based proteinMixtureValuesProtein (Solugel P®) in 100 ml tap water100Aspartic acid + asparagine5.2Glutamic acid + glutamine9.3Hydroxyproline12.1Serine3.1Glycine22.5Histidine0.5Arginine8.8Threonine1.1Alanine9.3Proline13.3Tyrosine0.4Valine2.1Methionine0.6Cysteine0.2Isoleucine1.4Leucine3.0Hydroxylysine1.4Phenylalanine1.9Tryptophan0.1Lysine3.6Carbohydrate (Malthodextrine) in 80 ml tap water50KCl0.094CaCl2·2H2O2.32l-Tryptophan (Tryp minus group)0l-Tryptophan (control group)0.28The composition of the administered mixture used in the current experiment is depicted in bold. The amino acid content (%) of the protein was obtained from PB Gelatins, Tessenderlo, Belgium. Blood sampling and quantification of TRP content On the experimental day, hourly blood samples were taken via the jugular vein catheter and analysed for TRP content. We collected six consecutive blood samples during the experiment. The first sample was taken after a stable microdialysis baseline was established and just before the first diet administration was given. Before the start of the experiment, the rats were habituated to blood sampling to minimise stress (i.e. a syringe was attached to the catheter and blood was withdrawn and immediately returned). Blood samples (200 μl) were collected in heparin-filled tubes and deproteinised with 5-sulfosalicylic acid (5%). After centrifuging (15 min at 3,220×g), the samples were stored at −80°C until analysis. After completion of the experiment, the plasma samples were thawed at 4°C, and 20 μl supernatants were transferred to a high-performance liquid chromatograph (HPLC) for analysis (Waters 600E pump and Waters 717plus autosampler, Waters Chromatography b.v., Netherlands; Jasco FP-920 intelligent fluorescence detector, Jasco, Benelux B.V., Netherlands and Shimadzu Class-vp™ software v.5.03 Shimadzu Duisburg, Germany). The mobile phase consisted of 0.06 mol/l sodium acetate, 9 mmol/l citric acid, 0.37 mmol/l sodium 1-heptane sulfonate (HSA) and 12.5% methanol in milliQ water. The flow rate was kept constant at 1.0 ml/min. Separation of TRP from other components was achieved with a Supelcosil column (LC-18-DB 25 cm × 4.6 mm × 5 μm), with a 2-cm guard-column of the same material (Supelco Superguard™, Supelco USA), at a constant temperature of 40°C. Quantification was achieved by means of fluorescence detection, with the excitation wavelength set at 335 nm and an emission wavelength of 440 nm. TRP content was measured against a calibration curve of an external TRP standard. Microdialysis probes and measurements Concentric dialysis probes, manufactured and routinely used in this laboratory (Feenstra and Botterblom 1996), were constructed from fused silica (i.d. 0.075 mm) protected by 25-G needles as inlet and outlet. A Hospal membrane (o.d. 0.32 mm, i.d. 0.24 mm) with 3-mm exposed length was fitted for these experiments. A week after surgery (Feenstra et al. 2000), the animals were transferred to the test room where the dialysis probes were connected to PEEK (polyetheretherketone)-tubing (o.d. 0.51 mm; i.d. 0.13 mm; Aurora Borealis, Netherlands) attached to a dual-channel swivel (Pronexus, Skärholmen, Sweden). A Univentor 801 microinfusion syringe pump delivered a ringer solution (145 mmol/l NaCl, 1.2 mmol/l CaCl2, 2.7 mmol/l KCl, 1.0 mmol/l MgCl2) at a flow rate of 1.3 μl/min. Dialysate was directly introduced into the electrically actuated HPLC injector (Valco, VICI, C6W), and 50 μl was injected every 15 min. DA and 5-HT were separated on a Supelcosil column (LC-18-DB 25 cm × 4.6 mm × 5 μm) and detected by means of an ANTEC VT-03 electrochemical flowcell (50-μm spacer) set at +0.65 V against a Ag/AgCl reference electrode. The mobile phase consisted of 0.06 mol/l sodium acetate, 9 mmol/l citric acid, 0.37 mmol/l HSA and 12.5% methanol in milliQ water. The flow rate was kept constant at 1.0 ml/min; a pulse damper (ANTEC, Leyden) was placed with the column inside the DECADE detector unit (ANTEC) and kept at a constant temperature of 40°C. The DECADE unit was programmed to automatically make an injection every 15 min and simultaneously start registration of the detector output. 5-HT and DA content was measured against a calibration curve of an external standard using Shimadzu Class-VP™ software (Shimadzu, v.5.03). The detection limit (ratio signal to noise = 2) was 0.15 pg of 5-HT and DA injected onto the column. Drugs and chemicals The Gelatine hydrolysate (Solugel P™) was obtained from PB Gelatins (Tessenderlo, Belgium). Glucodry 200 was obtained from the Amylumgroup (Koog aan de Zaan, the Netherlands). Potassium chloride (KCl), calciumchloride–dihydrate (CaCl2·2H2O), sodium chloride (NaCl), sodium 1-heptane sulfonate (HSA), methanol and tryptophan were obtained from Sigma-Aldrich Chemie, Germany. 5-Sulfosalicyclic acid dehydrate and magnesium chloride hexahydrate (MgCl2·6H2O) were obtained from Merck Schuchhardt. Sodium acetate and citric acid were purchased at Janssen Chimica, Geel, Belgium Histological examination After completion of the experiments, the rats were humanely killed by inhalation of a mixture of CO2/O2 (70/30) followed by 100% CO2. The brains were rapidly taken out of the skull and frozen at −20°C. Coronal (20 μm) sections were cut on a cryostat and stained with thionine for examination with the microscope for the precise location of the microdialysis probes. Data analysis Changes in extracellular concentration of DA and 5-HT were calculated as deviations from the baseline defined as the last two samples preceding the first diet administration. Changes in plasma levels of TRP were compared to the first sample preceding diet administration. ‘Movement-counts’, in arbitrary units, registered by the optical movement detector were taken as measure for possible changes in activity. All data were analysed using a repeated measures analysis of variance (ANOVA) with time as within-factor and group/treatment as between-factor. If indicated by the Mauchly’s test of sphericity, the number of degrees of freedom was adjusted by a Huynh–Feldt correction. If indicated by an ANOVA group or interaction effect, subsequent evaluation of the effects of the treatment on individual samples was done by the Student Newman–Keuls test with repeated measures in which all samples are mutually compared. The level of significance was p < 0.05. The effect of novelty exposure on 5-HT and DA release and general activity during the measurement was assessed by recalculation of a baseline from the two samples before novelty exposure. The data were then analysed as described above. All data are presented as means±SEM. Results Histological examination and animal exclusion Histological sections were examined for placement of the microdialysis probes. Data from rats with incorrectly placed probes were not included in the analysis. After exclusion, both the control group and the TRP minus group included eight animals. Plasma concentrations Figure 1 shows the change of plasma tryptophan concentrations for both the control and treated group. Due to blockade of several jugular vein cannulas, the control group included seven and the TRP minus group, five animals. Average plasma concentrations (μg/ml) of tryptophan for the initial baseline samples were 18.80 ± 11.92 for the control group and 21.51 ± 15.64 for the TRP minus group. Fig. 1Plasma tryptophan concentration. Relative concentration (mean±SEM) of tryptophan in plasma compared to baseline. Open triangles correspond to the TRP minus group; closed squares correspond to the control group. The vertical grey bars correspond to, respectively, first diet administration, second diet administration and novelty exposure. Asterisk indicates a significant difference between the two groups, p < 0.05; sigma indicates a significant difference compared to the baseline value, p < 0.05A repeated measures ANOVA with group (TRP minus/control) as between-subjects factor and time as within-subjects factor resulted in a main effect of time [F(4.45,44.50) = 3.07 p < 0.03] and group [F(1,10) = 38.76, p < 0.0005] as well as a group/time interaction [F(4.45,10) = 8.09, p < 0.0005]. An additional t test revealed a significant reduction of the TRP minus group compared to the control group (time point 3: t = 6.00, p < 0.0005; time point 4: t = 7.66, p < 0.0005; time point 5: t = 5.76, p < 0.0005; time point 6: t = 3.15, p = 0.01; Fig. 1). In contrast to the control group that showed an initial increase in plasma tryptophan (up to 167%) and returned to baseline values [F(5,30) = 3.50, p < 0.02], the TRP minus group showed a persistent decrease in plasma tryptophan after the second administration, with a minimum of 29% of baseline 2 h after the first diet administration [F(2.51,10.06) = 15.26, p = 0.001]. A simple contrast ANOVA for the control group revealed that time points 3 and 4 were significantly different from baseline; for the TRP minus, time points 3, 4, 5 and 6 were different from baseline. Microdialysis Microdialysis of 5-HT Figure 2a shows the relative change, compared to two baseline samples, in 5-HT efflux over the entire day. An ANOVA with repeated measures revealed no significant group [F(1,14) = 0.57, p > 0.4], time [F(13.49,188.78) = 1.19, p > 0.2] or interaction effect [F(13.49,14) = 0.97, p > 0.4]. Average transmitter concentrations (pg/50 μl) of serotonin for the initial baseline samples were 1.41 ± 0.27 (mean ± SEM) for the control group and 1.48 ± 0.23 (mean ± SEM) for the TRP minus group. Fig. 2Microdialysis of 5-HT and DA and general activity counts. Relative change in extracellular 5-HT and DA (mean±SEM) and movement compared to baseline. Open triangles correspond to the TRP minus group; closed squares correspond to the control group. a 5-HT, b DA and c movement counts. The vertical grey bars correspond to, respectively, first diet administration, second diet administration and novelty exposure. No significant changes were observedNo significant differences between treated and control group were observed during novelty exposure when values were expressed relative to the original baseline or to a recalculated baseline just before the exposure. Microdialysis of DA Figure 2b shows the relative change, compared to two baseline samples, in DA efflux over the entire day. An ANOVA with repeated measures revealed a significant effect of time [F(6.95,60.58) = 2.76, p < 0.02], but not of group [F(1,14) = 1.62, p > 0.2] or of interaction [F(8.14,14) = 0.98, p > 0.4]. Subsequent analysis of individual groups showed no time effect for either group.Average extracellular transmitter concentrations (pg/50 μl) of dopamine for the initial baseline samples were 0.64 ± 0.15 (mean±SEM) for the control group and 0.72 ± 0.16 (mean±SEM) for the TRP minus group.No significant differences between treated and control group were observed during novelty exposure when values were expressed relative to the original baseline or to a recalculated baseline just before the exposure. Movement detection Figure 2c shows the average number of activity counts for both the control group and the TRP minus group. Average movement counts for the initial baseline samples were 123.59 ± 17.71 (mean±SEM) for the control group and 110.38 ± 14.22 (mean±SEM) for the TRP minus group. These data support the behavioural observation of sustained general activity throughout the measurements (e.g. grooming and explorative movement). A repeated measures ANOVA revealed a significant effect of time over the entire day [F(10.27,143.72) = 2.69, p < 0.005]. No group [F(1,14) = 0.42, p > 0.5] or interaction [F(10.27,14) = 0.55, p > 0.8] effect was observed. The novelty stimulation did not cause any differential effect on overall activity. Discussion To gain insight into the relation between ATD and the release of serotonin (5-HT) in the prefrontal cortex, we measured plasma levels of tryptophan and, simultaneously, PFC efflux of 5-HT and DA in animals that underwent ATD. In line with current literature (Lieben et al. 2004a), ATD induced a rapid decline of free plasma tryptophan within 2 h of the initial administration that lasted throughout the experimental procedure. However, despite this reduction, efflux of prefrontal serotonin (5-HT), as measured in the medial PFC with microdialysis, was unaffected in these animals. In contrast to previous studies, these measurements, in behaviourally active animals, were performed in the absence of a serotonergic reuptake blocker. These results indicate that successful lowering of plasma tryptophan, to levels associated with behavioural effects, does not necessarily induce lowering of 5-HT efflux. Considering the widespread use of ATD as a method for studying the role of serotonin in behavioural and cognitive research, these findings are especially relevant. The current experiment combined blood sampling through a chronic jugular vein catheter with online microdialysis measurements of 5-HT and DA. This combination of techniques was used to ensure that blood sampling did not hinder the animals in their behaviour or cause unwanted effects on transmitter release. In line with current literature, we measured a rapid decrease of plasma tryptophan of more than 70%. Although the possibility exists that a greater reduction of plasma tryptophan, which can be achieved with chronic tryptophan depletion (Fadda et al. 2000a), could have induced a reduction of 5-HT efflux, it would not explain the effects on behaviour and affect observed after tryptophan depletions of a magnitude similar to those observed in the current experiment (e.g. Riedel 2004; Lieben et al. 2004b; Delgado 2006). The 4.5-h time-window in which we measured is well within the period for which behavioural effects have been reported, reducing the likelihood that a decrease in 5-HT efflux would have been observed had we measured for a longer period. Lack of sensitivity of the measurement equipment does not seem likely either; with the current detection limit, reductions of either transmitter would have been measurable. The absence of a clear effect of novelty stimulation on either general activity or transmitter efflux might suggest that the animals were inactive. However, behavioural observation, as well as quantitative movement detection, indicates otherwise, showing that the animals were in fact active throughout the measurements (e.g. grooming, and explorative behaviour), although the active novelty exposure did not induce the expected increase in locomotor activity. Literature data from both human and rodent studies show that ATD leads to lowering of plasma tryptophan (Biggio et al. 1974; Moja et al. 1989) and decreased synthesis (Gessa et al. 1974) and availability of central 5-HT (Ashley and Curzon 1981; Fadda et al. 2000b; Lieben et al. 2004a), but, to date, it has not been convincingly shown that the ATD lowers efflux of 5-HT in the brain. The few rodent studies that do show reduced efflux after ATD either used chronic treatment with a tryptophan-deficient diet (Fadda et al. 2000b; Van der Stelt et al. 2004) or combined the measurements with drugs that prevent reuptake of 5-HT (Fadda et al. 2000b; Bel and Artigas 1996; Stancampiano et al. 1997; but see Bel and Artigas (1996) who observe an increase, rather than a decrease, of 5-HT efflux in the PFC after ATD in rats on free food). To the best of our knowledge, the current experiment is the first to report on the effect of ATD on prefrontal 5-HT efflux after overnight fasting in the absence of a reuptake blocker, an experimental design that parallels the behavioural studies in both rodents and human subjects. As pointed out in the “Introduction”, this is especially important, as the addition of reuptake inhibition can lead to decreased 5-HT synthesis, and thus, augment the effect of ATD. In light of the behavioural effects observed after acute tryptophan depletion and the general notion that a reduction of synaptic 5-HT release underlies these effects, understanding the effect of ATD on 5-HT efflux, which is thought to reflect release, is essential. Our data seem to suggest that successful tryptophan depletion does not alter 5-HT release. ATD, however, has convincingly shown to be an effective method to manipulate mood and induce cognitive/behavioural effects in both humans (Fusar-Poli et al. 2006; Riedel et al. 2002) and rodents (Fadda 2000; Riedel 2004). This apparent discrepancy will be discussed in the next section. Although mood effects after ATD have been reported across populations of subjects, groups with vulnerability for 5-HT dysregulation (e.g. remitted depressed patients, women and people that show a genetic variation in the 5-HT transporter gene) seem particularly sensitive to ATD (Delgado 2006; Booij et al. 2002; Moore et al. 2000; Neumeister 2003; Riedel et al. 2002; Roiser et al. 2006; Jans et al., 2007). It is argued that especially these groups are sensitive to ATD because they already show a reduced availability of central 5-HT that is further reduced by tryptophan depletion. In contrast to non-vulnerable subjects, these subpopulations will be less able to compensate for a temporary reduction in 5-HT synthesis and will experience a greater impact of ATD. Likewise, SSRI-treated animals will experience depletion of 5-HT stores (Marco and Meek 1979) and more readily show decreased 5-HT efflux after ATD than unmedicated rats. Measurements of indices of 5-HT release in response to ATD in ‘vulnerable’ rodents, like serotonin transporter knockout animals (Holmes et al. 2003), or females have not been performed. In contrast to effects on mood, behavioural and cognitive effects are generally observed in a broader population that also includes healthy subjects (for a review, see Riedel et al. 2002) that do not show a particular sensitivity to 5-HT dysregulation. The most parsimonious explanation that can unify those observations with the current data is that brain areas outside the medial PFC, which have not been measured in the present experiment, might mediate the behavioural effects. As the innervation of 5-HT in the brain is not uniform (McQuade and Sharp 1997; see for an overview Frazer and Hensler 1999) and some areas appear to be more sensitive to ATD than others (Lieben et al. 2004a; Fusar-Poli et al. 2006), the absence of 5-HT efflux reductions in the medial PFC might not reflect ATD-induced changes in other (prefrontal) areas. However, this explanation seems to hold only partially at best. As mentioned in the “Introduction”, ATD studies have implicated prefrontal 5-HT in various ‘prefrontal’ tasks, like reversal learning (Rogers et al. 1999; Finger et al. 2007) and intra-, extradimensional set shifting (Park et al. 1994), and shown ATD-induced changes in PFC activity (Rubia et al. 2005; Allen et al. 2006; Evers et al. 2005). The possibility that other transmitter systems might mediate ATD effects (Fusar-Poli et al. 2006) was investigated by simultaneous measurements of DA. As we reported that DA efflux did not respond to the ATD, the current data cannot confirm this hypothesis, although other possible transmitters remain to be investigated. While it is not possible to extrapolate these findings to brain areas outside the medial PFC, our present data show that an effective reduction of plasma tryptophan with ATD does not necessarily imply a similar reduction in 5-HT efflux. As ATD continues to be an important and widely used paradigm to study specific roles of 5-HT, it is important to clarify through which mechanism ATD exerts its effects and to what extent changes in 5-HT contribute to the functional effects of ATD.

Ann_Biomed_Eng-2-2-1705491

Bioactive Hydrogel Substrates: Probing Leukocyte Receptor–Ligand Interactions in Parallel Plate Flow Chamber Studies

The binding of activated integrins on the surface of leukocytes facilitates the adhesion of leukocytes to vascular endothelium during inflammation. Interactions between selectins and their ligands mediate rolling, and are believed to play an important role in leukocyte adhesion, though the minimal recognition motif required for physiologic interactions is not known. We have developed a novel system using poly(ethylene glycol) (PEG) hydrogels modified with either integrin-binding peptide sequences or the selectin ligand sialyl Lewis X (SLeX) within a parallel plate flow chamber to examine the dynamics of leukocyte adhesion to specific ligands. The adhesive peptide sequences arginine–glycine–aspartic acid–serine (RGDS) and leucine–aspartic acid–valine (LDV) as well as sialyl Lewis X were bound to the surface of photopolymerized PEG diacrylate hydrogels. Leukocytes perfused over these gels in a parallel plate flow chamber at physiological shear rates demonstrate both rolling and firm adhesion, depending on the identity and concentration of ligand bound to the hydrogel substrate. This new system provides a unique polymer-based model for the study of interactions between leukocytes and endothelium as well as a platform to develop improved scaffolds for cardiovascular tissue engineering. Introduction Leukocyte adhesion to sites of inflammation is crucial to eliminate the cause of irritation and repair the surrounding tissue. After the initiation of inflammation by cytokines such as interleukin-1β (IL-1β), tumor necrosis factor-α (TNF-α), lipopolysaccharide, and interleukin-3, (IL-3), leukocyte velocity slows dramatically due to leukocyte contact with the vascular wall.9,27,34 Rolling, the initial transient interaction, is mediated by selectin molecules present on the activated endothelial cell surface. E-selectin is an inducible surface glycoprotein that is known to bind the carbohydrate ligand sialyl Lewis X (SLeX), which is present on various leukocytes10,11,13,35,41,43 and mediates stronger adhesions and slower rolling than P- or L-selectin.27 Following this primary contact, firm adhesion takes place via activated β2-integrins binding intracellular adhesion molecule-1 (ICAM-1) expressed on the endothelium, and through the most important member of the β1 integrin subfamily, very late antigen-4 (VLA-4; α4β1). VLA-4 binds vascular cell adhesion molecule-1 (VCAM-1), and is responsible for lymphocyte adhesion to vascular endothelium and leukocyte recruitment to the inflamed area.1,10,11,27 Though well studied, the exact mechanism of adhesion to the vascular endothelium is not known. There are two suggested mechanisms of integrin activation; one proposing that activation is a result of chemokine stimulation of leukocytes2,43 and another suggesting that selectin binding leads to integrin activation.12,42 Several systems have been developed to examine the dynamics of rolling and firm adhesion and to elucidate the exact mechanism of the binding cascade. Leukocytes isolated from human blood have been studied under static and flow conditions on activated endothelial cell monolayers as well as on cells engineered to express endothelial cell adhesion molecules.3,28,37,40 Polystyrene microspheres coated with adhesion ligands have also been shown to interact with stimulated cell monolayers17 and on substrates coated with selectins and other cell adhesion molecules.11 Yeast engineered to display E-selectin has been observed rolling on substrates coated with SLeX. 2 Recently, poly(ethylene glycol) (PEG) has been tethered to gold surfaces to create cell resistant surfaces and spatial gradients of PEG on gold surfaces have been used to study the kinetics of static cell adhesion.32,33 While each of these systems has revealed new insight into the dynamics of leukocyte interactions with endothelial cell adhesion molecules, the ultimate goal must be to establish a simple, cell-free system that more closely mimics the in vivo environment. PEG hydrogels are crosslinked hydrophilic networks that demonstrate excellent biocompatibility, being highly resistant to protein adsorption and cell adhesion, and causing minimal inflammatory responses.20,21,22,36 These highly swollen networks have similar water content and mechanical properties to soft tissues and may be engineered to contain cell adhesion peptides, growth factors, and therapeutics for localized drug delivery.5,6,26,29,30,39,44 PEG diacrylate hydrogels may be covalently modified with cell adhesive peptide sequences to encourage cell adhesion, spreading, and migration and the interactions of cells with these hydrogels has been extensively studied under static conditions.15,16,24,25,29,31,38 A key benefit of such materials is that the base material, PEG, is intrinsically resistant to protein adsorption, so the adhesive interactions with cells are limited to the factors that are specifically engineered into the hydrogel network. In this work we propose a novel system to study leukocyte adhesion under shear using photopolymerized PEG copolymer hydrogels. After forming thin flat PEG diacrylate base hydrogels, we are able to polymerize a layer of monoacrylate PEG-peptide (or SLeX) to the surface, and using a parallel plate flow chamber, we can observe cell adhesion using video microscopy. Materials And Methods All chemicals were purchased from Sigma–Aldrich (St. Louis, MO) unless otherwise stated. Synthesis of Polyethylene Glycol Diacrylate Polyethylene glycol diacrylate (PEG-DA) was synthesized by dissolving 12 g dry PEG (MW: 6000; Fluka, Milwaukee, WI) in 16 ml anhydrous dichloromethane (DCM) with an equimolar amount of triethylamine and 0.72 g acryloyl chloride (Lancaster Synthesis, Windham, NH) added dropwise. The mixture was stirred under argon for 24 h, washed with 2 M K2CO3, and separated into aqueous and DCM phases to remove HCl. The DCM phase was dried with anhydrous MgSO4 (Fisher Scientific, Pittsburgh, PA), and the PEG diacrylate was then precipitated in diethyl ether, filtered, and dried under vacuum at room temperature overnight. The resulting polymer was dissolved in N,N-dimethylformamide-d7 and characterized via proton NMR (Avance 400 MHz; Bruker, Billerica, MA) to determine the extent of acrylation. Synthesis of PEG Derivatives Containing Cell Adhesion Molecules The cell adhesive peptide sequence used in this study include arginine–glycine–asparagines–serine (RGDS; American Peptide Company, Inc., Sunnyvale, CA), and a peptide containing the cell adhesive leucine–asparagines–valine sequence (glycine–proline–glutamic acid–isoleucine–leucine–asparagines–valine–serine–threonine, GPEILDVST), which was synthesized using standard fluorenylmethoxycarbonyl (Fmoc) chemistry on an Applied Biosystems 431A peptide synthesizer (Foster, CA). The high affinity 4-((N′-2-methylphenyl)ureido)-phenylacetyl-leucine-aspartic acid-valine-proline (Bio1211; Commonwealth Biotechnologies, Inc., Richmond, VA)5,6 was also used as an alternate LDV-containing compound. The non-adhesive sequences used as negative controls were arginine–glycine–glutamic acid–serine (RGES; American Peptide Company, Inc., Sunnyvale, CA) and glycine–proline–glutamic acid–isoleucine–leucine–glutamic acid–valine–serine–threonine (GPEILEVST), also synthesized using Fmoc chemistry on a peptide synthesizer. Peptides were conjugated to PEG monoacrylate by reaction with acryloyl-PEG-N-hydroxysuccinimide (PEG-NHS; MW 3400; Nektar Therapeutics, Huntsville, AL) in 50 mM sodium bicarbonate (pH 8.5) at a 1:1 molar ratio for 2 h. The mixture was then dialyzed (MWCO 1000), lyophilized, and stored at −20°C. Gel permeation chromatography with UV and evaporative light scattering detectors (Polymer Laboratories, Amherst, MA) was used to determine the coupling efficiency. The selectin ligand sialyl Lewis X (SLeX) was conjugated to PEG using an avidin–biotin bridge. PEG-NHS (25 mg) was reacted with a lysine–biotin conjugate (biocytin) at a 1:2 molar ratio in 50 mM sodium bicarbonate (pH 8.5) for 2 h. SLeX-biotin (500 μg; Glycotech, Gaithersburg, Maryland) was reacted with avidin (10–15 units mg−1) at a ratio of 2 units avidin per mole of SLeX-biotin in 50 mM sodium bicarbonate (pH 8.5) for 2 h. The reaction mixtures were then combined to allow further conjugation of avidin and biotin, thereby linking acryloyl–PEG–NHS–lysine–biotin to avidin–biotin–SLeX. Synthesis of Bilayered PEG Copolymer Hydrogels Hydrogels were formed by first dissolving 0.2 g ml−1 PEG diacrylate in 10 mM HEPES buffered saline (HBS, pH 7.4); the polymer solution was then filter sterilized using a 0.22 μm filter (Gelman Sciences, Ann Arbor, MI). The photoinitiator 2,2-dimethoxy-2-phenyl acetophenone in N-vinylpyrrolidinone (300 mg ml−1) was added at 10 μl ml−1 polymer solution. This mixture was injected between rectangular glass plates separated by 0.5 mm spacers and polymerized under UV light (365 nm, 10 mW cm−2) for 30 s. The top plate was removed and the hydrogel surface rinsed with sterile PBS. A second layer, consisting of 5 μmol ml−1 of either acryloyl-PEG-peptide or acryloyl-PEG-SLeX in HBS and 10 μl ml−1 2,2-dimethoxy-2-phenyl acetophenone in N-vinylpyrrolidinone was then layered on top of the PEG diacrylate base gel, the upper glass plate replaced, and the second layer photopolymerized by exposure to UV light (365 nm, 10 mW cm−2) for 1 min. Cell Maintenance JURKAT cells (human T-lymphocytes; ATCC, Manassas, VA) were maintained in RPMI-1640 prepared with 10% fetal bovine serum (FBS; BioWhitaker, Walkersville, MD), 2 mM l-glutamine, 1 unit ml−1 penicillin, and 100 mg l−1 streptomycin (GPS). 300.19/E cells (mouse pre-B lymphoblast; ATCC, Manassas, VA) were sustained in RPMI-1640 prepared with 10% FBS, 1% GPS, and 0.1 mM 2-mercaptoethanol. Cells were maintained at 37°C in a 5% CO2 environment. Cell Adhesion and Rolling on Adhesive PEG Gels Flow assays were performed using a circular parallel plate flow chamber (Glycotech, Gaithersburg, MD) mounted on the stage of a Zeiss Axiovert 135 microscope (Carl Zeiss Inc., Thornwood, NY). The chamber was placed on top of photopolymerized PEG copolymer hydrogels and vacuum sealed to the surface using a portable vacuum pump (Fisher Scientific, Pittsburgh, PA) as shown in Fig. 1. Cell suspensions were drawn through the flow field (1 cm path width, 0.01 in thickness) using a programmable syringe pump (BS-8000 Multi-Phaser™ Programmable Syringe Pump, Braintree Scientific Inc., Braintree, MA) at varying flow rates corresponding to a shear stress range of 3.5–35 dynes cm−2, which is comparable to in vivo shear rates. Cellular interactions with the hydrogels were monitored using a Nikon CoolPix 5000 camera (Nikon Inc., Melville, NY) and transferred to videotape for further analysis. Figure 1.Schematic of parallel plate flow chamber system. A small pump is used to create a vacuum, sealing the chamber on the surface of the gel. Cells are then perfused over the gel surface using a programmable syringe pump. Cation Dependent Binding JURKAT cells were treated with 2 mM magnesium (Mg2+), calcium (Ca2+), or manganese (Mn2+) or with 10 mM EDTA and then perfused through the flow chamber and allowed to settle on the gel for 5 min. Controls were exposed to standard formulations of RPMI-1640 containing 0.4 mM Ca2+ and 0.4 mM Mg2+. Ten fields of view were scanned to get an average number of cells per field of view. Flow rates corresponding to shear stresses of 0.5, 1.0, and 10 dynes cm−2 were used to wash away unbound cells. The number of cells remaining for each shear stress was counted and averaged over several fields of view. LDV Specificity JURKAT cells treated with 2 mM Mg2+ were allowed to settle on the LDV gel for 5 min and an average number of cells per field of view was determined. Specificity was demonstrated by the addition 7 μg ml−1 of either a mouse anti-human monoclonal antibody that blocks VLA-4 binding to VCAM-1 (anti-CD49d, clone BU49; Ancell Corporation, Bayport, MN) or an IgG1 isotype control (purified mouse myeloma IgG1; Invitrogen Corporation, Carlsbad, CA) at a shear stress of 0.5 dynes cm−2. The average number of cells bound per field of view was again counted to determine the amount of cells remaining bound to the surface. Specificity was also demonstrated by the addition of a solution of either 10 mM EDTA or 150 μM Bio1211 introduced into the flow chamber under a shear stress of 0.5 dynes cm−2. An average number of cells bound to the gel under flow was determined every minute for 13 min. Video Analysis Cells were allowed to settle on each gel for 5 min. An average of 10 fields of view was scanned and the number of cells settled on the peptide gel was counted. After flow began, fields of view were scanned again and the number of cells remaining (bound to the gel) was counted. After flow was established on the SLeX gels, video was paused and the number of interacting cells was counted. The numbers were averaged over 10 fields of view for each shear stress. Statistical Analysis Data were compared with two-tailed, unpaired t-tests; p-values less than 0.05 were considered to be significant. Results Synthesis of PEG Hydrogels PEG hydrogels were formed under UV light in the presence of a photoinitiator between two glass plates. Hydrogels 0.5 mm thick were formed after 30 s of exposure. The addition of an acryloyl-PEG-peptide derivative mixed with the photoinitiator to the surface of the hydrogels resulted in a covalently bound layer of cell adhesive moiety on the surface (Fig. 1). Quantification of Cell Adhesion and Rolling Cell rolling on SLeX hydrogels was quantified over a range of shear stresses, and 86.8 ± 11.6 cells per field of view rolled on the gel surface at 3.5 dynes cm−2 with an average rolling velocity of 141.3 μm s−1 (Fig. 2). Rolling decreased with increasing shear stress (7.0 dynes cm−2: 15.2 ± 3.8 cells, average rolling velocity 232.7 μm s−1; 14 dynes cm−2: 8.2 ± 1.9 cells, average rolling velocity 486.8 μm s−1; 21 dynes cm−2: no rolling observed). Figure 2.300.19/E cells rolling on SLeX modified gel. 300.19 cells transfected with human E-selectin was perfused over a SLeX modified gel. The number of rolling cells were counted per field of view. SLeX ability to support rolling decreased as shear stress increased (*p < 0.02 as compared to shear rate of 3.5 dyne cm−2). Approximately 98.5 ± 18.6% of cells contacting hydrogel surfaces modified with Bio1211 adhered to the gel surface. In comparison, 41.5 ± 13.4% of cells in contact with the surface of PEG-LDV, an interaction specific for the α4β1 integrin expressed on JURKAT cells, adhered to the hydrogels and 23 ± 8.5% of cells contacting PEG-RGDS hydrogels adhered to the gel surface (Fig. 3). No cells adhesion to PEG-DA or PEG-RGES hydrogels was observed; however the control LEV peptide-modified gels had 1.6 ± 1.2% of cells contacting the surface adhere. Figure 3.Peptide ability to bind VLA-4. JURKAT cells were allowed to settle on PEG-DA gels conjugated with different peptides at similar densities. After 5 min, cells were subjected to a shear stress of 0.5 dynes cm−2 to remove unbound cells and the remaining cells were counted (*p < 0.02 compared to Bio1211). Cation Dependant Binding Cation sensitivity was assessed by the addition of cations or EDTA to cell cultures prior to their exposure to PEG-LDV hydrogels. The presence of 10 mM EDTA had the greatest inhibitory effect on binding, followed by 2 mM Ca2+. Mg2+ and Mn2+ had little effect on the ability of JURKAT cells to bind LDV, and a higher number of cells exposed to both calcium and magnesium were able to bind to the gel surface than those exposed to calcium alone (Fig. 4). Figure 4.Cation dependent binding of VLA-4 to LDV peptide. JURKAT cells were incubated with 2 mM cations or 10 mM EDTA prior to settling on the LDV gel; controls were exposed to standard formulations of RPMI-1640 containing 0.4 mM Ca2+ and 0.4 mM Mg2+. The percentage of cells on the gel was determined based on the total number of cells before initiating a shear stress of 0.5 dynes cm−2 (*p < 0.05 compared to Control). LDV Specificity JURKAT cell binding to PEG-LDV hydrogels was reversed upon the addition of a monoclonal anti-VLA-4 (3.4 ± 2.1% bound). Cells exposed to an isotype-matched control antibody were still able to adhere to the hydrogel surface (88.8 ± 33.4% bound), demonstrating specificity of LDV for the VLA-4 integrin. The higher affinity of Bio1211 compared to the LDV peptide was also observed by the addition of unbound Bio1211 after cells were allowed to settle on PEG-LDV hydrogels (Fig. 5). Bio1211 was able to remove cells from the gel surface better than EDTA through competitive binding of VLA-4. Figure 5.LDV Specificity for VLA-4. 10 mM EDTA or 1.5 uM Bio1211 was introduced into the flow chamber at a wall shear stress of 0.5 dynes cm−2 after JURKAT cells were allowed to settle on the LDV gel for 5 min. Bio1211, which has a higher affinity for VLA-4 was able to compete the cells off the LDV modified gel surface. EDTA was not as effective as Bio1211 as cells detached more slowly and not completely. The dashed line represents the uninhibited controls, in which 40 ± 15.2% of the cells initially exposed to the hydrogel remained bound through the duration of the experiment. Each data point represents the average percent of bound cells per 10 fields of view and error bars represent the standard deviation of the percent of bound cells within those 10 fields of view. The influence of peptide concentration on cell adhesion was determined by varying the amounts of LDV bound to the hydrogel surface (Fig. 6). The number of cells bound to the gel increased with LDV concentration, and LDV was better able to support adhesion at lower shear stresses for all concentrations when compared to higher rates of shear. Figure 6.Dilution effect of LDV. The number of JURKAT cells bound to the gel decreased as the LDV peptide concentration was lowered. LDV was better able to support adhesion at lower shear stresses for all concentrations (*p < 0.05 comparing shear rates at each concentration). Discussion The system of hydrophilic hydrogels modified with cell adhesion peptides implemented in this study exhibits the ability to mimic the cell adhesion cascade that occurs during the onset of inflammation in the vascular system. The capacity to modify hydrogels with cell adhesion molecules for the study of cellular interactions with biomaterials for drug delivery and tissue engineering is well documented.4,7,15,16,18,24–26,29 –31,38,39,44 This study implements the use of these materials under physiological flow conditions to examine the mechanisms of leukocyte rolling and firm adhesion. The formation of thin flat PEG diacrylate hydrogels for use with a parallel plate flow chamber improves upon earlier systems that utilize hydrophobic surfaces with adsorbed cell adhesion molecules. The highly crosslinked structure of swollen PEG networks have water content similar to native vessels, and covalent incorporation of cell adhesive peptides guarantees control of concentration and allows for patterning of one or more adhesion sequences on the gel surface.23 This system is based on a flexible substrate with tunable stiffness, the properties of which can be exploited to examine the responses of different cell types in microenvironments that mimic native tissues in various states of development, remodeling, regeneration, and disease. It has been suggested that the differentiation of cellular function and response could depend significantly on matrix elasticity,8,14,45 and varying the polymer composition or concentration in the hydrogel can alter the permeability and mechanical properties,36 simulating a range of biologically relevant conditions. In addition, the local mobility of adhesive ligands in the solvated hydrogel system should be considerably better than on solid substrates, allowing more variable orientations which could contribute to a greater fraction of accessible ligands available to receptors on the cell surface. Leukocyte cell adhesion to the surfaces of modified PEG-DA hydrogels was highly specific, reversible, and sensitive to ligand site density and affinity, demonstrating the efficacy of the system to mimic the events leading to the firm adhesion of immune cells on activated endothelium. The slow rolling of 300.19/E cells on the surfaces of PEG-SLeX hydrogels confirms the integration of SLeX into the system through the use of an avidin–biotin linkage, which did not disrupt the active binding site of the carbohydrate. Incorporation of the RGD and LDV peptides encouraged firm cell adhesion to the materials surface under shear in a concentration dependant manner, and cells remained adherent throughout the duration of the flow experiment. These results illustrate that this method of studying leukocyte adhesion succeeds in mimicking the interactions seen on the in vivo vascular wall under shear. With further optimization, such as incorporation of signaling molecules, this system using of PEG gels can improve insight into the mechanisms of rolling and firm cell adhesion at sites of active vascular disease. Conclusions The covalent modification of PEG hydrogel surfaces with cell adhesive peptides was accomplished in continuous layers of adhesive regions. These materials encouraged vascular cell adhesion through both transient interactions with selectin molecules and firm binding via integrins. The system presented here to expose adhesive hydrogels to cells under flow conditions represents a method to study cell–material interactions in environments that closely mimic in vivo environments.

Pflugers_Arch-3-1-2137943

A refined radio-telemetry technique to monitor right ventricle or pulmonary artery pressures in rats: a useful tool in pulmonary hypertension research

Implantable radio-telemetry methodology, allowing for continuous recording of pulmonary haemodynamics, has previously been used to assess effects of therapy on development and treatment of pulmonary hypertension. In the original procedure, rats were subjected to invasive thoracic surgery, which imposes significant stress that may disturb critical aspects of the cardiovascular system and delay recovery. In the present study, we describe and compare the original trans-thoracic approach with a new, simpler trans-diaphragm approach for catheter placement, which avoids the need for surgical invasion of the thorax. Satisfactory overall success rates up to 75% were achieved in both approaches, and right ventricular pressures and heart and respiratory rates normalised within 2 weeks. However, recovery was significantly faster in trans-diaphragm than in trans-thoracic operated animals (6.4 ± 0.5 vs 9.5 ± 1.1 days, respectively; p < 0.05). Stable right ventricular pressures were recorded for more than 4 months, and pressure changes, induced by monocrotaline or pulmonary embolisms, were readily detected. The data demonstrate that right ventricular telemetry is a practicable procedure and a useful tool in pulmonary hypertension research in rats, especially when used in combination with echocardiography. We conclude that the described trans-diaphragm approach should be considered as the method of choice, for it is less invasive and simpler to perform. Introduction Over the last decade, new medical treatments became available for the treatment of pulmonary hypertension (PH) patients, and this development set off a renewed interest in the pathophysiology and haemodynamics of the pulmonary circulation [6, 12, 22]. Experimental PH studies are still largely confined to single measurements of pulmonary arterial (PA) or right ventricular (RV) pressure in anaesthetised, open-chest animals undergoing artificial respiration, which does not allow determining changes of pressures in time. Measuring pulmonary haemodynamics have improved considerably since RV telemetry was firstly introduced by Hess et al. [4], as it allows for continuous monitoring of RV or PA pressures, without introducing artefacts caused by stress or anaesthetics [9]. Moreover, animals fitted with telemetry can serve as their own controls, thus permitting the use of smaller groups of animals than is possible with the single measurement approach in anaesthetised animals [9]. Although many investigators have embraced this new technology with enthusiasm and despite its proven potential, studies actually using this methodology in rats are scarce [4, 13, 16, 18, 19]. One possible explanation for this discrepancy is that the original procedure was only described briefly before [4], complicating its implementation for potential new users. Secondly, the described trans-thoracic approach involves RV catheterisation through complex and highly invasive thoracotomy [4]. In the present study, we hypothesised that RV catheterisation performed via an opening in the diaphragm, thereby omitting a thoracotomy, may be a simpler, less invasive alternative. In the present report, we provide a detailed description of the alternative trans-diaphragm approach together with a comparison with the original trans-thoracic approach. Furthermore, we illustrate its usefulness when studying the pulmonary circulation, by presenting novel comprehensive pulmonary and cardiac haemodynamic data during the development of monocrotaline-induced pulmonary hypertension. Materials and methods All experiments were approved by the Institutional Animal Care and Use Committee of the VU University and were conducted in accordance with the European Convention for the Protection of Vertebrate Animals used for Experimental and Other Scientific Purposes, and the Dutch Animal Experimentation Act. Animals Thirty-eight male Wistar rats were used (250–300 g; Harlan, Horst, The Netherlands), of which 33 were operated and 5 served as controls (no surgery). The animals were allowed to adapt to their new environment for at least 1 week before surgery. The animals were conventionally housed in pairs under controlled conditions (temperature 21–22°C; humidity 60–65%; 12:12 h light–dark cycle) and had free access to filtered water and standard rat chow (Global 2016, Harlan Teklad, Blackthorn-Bicester, England). Telemetry system An implantable radio-telemetry system for blood pressure measurements for small laboratory animals was used, comprising a radio-transmitter (TA11PA-C40) fitted with a 10-cm-long catheter (Data Science International [DSI], St. Paul, MN). Transmitters were turned on a day before implantation and stored in sterilised saline, following the manufacturer’s instructions (http://www.datasci.com/information/index.asp). The pressure was monitored continuously during surgery to ensure proper position of the tip of the pressure catheter. After successful implantation of the transmitter, RV or PA pressures and locomotor activity were recorded for 10 s every 5 min. From the pressure recordings mean, systolic and diastolic pressure, heart rate, respiratory rate and circadian rhythm were subsequently derived (Dataquest A.R.T. software 4.0, DSI). Pre-operative care and anaesthesia All animals received pre-surgical intramuscular buprenorphine analgesia (0.10 mg/kg; Schering-Plough, Maarssen, The Netherlands). For general anaesthesia, isoflurane (2.0% in 1:1 O2/air mix; Pharmachemie, Haarlem, The Netherlands) was used via induction chamber and tracheal intubation (16 G × 51-mm Teflon tube; ventilator settings: breathing frequency 80/min, pressures: 9/0 cmH2O, inspiratory/expiratory ratio 1:1). For extra local analgesia, lidocaine spray (100 mg/ml; AstraZeneca, Zoetermeer, The Netherlands) was applied at the surgical region. Animals were maintained under anaesthesia for an average time of 80 min. Animals were placed on a heating pad to maintain body temperature and positioned in dorsal recumbency. Hydromellose drops (0.3%; Ratiopharm, Zaandam, The Netherlands) were applied to prevent drying of the eyes. After shaving and disinfection with 70% ethanol of the chest and abdomen, animals were covered with sterile incision foil and the animals were then covered with sterile incision foil (Opraflex, Lohmann & Rauscher, Almere, The Netherlands). A surgical microscope was used (magnification ×16–64; Carl Zeiss, Sliedrecht, The Netherlands) for optimal view. Surgery: trans-thoracic approach Implantation of the telemetry transmitter in the abdomen The abdominal cavity was accessed via a 5-cm midline laparotomy, starting just below the xiphoid process. The transmitter was placed in the peritoneal cavity, with the catheter pointing caudally to prevent liver injury and tissue reaction. The abdominal cavity was covered with gauzes soaked in warm saline and left open until the end of the procedure. Routing of the pressure catheter to the right ventricle The heart was exposed by means of a left thoracotomy performed at the sixth intercostal space and mid-clavicular line. The thorax was opened with blunt scissors, respecting the anatomy of the overlying muscle layers and cautiously avoiding injury to the lungs. The pressure catheter was then tunnelled subcutaneously from the peritoneal cavity to the opening in the thorax and temporarily laid aside. Four individual small hooks were used to retract the ribs and expose the heart through a wider, 2 × 2-cm window. The pericardium was then opened with two dressing forceps. Catheterisation of the right ventricle A superficial purse string (6–0 Prolene, Ethicon, St-Stevens-Woluwe, Belgium) was placed on the right ventricle free wall, near the apex. Through the purse string, the right ventricle was punctured with a 19 G syringe needle in the direction of the RV outflow tract, avoiding any coronary vessel. After removing the needle, the site was wiped with a sterile cotton stick. The tip of the catheter was then carefully inserted into the right ventricle, using a vessel canulation forceps (0.5–1.0 mm outer diameter, Fine Science Tools, Heidelberg, Germany), avoiding loss of gel from the tip at all times. Live pressure waveform trace confirmed proper RV catheterisation (RV systolic/diastolic pressures are typically ∼25/1 mmHg). Depending on the protocol, the pressure catheter was subsequently advanced for about 2 cm more, positioning the tip of the catheter beyond the pulmonary valves. Again, live pressure waveform trace confirmed correct positioning of the pressure catheter in the pulmonary artery (PA systolic/diastolic pressures are typically ∼25/10 mmHg; Fig. 1, left panel). Finally, the catheter was fixed by the purse string and a small drop of tissue adhesive at the site of insertion (10 μl dispensed by a pipette; Vetbond, 3M, St. Paul MN; Electronic Supplementary Material (ESM) Fig. E-1a). Fig. 1left panel: Live pressure recordings while advancing the pressurecatheter into the pulmonary artery. Notice the stepwise increase in diastolic pressures when manoeuvring the catheter beyond the pulmonary valves (arrow), whereas the systolic pressures remained unchanged. right panel: Live pressure recordings while injecting a bolus injection of microspheres (arrow; 19 μm, 1.5 million/100 g i.v.). Upon embolisation, a significant increase in systolic, diastolic and developed RV pressures was observed. Furthermore, in both recordings, the effect of breathing is noticeable Closing of the chest and abdomen After the catheterisation, blood clots if any were removed from the pleural cavity with a moist cotton stick. The ribs were approximated with two individual sutures (5–0 Vicryl, Ethicon). To promote full expansion of the lungs, a small burst of positive pressure was applied (maximally 10 cmH2O for 1 s). A chest tube (18 G), used as air outlet, was removed after proper thorax excursions were observed. The chest muscles were laid back in layers (no suturing necessary). Then, the transmitter was sutured to the abdominal muscle (5–0 Perma-Hand, Ethicon) incorporating the suture rib of the device. The abdominal wall was sutured in a running subcuticular pattern (5–0 Vicryl). Finally, the skin of the chest and abdomen was closed with a running suture (5–0 Vicryl). Surgery: trans-diaphragm approach Opening of abdomen and the diaphragm The abdominal cavity was accessed via a 5-cm midline laparotomy, with the abdominal wall retracted by two individual hooks. To expose the diaphragm, a retraction suture was used to lift the xiphoid process during the whole procedure. The visceral organs were covered temporarily with wet gazes, and the liver was gently pushed down by the weight of a blade holder. A small midline incision in the diaphragm was then made, with small sharp scissors, from the xiphoid process until the tendinous part of the diaphragmatic membrane. The two sides of the diaphragm were held aside with two retraction sutures, providing optimal exposure of the heart (ESM Fig. E-1b). Catheterisation of the right ventricle RV catheterisation was performed similarly, as described above. In short, the pericardium was opened by two dressing forceps, and a purse string was placed just right of the apex. The pressure catheter was then inserted into the right ventricle through a small puncture made with a syringe needle and fixed in place. Closing the diaphragm and abdomen After catheterisation, blood clots in the thorax, if any, were removed. The diaphragm was closed with a running suture (5–0 Vicryl), starting from the ventral side, with the pressure catheter sticking out the diaphragm at the dorsal end of the incision. After that, the retraction suture through the xiphoid process was removed. To promote full expansion of the lungs, a small burst of positive pressure was applied, as described above, with the chest tube between the sutures of the diaphragm, until proper thorax excursions were visually confirmed. To further secure closure of the diaphragm, a small amount of tissue adhesive (Vetbond, 3M) was applied. Finally, the transmitter was implanted in the peritoneal cavity, fixed to the abdominal wall with the catheter facing caudally, and the abdomen was closed, as described above. Post-operative care To compensate for loss of fluids, all animals received warm sterile saline at the end of the procedure (5 ml intraperitoneal). After final skin closure, the animals were allowed to regain consciousness. In the first 24 h of recovery, the animals were housed in individual cages. Each cage was placed halfway on a heating pad, in such a way that half of the cage was maintained above it and the other at room temperature. During this period, the animals were monitored three times, received post-surgical analgesia when clinically indicated (buprenorphine 0.10 mg/kg subcutaneous) and were provided with drinking gel pads and softened rat chow. After 24 h, the animals were conventionally housed in pairs for a full recovery, and they were inspected and weighted daily. Animals were considered fully recovered from surgery when their appearance and behaviour were normal, their surgical wounds healed, their pre-surgical body weight regained, when the pressures normalised and circadian rhythm was restored. The study period was ended after 4 months. Experimental protocol To induce acute and chronic pulmonary hypertension, four animal received an intravenous bolus injection of microspheres [25] (tail; 1.5 million /100 g body mass in sterile saline; polystyrene, mean diameter 19 μm (#7520), Duke Scientific, Fremont, CA) and three animals monocrotaline subcutaneous [24] (MCT; 80 mg/kg in sterile saline; Sigma-Aldrich, Zwijndrecht, The Netherlands), respectively. Experimental interventions were performed no earlier then 2 weeks after the surgery. RV echocardiography After a 2-week recovery, trans-thoracic echocardiographic measurements (ProSound SSD-4000 system equipped with a 13-Mhz linear transducer [UST-5542], Aloka, Tokyo, Japan) were performed on anaesthetised but spontaneously breathing rats (isoflurane 2.0% in 1:1 O2/air mix), according to the standards of the American Society of Echocardiography [10, 15]. To minimise the effects of isoflurane on cardiac function, the time under anaesthesia was set to a maximum of 15 min [14]. The measurements were compared with untreated rats to check for possible adverse effects related to the procedure. Cardiac output, tricuspid annular plane systolic excursion (TAPSE), RV wall thickness (RVWT) and RV end diastolic diameter (RVEDD) were measured, as described before by Hardziyenka et al. [3]. TAPSE is a parameter for RV systolic function, which is measured in the apical four-chamber view of the heart and expresses the displacement of the lateral tricuspid annulus towards the apex during systole [10]. Estimation of pulmonary vascular resistance and RV power output Additional echocardiographic measurements were performed in MCT-treated rats and untreated but operated controls, just before the injection, 2 weeks after injection and when the first clinical signs of RV heart failure developed (as indicated by weight loss, dyspnoea and lethargy, after which the rats were euthanised). Pulmonary vascular resistance (PVR) and RV power output (RV-power) were estimated, by combining telemetric pressure data and echocardiographic flow data. PVR was estimated by Poiseuille’s law: [PVR] ≈ [mean PAP]/[cardiac output] ≈ (0.61 × [systolic RV pressure] + 2 mmHg)/[cardiac output] [1, 23]. RV-power was estimated using a simplified pressure–volume analysis, assuming rectangular pressure–volume loops: [RV-power] ≈ [systolic RV pressure] × [stroke volume] × [heart rate] = [systolic RV pressure] × [cardiac output] [21, 23]. Autopsy To investigate potential effects of chronic implantation of the telemetry transmitters, the animals were euthanised at the end of the study protocol by exsanguination under isoflurane (4.0% in 1:1 O2/air mix). Macroscopic examination of right ventricle, lungs, diaphragm and intercostal muscles were performed, and organ weights (heart, lungs, liver, spleen, kidneys) were measured and compared with controls. Animals treated with MCT or microspheres were excluded from these analyses. Statistical analyses All measurements, unless otherwise stated, are presented as mean ± standard error of the mean. Normal distribution was verified, and Student’s t test, analysis of variance or Fisher Exact test were used, when appropriate (SPSS 13.0, SPSS, Chicago IL). A p value less than 0.05 was considered statistically significant. Results Success rate Seventeen animals were operated using the trans-thoracic approach, and 16 animals were operated using the trans-diaphragm approach. In the trans-thoracic group, 8 animals showed stable pressure signals for the whole study period of 4 months. Five animals did not recover form surgery, and 4 animals developed instable pressure signals, caused by clot formation inside the pressure catheter within 2 weeks of recovery. Of the 9 failed procedures, 4 occured in the first four attempts. In the trans-diaphragm group, 9 animals showed stable pressure signal for the whole study period of 4 months. Six animals did not recover form surgery, and one animal developed instable pressure signals, caused by clot formation inside the pressure catheter within 2 weeks of recovery. As in the trans-thoracic group, 4 of the failures occurred in the first four attempts. By discarding the first 4 animals in each procedure, as we consider this the learning curve, the overall success rate was 8 of 13 (62%) for the trans-thoracic approach and 9 of 12 (75%) for the trans-diaphragm approach. Recovery from surgery Within a week post-surgery, heart and respiratory rates and circadian rhythm normalised in all rats, and no differences were observed between the two approaches (Table 1). Although all animals regained their pre-surgical body weight within 2 weeks, full recovery was significantly faster in the trans-diaphragm- than the trans-thoracic-operated animals (6.4 ± 0.5 vs 9.5 ± 1.1 days, respectively; p < 0.05). Table 1Main results of this study, all values in mean ± SEM Trans-thoracic approach (n = 17)Trans-diaphragm approach (n = 16)Controls (n = 5)Success ratea8/13 (62%)7/9 (79%)n.r.Recovery BW to pre-surgical level (days)d9.5 ± 1.16.4 ± 0.5*n.r. Heart rate (bpm)e362 ± 4.4359 ± 5.7350–400b [8, 17] Respiratory rate (rpm)e90 ± 2.491 ± 3.880–100b [11, 17, 20] Circadian rhythm (days)e3.4 ± 0.43.7 ± 0.6n.r.Pressurese RVSP (mmHg)25 ± 1.124 ± 0.921–26b [16] RVDP (mmHg)1.7 ± 0.32.2 ± 0.21–5c [5]RV echocardiographyd Cardiac output (ml/min)107 ± 4.9116 ± 6.9110 ± 4.9 TAPSE (mm)3.6 ± 0.13.7 ± 0.23.4 ± 0.1 RVWT (mm)1.0 ± 0.11.0 ± 0.10.9 ± 0.1 RVEDD (mm)3.6 ± 0.13.7 ± 0.13.6 ± 0.1Autopsyf, organ weights (g) Heart1.4 ± 0.11.4 ± 0.21.3 ± 0.1 Lungs1.5 ± 0.11.5 ± 0.21.5 ± 0.2 Liver15.4 ± 0.515.0 ± 0.615.1 ± 0.5 Spleen0.65 ± 0.030.68 ± 0.040.64 ± 0.04 Kidneys2.4 ± 0.22.3 ± 0.22.3 ± 0.1n Number of animals, n.r. not relevant, BM body mass, RVSP RV systolic pressure, RVDP RV diastolic pressure, TAPSE tricuspid annular plane systolic excursion, RVWT RV wall thickness, RVEDD RV end diastolic diameter*p < 0.05aFirst four attempts were discarded.bNormal values for Wistar rats, awake and at rest, measured by radio-telemetrycNormal values for Wistar rats, measured during acute pressure measurements under anaesthesiadAnalyses on surviving animals onlyeAnalyses on successfully operated animals only (stable signal)fAnalyses on surviving and untreated animals only (no microspheres or monocrotaline) RV pressures normalised in all animals within 2 weeks after surgery (RV systolic/diastolic pressures ∼25/1 mmHg), irrespective of the method used (Table 1). PA systolic/diastolic pressures were estimated to be ∼25/10 mmHg (Fig. 1, left panel). Activity had no significant effect on the stability of the pressure signals. Echocardiography revealed normal cardiac output (∼110 ml/min), TAPSE (∼3.5 mm), RVWT (∼1.0 mm) and RVEDD (∼3.6 mm) in all rats, and no differences were seen between the differently operated animals and controls (Table 1). Furthermore, other signs of RV dysfunction, like pericardial effusion and tricuspid regurgitation, were not present in any animal. Acute pressure effects of pulmonary embolisation by microspheres RV telemetry could readily detect acute changes in RV pressure, induced by a bolus injection of microspheres (Fig. 1, right panel). Upon embolisation, systolic RV pressures increased from 21 ± 1 to 38 ± 3 mmHg (p < 0.01) and diastolic RV pressures increased from 2 ± 1 to 7 ± 2 mmHg (p < 0.05). Haemodynamic changes during the development of MCT-induced pulmonary hypertension Changes in haemodynamic parameters during the development of MCT-induced PH are shown in Fig. 2 (also in ESM Fig. E-2). Four weeks after MCT injection, the first clinical signs of RV heart failure developed. During this period, there was a gradual rise in RV systolic pressure, from 22 ± 2 mmHg at baseline to 38 ± 4 mmHg at 2 weeks and 72 ± 5 mmHg at 4 weeks (p < 0.01 vs control). At 4 weeks, cardiac output dropped dramatically, from 100 ± 7 ml/min at baseline to 107 ± 8 at 2 weeks and 52 ± 4 ml/min at 4 weeks (p < 0.01 vs control). In addition, the poor haemodynamic condition of the MCT-treated rats at 4 weeks was confirmed by a simultaneous decline in TAPSE from 3.6 ± 0.1 mm during the first 2 weeks to 1.8 ± 0.2 mm at 4 weeks (p < 0.01 vs control). During the development of PH, the right ventricle remodelled initially by increasing RVWT (1.0 ± 0.1 mm at baseline, 1.1 ± 0.1 mm at 2 weeks and 1.3 ± 0.1 mm at 4 weeks, respectively; p < 0.01 vs control) but subsequently by increasing RVEDDs as well (3.5 ± 0.1 mm at baseline, 3.6 ± 0.1 mm at 2 weeks and 7.0 ± 0.5 mm at 4 weeks, respectively; p < 0.01 vs control). Fig. 2Haemodynamic changes during the development of MCT-induced pulmonary hypertension. Four weeks after the rats received MCT, clinical signs of RV heart failure developed. In these 4 weeks, systolic pressures continued to rise, while cardiac output could not be maintained. The presence of RV remodelling was indicated by increased wall thickness and diameter. The pulmonary vascular resistance continued to rise as well, while at 4 weeks, the increase in the RV-power was inadequate to compensate for the dramatically increased afterload. Triangles connected by dashed lines: MCT-treated; squares connected by uninterrupted lines: control group; n = 3 for both groups. All values are in mean ± SEM, Asterisk, p < 0.05; double asterisk, p < 0.01 compared to control (post-hoc analyses). Please note that sometimes the icons conceal the error bars. For explanation of the abbreviations, see corresponding text During the development of PH, PVR increased continuously, from 1.2 ± 0.2 × 109 N m−5 s at baseline, to 1.9 ± 0.2 × 109 N m−5 s at 2 weeks and 7.9 ± 0.9 × 109 N m−5 s at 4 weeks (p < 0.01 vs control). RV-power was increased at 2 weeks, but no further increase was seen at 4 weeks (RV-power = 4.9 ± 0.7 mW at baseline, 9.1 ± 0.9 mW at 2 weeks, 8.3 ± 0.7 mW at 4 weeks, respectively; p < 0.01 vs controls). Autopsy Chronic implantation of the telemetry transmitter in both approaches did not result in any differences in organ weight as compared to control (Table 1), and no apparent pulmonary embolisms were seen. Autopsy revealed only local fibrosis of the right ventricle near the insertion point of the catheter (ESM Fig. E-3a). In the trans-thoracic approach, animals showed normal healing of the intercostal muscles. In the trans-diaphragm approach, no herniation or dehiscence of the diaphragm were seen in any animal, and often, the liver was partially accreted to the healed wound of the diaphragm (ESM Fig. E-3b). Discussion The present study describes in detail two surgical techniques to monitor RV or PA pressures over time by radio-telemetry in rats. To our best knowledge, we are the first research group to describe the trans-diaphragm approach and its comparison with the previously used trans-thoracic approach. We have demonstrated that: (1) in our hands, both the trans-diaphragm as well as the trans-thoracic approach have satisfactory success rates, especially when considering the complexity of the procedures, (2) the time-to-recovery was significantly shorter with the trans-diaphragm approach than the trans-thoracic approach, (3) measured physiological parameters recovered fully in both methods, and there were no permanent detrimental effects on RV, intercostal muscles or diaphragm, (4) using RV-telemetry, acute and chronic pressure changes in the pulmonary circulation can be readily detected and (5) RV-telemetry, in combination with echocardiography, allow thorough monitoring of pulmonary and cardiac haemodynamic changes over time. Fully recovered animals showed normal RV structure, as expressed by RVWT and RVEDD, and normal function, as expressed by cardiac output and TAPSE. In addition, in both approaches, RV pressure normalised shortly after surgery, and autopsy revealed only minor fibrosis in the direct proximity of the catheter. Therefore, permanent detrimental effects of RV-telemetry on the RV can be excluded for both techniques. In addition, examination of other organs revealed no abnormalities, as was described by others [4]. In animals operated using the trans-diaphragm approach, macroscopic examination revealed normal wound healing of the diaphragm. Although we have not tested the strength of the healed diaphragm, others did not find significant differences between a normal (untouched) and a sutured diaphragm [7]. The natural history of small defects in the diaphragm is relatively benign, as they almost always heal spontaneously [2], especially when the diaphragm is fixated by other organs [26]. Furthermore, the observation of a normal respiratory rate seen in the recovered animals confirms that the trans-diaphragm approach is safe and a good alternative for the trans-thoracic approach. In our experience, insertion of the pressure catheter into the RV and its advance further into the pulmonary artery when desired is comparatively easy to perform in the trans-diaphragm approach. This relative ease of RV catheterisation is reflected in the higher success rate compared to the trans-thoracic approach and might be well explained by the more caudal and inferior direction by which the catheter enters the heart. Approaching the heart in this manner entails minimal manipulation and manoeuvring of the catheter during the process of RV catheterisation, resulting in a lower incidence of clot formation inside the catheter. This complication occurred more frequently in trans-thoracic-operated animals as a direct consequence of excessive manoeuvring and accidental squeezing of the catheter causing it to lose or displace some of its anti-thrombolitic gel. The strongest advocate of the trans-diaphragm approach over the trans-thoracic approach, however, is the significantly faster recovery of the animals. Prompt recovery from surgery is particularly important in rats with MCT-induced PH because of low survival rates beyond 4 weeks after MCT injection. In this study, we used echocardiography to validate RV-telemetry in vivo. Additionally, through the combination of pressure data, obtained by RV-telemetry, with echo-Doppler flow data, we were able to study the haemodynamics of the pulmonary circulation in greater detail [23]. MCT induced pulmonary and cardiac remodelling, at the same time as PVR, RVWT and RVEDD increased. Cardiac adaptation to overcome the increasing afterload became inadequate by the fourth week of MCT injection, in such that there was no further increase in RV-power to maintain adequate cardiac output while RV systolic pressure continued to rise. Detailed studies of these observations are planned for the near future. In the present study, we have focussed on the refinement of the RV-telemetry technique in rats, but this does not exclude its applicability in mice and other rodents. Indeed, in their recent study, Schwenke et al. [18] reported long-term monitoring of PA pressure in mice using radio-telemetry. In conclusion, we described a new, easier-to-perform, mildly invasive trans-diaphragm-based RV-telemetry approach for long-term monitoring of PA and RV pressures in the rat model of MCT-induced PH. Our findings may be applied to improve our understanding of the disease processes involved in PH and develop better treatment strategies for the disease. Electronic supplementary material Below is the link to the electronic supplementary material. Electronic Supplementary Material M.L. Handoko et al. A refined radio-telemetry technique to monitor right ventricle or pulmonary artery pressures in rats. Pflugers Arcg – Eur J Physiol 2007. Fig. E-1, E-2, E-3. (PDF 292 kb)

J_Comp_Physiol_A_Neuroethol_Sens_Neural_Behav_Physiol-4-1-2248214

Far field scattering pattern of differently structured butterfly scales

The angular and spectral reflectance of single scales of five different butterfly species was measured and related to the scale anatomy. The scales of the pierids Pieris rapae and Delias nigrina scatter white light randomly, in close agreement with Lambert’s cosine law, which can be well understood from the randomly organized beads on the scale crossribs. The reflectance of the iridescent blue scales of Morpho aega is determined by multilayer structures in the scale ridges, causing diffraction in approximately a plane. The purple scales in the dorsal wing tips of the male Colotis regina act similarly as the Morpho scale in the blue, due to multilayers in the ridges, but the scattering in the red occurs as in the Pieris scale, because the scales contain beads with pigment that does not absorb in the red wavelength range. The green–yellow scales of Urania fulgens backscatter light in a narrow spatial angle, because of a multilayer structure in the scale body. Introduction Many butterflies have vivid, colourful wing patterns, created by rows of partly overlapping scales with intricate spatial structures that scatter incident light. The scattered light can interfere coherently or incoherently. When the scale structures have spatial periodicity, the interference of coherent light waves often results in striking iridescences. The displayed colours are then called structural or physical. Without periodicity, light scattering is incoherent or random, resulting in white scales, unless they contain pigment that selectively absorbs in a certain wavelength range. In the latter case a pigmentary or chemical colour results (Fox and Vevers 1960; Vukusic and Sambles 2003; Kinoshita and Yoshioka 2005). Many butterfly species feature structural as well as pigmentary colours (Rutowski et al. 2005; Yoshioka and Kinoshita 2006a). Each butterfly wing scale is the cuticular product of a single cell, with a rather flat, unstructured lower scale lamina and a highly structured upper lamina, typically consisting of longitudinal ridges connected by crossribs (Ghiradella 1998). The fine structure of butterfly scales is highly variable (Vukusic et al. 2000). For instance, the crossribs of the white scales of pierids are adorned with granules (Yagi 1954; Giraldo and Stavenga 2007; Morehouse et al. 2007), the ridges of Morpho scales and many male Pieridae are elaborated into multilayer structures (Ghiradella et al. 1972; Vukusic et al. 1999; Kinoshita et al. 2002), and many papilionids and lycaenids have scales with photonic crystal properties (Vukusic et al. 2002; Vukusic and Sambles 2003; Kertesz et al. 2006). The scales reflect only part of the incident light flux, and the remaining part is transmitted unless it is absorbed by pigment. Because the scales are arranged in layers on both sides of the wing, incident light suffers reflection and transmission in each layer of the scale stacks, and therefore the wing reflectance is not solely due to backscattering by the top layer, the cover scales, but it is the cumulative effect of the scale stacks on both wing sides. Yoshioka and Kinoshita (2006b) investigated this phenomenon in Morpho wings, and to explain the reflectance spectra of intact wings from the spectra of individual scales they used a simplified scale stack model, assuming that normally incident light rays are also reflected and transmitted normally. Stavenga et al. (2006) developed a more general model for the reflectance of butterfly wings to interpret reflectance measurements performed on intact as well as partly or completely denuded wings of the small white butterfly, Pieris rapae. The basic assumption of the latter model, that the scales scatter randomly, was however not specifically demonstrated. Knowledge of the spatial and spectral reflectance characteristics of single scales is essential to further develop our understanding of the coloration principles applied by butterflies. In the present study we investigate the scattering properties of wing scales of a variety of butterflies, and we correlate the scattering diagrams with the electron micrographs of the scale structure. We show that white scales of pierid butterflies are approximately random scatterers. Iridescent scales, with multilayer structures, exhibit directional reflection. Whereas some butterfly species appear to have scales with dominant iridescence, other species combine iridescence and scattering properties. Materials and methods Animals We investigated the scales of a variety of butterflies. The small white, P. rapae, was obtained from a continuous culture maintained by Dr J. J. van Loon, Entomology Department, University of Wageningen (the Netherlands). The black jezebel, Delias nigrina, was captured near Bateman’s Bay, Australia. The Morpho aega was purchased. The purple tip, Colotis regina, was received from the butterfly collection of the Royal Museum for Central Africa, Brussels (Belgium; curator Dr U. Dall’Asta). The moth Urania fulgens was a gift from Dr Marta Wolff, Entomology group, University of Antioquia (Medellín, Colombia). Angular distribution of scattering by single scales Single scales were isolated by gently pressing the wing of a butterfly on a glass plate. Subsequently, an isolated scale was glued to the tip of a glass micropipette (tip diameter ca. 5 μm). The micropipette was then mounted on a micromanipulator in the optical setup shown in Fig. 1a. The scale was always hanging with the longitudinal ridges of the scale oriented vertically. Light from a xenon lamp was focused on a pinhole with diameter 50 μm, which was subsequently imaged on the scale. The light beam, which had an aperture of <10°, passed a small hole in a screen before it reached the scale. The scale reflected (that is, back-scattered) part of the incident light. The angular spread of the reflected light was documented by photographing the light distribution at the white backside of the screen (Fig. 1a). We used a Nikon Coolpix 990 (Fig. 2) and an Olympus DP70 (Fig. 6). The digital images were processed with MatLab (Fig. 6). Fig. 1Diagram of the optical system used for measuring the angular distribution of the scale reflectance. Light from a light source was focused on a pinhole, which was imaged on the scale. a A white screen with a small hole was placed in between the imaging lens and the scale. The light reflected by the scale caused a light pattern on the screen, which was photographed. b The light reflected by the scale was collected by a lightguide, which relayed the captured light to a spectrometer. The lightguide was mounted on a motorized stage rotating in the horizontal plane (top view drawing) Fig. 2Angular and spectral reflectance of isolated butterfly scales. a–d White scale of the dorsal forewing of a male Pieris rapae. e–h White scale from the dorsal forewing of a male black jezebel, Delias nigrina. i–l Blue scale of a Morpho aega. m–p Purple scale from the tip of the dorsal forewing of a male purple tip, Colotis regina. q–t Orange–white scale of Urania fulgens. First column: scanning electron microscopy of scale cross-sections made by cutting pieces of wing with a razor blade. Bars 1 μm. Second column: photographs of the angular distribution of light reflected by a scale and subsequently reflected by a white screen, where the scale was illuminated with a narrow beam of white light (see Fig. 1a). The scales were glued at the tip of a glass micropipette, which was suspended from above on a micromanipulator; see the black shadow in the upper part of b and f. The bright spot at the right of the central black hole is due to light forward scattered by the scale. The ellipses in b (due to the slightly oblique position of the camera) represent directional angles in steps of 10°, from 10° to 60°, as well as that for 65°. The hole in the screen is ca. 8°. Third column: angular dependence of the reflectance in the horizontal plane, measured with the setup of Fig. 1b. Fourth column: reflectance spectra of single scales measured with a microspectrophotometer Spectrophotometry The angular dependence of the light scattering by the scale was measured with a lightguide, which was mounted on a rotating motor and connected to a spectrometer (Yoshioka and Kinoshita 2006b). The scales were adjusted so that the scale plane was perpendicular to the light beam, as judged by the symmetrical reflection pattern with respect to the axial direction. The experiments were run using a Labview interface, which allowed the measurement of reflectance spectra in angular steps of 1° over a 180° angle (Fig. 1b). Angular reflectance curves were calculated for a series of wavelengths with 10 nm interval by sequentially averaging the reflectance spectra over 10 nm wavelength ranges. In addition, the spectral reflectances of single scales were measured with a microspectrophotometer, consisting of a xenon light source, a Leitz Ortholux microscope, and a fiber optic spectrometer. The microscope objective was an Olympus 20×, NA 0.46. A white reflectance standard (Spectralon, Labsphere, North Sutton, NH, USA) served as the reference in all cases. Electron microscopy Subsequent to the microspectrophotometry, the single scales were prepared for scanning electron microscopy (SEM). Additionally, pieces of wing were cut and put on the SEM-holder in different positions to visualize the upper surface as well as cross-section of the scales. Samples were sputtered with palladium for 5 min with 800 V and 200 mTorr (Hummer, Alexandria, VA, USA). A Philips XL-30 scanning electron microscope with a voltage of 3 kV was used to investigate the scale anatomy. For transmission electron microscopy, wing pieces were prepared as usual. In brief, samples were embedded in agar for better handling, prefixed in 2% glutaraldehyde in 0.1 M sodium cacodylate buffer, and postfixed in 1% OsO4/1.5% K4 Fe(CN)6 in 0.1 M cacodylate. Subsequent washing with double-distilled water and dehydration with an alcohol series that ended with 100%, were followed by propylene oxide during 30 min and embedding in Epon. Post-microtomed samples were contrasted with uranyl acetate in methanol for 2 min and lead in water for 1 min, and were then examined with a Philips 201 transmission electron microscope. Results The white scales of the small white, P. rapae, are marked by ovoid beads that adorn the crossribs (Fig. 2a). The angular light scattering of a single scale can be readily visualized with the setup of Fig. 1a, where a beam of white light is focused on a scale via a small hole in a screen. Figure 2b is a photograph of the screen, showing the angular distribution of the light reflected by a white scale taken from the dorsal wing of a male small white, P. rapae. The scale scatters light approximately circular-symmetrically, suggesting that the scale acts as a diffuse scatterer. The angular distribution of the scattering was quantitatively investigated with the setup of Fig. 1b, where the light reflected by the scale is measured as a function of angle in the horizontal plane. Figure 2c presents the reflectance of the scale of Fig. 2b as a function of angle for a number of wavelengths, normalized to the maximal reflectance value; the angle is 0° for the normal to the scale. The scale’s reflectance spectrum for normally incident light is given in Fig. 2d. The reflectance is high in the visible wavelength range, but it is low in the ultraviolet, because of an ultraviolet-absorbing pigment, presumably leucopterin (Wijnen et al. 2007), which is concentrated in pigment granules, the ovoid beads (Fig. 2a). The beads act as strong scatterers at wavelengths where the pigment absorption is negligible (Stavenga et al. 2004; Giraldo and Stavenga 2007; Morehouse et al. 2007). In addition to the beads, the other structures of the scale, that is the ridges and crossribs of the upper lamina of the scale as well as the lower lamina, contribute to the scattering (Fig. 2c). Figure 2e–h presents a similar set of data for a white scale of the dorsal forewing of a male black jezebel, D. nigrina. This case is obviously very similar to that of the white P. rapae scale. The scale anatomy with crossribs studded with beads is very similar (Fig. 2e), the scattering is again approximately random (Fig. 2f, g), and the reflectance spectrum is also high in the visible and low in the UV (Fig. 2h). The strikingly blue M. aega has scales where the lamellae of the ridges form multilayers (Fig. 2i). Different from Morpho didius, for example, which has glass cover scales and strongly pigmented ground scales (Vukusic et al. 1999), cover and ground scales of M. aega cannot be distinguished. The scattering diagram of a scale of a M. aega is a horizontal stripe (Fig. 2j), perpendicular to the vertically oriented scale ridges. The angular dependence of the reflectance varies strongly with the wavelength (Fig. 2k), and the reflectance measured with normally incident light features a distinct peak in the blue (Fig. 2l), which is due to the multilayered structure of the scale ridges (Fig. 2i; see also Vukusic et al. 1999; Kinoshita et al. 2002; Yoshioka and Kinoshita 2004). A reflectance peak value of more than two results, because the scale’s scattering is highly directional and the reflectance was measured relative to the diffusely scattering white standard (Fig. 2l). The purple scales at the dorsal wing tips of the male purple tip, C. regina, have ridges that are fine-structured similarly as in the case of M. aega (Fig. 2m). The purple scale features a scattering diagram with a blue stripe and a red diffuse pattern (Fig. 2n), which is a mixture of the stripe phenomenon of Fig. 2j and the diffuse patterns of Fig. 2b, f. The blue stripe is due to light backscattered by the fine-structured ridges, and the red diffuse pattern results from randomly scattered light, filtered by a pigment contained in granules below the multilayered ridges (Fig. 2m). The blue peaking reflectance spectrum, shown in Fig. 2p, is mainly due to reflection by the ridges, and the red band, above 600 nm, is mainly caused by the light scattering pigment granules (Fig. 2p). A green–yellow reflecting scale of the moth U. fulgens has between the upper and lower scale laminae (Fig. 2q) an elaborate multilayer system, which yields a spatially restricted, directional scattering diagram (Fig. 2r, s). The high amplitude of three of its reflectance spectrum is again due to the directionality of the scale reflectance (Fig. 2t). The reflectance spectrum features a distinct band, peaking at 590 nm, with halfwidth about 120 nm, indicating that an interference reflector is involved. To evaluate whether the beaded scales of P. rapae, D. nigrina and C. regina act as diffuse scatterers, the angular reflectances of these species were normalized and compared with a Lambertian reflector (Fig. 3). The wavelengths selected are outside the absorption bands of the pigments, which means that the wavelengths chosen for P. rapae and D. nigrina are in the visible spectrum, but for C. regina only in the red wavelength range. The white scale of D. nigrina (Fig. 3b) well approximates a Lambertian reflector. The angular distribution patterns of the white scale of P. rapae slightly deviate from the ideal curve (Fig. 3a), but notably the purple scale of C. regina is not a perfectly diffused red scatterer. Presumably this is due to the highly structured, multilayered ridges. Fig. 3Normalized angular reflectances measured from the scales of the three pierid butterflies, P. rapae, D. nigrina, and C. regina, compared to a Lambertian reflector (bold line). Angular reflectance curves are shown in the wavelength range where the reflectance is high, that is, where pigment absorption is low. Four wavelengths covering the visible range are shown for P. rapae (a) and D. nigrina (b). The scattering diagrams of D. nigrina well approximate a Lambertian reflector, but the angular dependence of the scattering by C. regina shows regular peaks, which indicates that the light scattered by the pigment granules is affected by the multilayered ridges Discussion The relationship between the optical properties of butterfly scales and their structure has been the topic of several studies (e.g., Vukusic et al. 1999, 2002; Kinoshita et al. 2002). Most of the previous investigations have focused on iridescent scales. Here, we have compared the reflection pattern of five differently structured single butterfly scales that scatter light coherently or incoherently, or both. We started with a simple white scale common to many species of the pierid subfamily Pierinae. Due to the characteristic beaded structure light scattering is strong, thus causing the intense white colour. We find that the white pierid scales approximate the properties of a Lambertian diffuser, at least in the wavelength range where pigment absorption is negligible (Fig. 3). The ridges of the scales of Morpho butterflies cause a blue colour. Melanin pigment below the multilayered ridges absorbs stray light over the whole visible wavelength range, including the ultraviolet, thus supporting the strikingly blue wing colour (Yoshioka and Kinoshita 2006a). The pigment of the scales in the dorsal wing tips of C. regina also absorbs stray light, but not in the long-wavelength range. The remaining red light together with the blue iridescence causes the purple colour. A similar, however, short-wavelength-shifted case is formed by the yellow scales of the dorsal wings of many male butterflies of the pierid subfamily Coliadinae, where a UV and blue absorbing pigment is combined with UV iridescence (Ghiradella et al. 1972; Rutowski et al. 2005). Figure 4a is a transmission electron micrograph of a yellow cover scale from the dorsal wing of a male brimstone, Gonepteryx rhamni. The UV iridescent, yellow scales of the male brimstone have numerous beads, which contain the yellow pigment xanthopterin (Wijnen et al. 2007). The beads are seen in Fig. 4a as empty ovoids, but this is presumably due to the procedures for transmission electron microscopy (Morehouse et al. 2007). Fig. 4Transmission electron microscopy images of a cover scale of the dorsal wing of a male brimstone, Gonopteryx rhamni (a), and of a scale of Urania fulgens (b). Ridges are indicated by closed arrows. The brimstone scale has multilayers restricted to the ridges. The space between ridges and crossribs is studded with pigment granules (open arrow). U. fulgens has continuous multilayers connected by numerous pillars that result in spreading of the scattering. Bar: 1 μm The scattering diagram of U. fulgens is not perfectly directional (Figs. 2r, s), which should have been the case when the scales consisted of an ideal multilayer. Transmission electron microscopy shows that between the multilayers exist pillars (Fig. 4b), which presumably cause the spread in the scattering diagram. Of course, the scales—and therefore the multilayers—are also not perfectly flat. The ridges (see Figs. 2q, 4b) will further contribute to some diffuse scattering. Knowing how light is scattered by individual scales is necessary for a proper understanding of butterfly wing coloration. As has been previously demonstrated, wing reflectance is a cumulative effect due to multiple reflection and transmission by the layers of scales on the butterfly wing (Yoshioka and Kinoshita 2006b; Stavenga et al. 2006). The scales are often arranged in quite regular rows and have approximately the same orientation with respect to the wing surface, but of course there are deviations from perfect order. This will be rather irrelevant in the case of the small white, P. rapae, where scattering is approximately diffuse, but in cases where scattering is highly directional, as in Morpho butterflies, disordered scale arrangement will result in broadening of the scattering diagrams. This is illustrated in Fig. 5, where scattering by a single scale of M. aega (Fig. 5a) is compared with scattering by an array of scales on the wing (Fig. 5c). The far-field scattering diagram of a single scale is a narrow stripe (Fig. 5b; see also Fig. 2j), but that of a set of scales is a distinctly broadened stripe (Fig. 5d). The angular extent of the scattering by the single scale in the vertical plane is <10°, but that of the scale set is about 30° (Y, Fig. 6). The broadening of about 20° will be due to a rotational variation of the scale plane around the long axis of the scales. The tilt angle, that is the angle of the scale plane with the wing, will have a similar variation, as the scattering diagram of the scale set is about 20° wider than that of the single scale (X, Fig. 6). Further broadening of the scattering diagram will occur when reflection by the complete wings of a Morpho butterfly is considered. The dynamic changes of light scattering by a Morpho flying in a natural environment, and how the butterflies will be perceived by conspecifics and predators will be interesting themes for future research. Fig. 5Scattering by scales of Morpho aega. a Single scale about normally illuminated by a beam with about 50 μm diameter; the outline of the scale is visible because of an additional, obliquely illuminating beam (bar: 50 μm). b Far field scattering pattern of the single scale photographed at a white screen (see Fig. 2j). The bright white spot is due to light scattered at the hole in the white screen and direct scattering by the scale. The interrupted circles indicate the angular directions of the scattering in steps of 10°. c A wing piece about normally illuminated by a beam with about 900 μm diameter; cover and ground scales cannot be distinguished (bar: 200 μm). d Far field scattering pattern of the wing piece. The white spot is due to light scattered at the hole in the white screen and to direct scattering by the scaleFig. 6Spatial profiles of the scattering diagrams of Fig. 5b, d. The extent of the scattering in the horizontal plane (X) is much broader than the scattering in the vertical plane (Y) for both the single scale as for the intact wing piece. The angular width in the vertical plane of the scattering by the single scale is <10°, but the width in the horizontal plane is about 100°. The vertical angular extent of the scattering by the wing piece is about 30°, meaning a broadening of about 20° with respect to the single scale, and the extent in the horizontal plane is similarly broadened

BMC_Bioinformatics-4-_-521197

EasyGene – a prokaryotic gene finder that ranks ORFs by statistical significance

Background Contrary to other areas of sequence analysis, a measure of statistical significance of a putative gene has not been devised to help in discriminating real genes from the masses of random Open Reading Frames (ORFs) in prokaryotic genomes. Therefore, many genomes have too many short ORFs annotated as genes. Background As of February 2003, 106 microbial genomes have been sequenced and made publicly available and the race is now on to mine genomes such as these for interesting and/or valuable genes and motifs. It has been estimated [1] that 60–80% of the genes in newly sequenced organisms have known homologues in other species. This percentage will grow as genomic annotations progress and perhaps there will be a time when virtually all genes can be found by homology matches to known proteins. That day, however, is not around the corner and even if it were, the occasional odd genes which would nevertheless escape detection by homology may very well be the truely novel and most wanted ones. It is a common misconception that identification of genes in prokaryotes is almost trivial. Any random sequence, as well as non-coding regions in real genomes, contain a large number of open reading frames (ORFs). Most of these are too short to be possible protein coding genes, but in many genomes there are many 'random' ORFs longer than e.g. 100 amino acids, a cut-off that is often used for considering an ORF a real gene. The large number of short 'random' ORFs makes is difficult to discriminate real genes from random ORFs below a certain length, which depends on the genome and in particular its GC content. Therefore many genomes are over-annotated [2]. In one genome, that of A. pernix [3], all ORFs longer than 100 amino acids are annotated as genes, but probably only around half the annotated genes are real [2]. The most severe problem today is to disciminate between short genes and random ORFs, and here the meaning of 'short' is quite organism dependent. One of the most important contributions of this paper is a way to deal with this problem by introducing a statistical significance for an ORF being a gene. Computational gene finding exploits the statistical differences in codon usage between coding and non-coding regions of DNA [4-6]. The search for a mathematical frame work to efficiently capture these differences in codon usage led to Markov chain models and the GeneMark algorithm [7]. In order to facilitate the combination of various Markov chain scores, the application of Hidden Markov Models (HMMs) to gene finding was introduced in a gene finder for E. coli, Ecoparse [8]. These methods relied on a set of known genes for estimating parameters. More recently methods have been developed which start from a raw genome and automatically extract data for estimation. One of these, Glimmer [9], employs interpolated Markov models in order to use the maximum Markov chain order which can reliably be estimated for every oligomer. Another one, Orpheus [1], appeared the same year and calculates coding potentials of ORFs based on codon frequency of similarity-derived genes. Most of these gene finders also extract Shine-Dalgarno sequences in order to improve prediction of start codons. Due to their modular structure, HMMs are a suitable frame work for gene finding, and they are now used in GeneMark.hmm [10], GeneMarkS [11,12] and Frame-by-Frame [13]. GeneMarkS uses a mixture of Markov chains of order 0, 1 and 2 in combination with features of already annotated genomes to bootstrap an initial estimation of coding statistics, which is then iteratively improved by the GeneMark.hmm2.1 algorithm. Gibbs sampling is also used to detect Ribosome Binding Sites (RBS). The Frame-by-Frame method was conceived to improve the accuracy of GeneMark.hmm and it employs standard Viterbi parsing of all six reading frames independently and subsequently combines these into a global parse. In this paper, we describe a fully automated method for making an organism specific gene finder. It starts from a raw genome and searches for protein matches to get a good training set. Then an HMM with states for coding regions as well as RBS is estimated from the data set. This HMM is used to score all the ORFs in the genome and finally the score is converted to a measure of significance – R – which is the expected number of ORFs that would be predicted in one megabase of random DNA. The main advantage of this significance measure is that it takes the length distribution of random ORFs properly into account. The method is shown to match or exceed other gene finders currently available. Methods Automatic extraction of training set In order to fully automate the construction of the gene finder, a data set of reliable genes was obtained through the following procedure, which is essentially the same as that used in [1]. All ORFs with a significant match to a protein from a different organism are assumed to be real genes. A subset of those have only one possible start codon, because there is only one start between the most upstream protein match and the nearest upstream in-frame stop codon. The details of the method are: 1. Extract the maximal ORFs longer than 120 bases from the query-genome. For every stop codon, extract the region from the first downstream (in frame) start codon to the next downstream (in frame) stop codon. 2. Translate the ORFs to proteins and search for significant protein matches in Swiss-Prot [14] using BLASTP without gaps and a threshold of 10-5 [15]excluding proteins from the query-genome and proteins listed with one or more of the keywords putative, unknown, possible, hypothetical, probable, bacteriophage, transposon, insertion, reverse transcriptase. 3. For each ORF with at least one significant protein match, identify the most upstream position of these matches (to compensate for random matches, we actually remove 6 bases from the most upstream match before proceeding). If there is no alternative start codon between this position and the start of the ORF, put it in set A' of genes with certain starts. The remaining ORFs are put in set B' of certain genes with uncertain starts. 4. Reduce similarity in set A' by removing genes with similarity within the set. All pairs of genes are compared using BLASTN with a significance cut-off of 0.001. Two genes that match are called neighbours. Genes are removed starting with the one having the largest number of neighbours. This continues until no gene has neighbours left in the set [16]. The reduced set is henceforth referred to as A. 5. Unite set A' and B' into set T' and reduce similarity of T' to obtain set T . 6. Add 50 bases of upstream flank to genes of all sets and 10 bases downstream flank. This procedure is a means to identify ORFs in a genome which are almost certain to be (protein coding) genes. The ORFs in set A and T make out reliable and balanced sets of positive examples which may be used to estimate the model parameters. HMM architecture We use a hidden Markov model (HMM) to model the gene structure. An HMM is a probabilistic model in which it is possible to model the various types of signals in a gene in one integrated model. For introductions to HMMs, see e.g. [17-19]. We model standard 'text book genes' with an unbroken open reading frame, i.e. genes with no programmed frame shifts, no sequencing errors nor any other special feature obstructing the reading frame. The general architecture of the EasyGene HMM is shown in figure 1. There is a begin and end state marked B and E respectively. Then, at each side of the gene model, there are null models to model everything that is not part of a gene nor lies in the immediate vicinity of a gene. The next submodel is the RBS model which models the RBS as well as the nucleotides between the RBS and the start codon. After the start codon we model 3 bases explicitly since it appears that the codon usage immediately downstream of start codons differs from the rest of the gene [13]. Similarly we model the last codon before the stop codon explicitly and 6 bases after the stop in order to capture information present around the stop codon. Figure 1 The overall HMM architecture. Each box corresponds to a submodel with more than one state. The number above the boxes indicates the number of bases modelled by the submodel. An 'X' indicates a variable number. We employ 3 looped codon submodels of the interior of the gene as depicted on the right hand side of figure 2. The reason for using several codon models is to embed a realistic length distribution in the HMM instead of the geometric distribution which would be implicit in having only one looped codon submodel [[17], Section 3.4]. We chose 3 because it results in a good fit to the empirical length distribution (see Results). The states in the codon model are of 4th order [20] in order to capture the inter-codon dependencies. The three looped codon models are identical, i.e. they have the same emission and transition probabilities (the states are tied). Figure 2 Enlargement of null model and internal looped codons. LEFT: The state structure of the NULL model. The background state is of third order and models the general composition of the genome. The three shadow states model coding regions on the complementary strand. There are transitions from the background state to the first RBS state and to the first state modelling the start codon. RIGHT: Details of model of internal codons. A codon is modelled by three states with a transition from the last state back to the first and one out of the codon model. By putting several codon models in series, the length distribution of coding regions can be captured. From the last state there is a transition to the first state of the 'BSTOP' model, which models the last codon before the stop codon. The RBS model in figure 3 models the ribosome binding site as well as the nucleotides between the RBS and the start codon. It has 7 states to capture any ribosome binding patterns and 12 tied spacer states for modelling the region between the RBS and the start codon. From the first spacer state there are transitions to all but the last of the following states, so the length distribution (with a minimum of 3 and maximum of 12) can be modelled exactly without imposing Gaussian assumptions as is done in e.g. [21]. These spacer states are of order zero. Figure 3 The state structure of the RBS model. The RBS model consists of seven states for modelling the ribosome binding site followed by a set of tied states for the variable region between the RBS and the start codon. From the last state there is a transition to the first of the three states modelling the start codon. The null model depicted on the left side of figure 2 has a third order state for capturing intergenic regions and a reverse codon model for modelling reverse genes ('shadows') with states of second order. Note that transitions are allowed directly from the null model to a start codon; this facilitates detection of genes inside operons which may not have a clear RBS. We found that the inclusion of two more branches of internal codons improved performance. This is presumably because it allows the HMM to keep separate statistics for atypical genes, some of which may be horizontally transfered. Adding a fourth branch did not improve performance further, so we stopped at three lending some support to the hypothesis that there are essentially three classes of genes in prokaryotic genomes [22]. Model estimation The HMM parameters (transition and emission probabilities) are estimated with the Baum-Welch algorithm, which is a maximum likelihood approach that finds the parameters maximizing the probability of the training set, see e.g. [17]. The estimation is done in these stages: 1. The emission probabilities of the background state are estimated from both strands of the complete genome. 2. The genes from start to stop codons are extracted from the training set and reverse complemented. The shadow model (consisting of three states) is estimated from these sequences. The parameters of this model are fixed. 3. The RBS, start and astart submodels (cf. figure 1) are estimated using set A exclusively since this set has reliable gene starts and therefore aligned upstream regions. Regions of 50 bases upstream from the start codon are extracted from all genes in set A. A null model (with fixed parameters) is included before the RBS model. The RBS model is initially primed with a high probability for the consensus GG (a dinucleotide common to most RBS), but the precise pattern is found by the estimation method. During estimation, a type of simulated annealing is used, where noise of a decreasing amount is added to the parameters in each iteration of the Baum-Welch algorithm [23]. After training on set A, the parameters of the RBS, start and astart models are fixed. 4. The null model, RBS, start and astart models are now combined with the internal codons, bstop, stop and astop models to make up the complete model. The non-fixed parameters of this model are then trained on the (larger) set T . The whole procedure can be completely automated. Note that no experimentally mapped RBSs are used for estimating the RBS model, the RBS is discovered during the estimation procedure. When estimating the complete model (stage 4 above) we use labelled estimation [17,20], where each base of the sequence is labelled as coding or non-coding. For the part of set T where the exact start is not known, we leave the part of the sequence from the most upstream start codon until the first significant protein match unlabelled. The weight with which each base in the unlabelled region contributes to the estimation of the parameters in the coding states is automatically determined during the iterative estimation procedure. Decoding By decoding we mean the process of finding an optimal parse of the DNA string into coding and non-coding regions. The commonly used Viterbi decoding returns the most probable path of the sequence given the model [17], but this is not appropriate when the length is modelled by duplicating codon states, since this length modelling is realized only as a sum over many HMM paths. Therefore we use posterior decoding where one calculates, for each nucleotide i, the probability that it was emitted by a given state S. The calculation is done by adding the probabilities of all paths compatible with having state S emit nucleotide i [24,17]. We use this to calculate the posterior probability of the first state of the start codon model, and thus obtain the probability that a gene starts at any position in the sequence. Given our assumption of perfect 'textbook-genes' with no errors or frame shifts, there is exactly one stop codon for each start, and thus the probability of a gene start is equal to the probability of the whole gene. The independent scoring of start codons makes it trivial to report several possible start codons for a gene in cases where there is no clear "winner". Note that the HMM architecture in figure 1 is non-looped – ie. it would find only one gene if we were using Viterbi decoding, which only gives the single most likely parse. This architecture however, is the correct one for scoring ORFs with posterior decoding. It has the further advantage that overlapping genes are easily handled since each gene start will be scored independently whether or not it overlaps other genes. In contrast, using a looped model and Viterbi decoding would not facilitate detection of overlapping genes unless the model contains explicit states for overlapping genes as described in e.g. [8]. The state posterior probability itself is not a useful score, because it is a probability of the whole sequence, not just a single gene, and it therefore depends on the length of the sequence it is part of. By dividing the posterior probability by the probability of the whole sequence (the genome) according to the null model, the contribution to the state posterior probability of the sequences flanking a gene will cancel and effectively make the ratio independent of the flanking sequences (except the parts very close to the gene), see the Appendix. The log of this ratio is called the log-odds score, and that is our basic score for a gene. Significance As mentioned above, it is important to take into account the chance that a random ORF of the same length scores as high as a given gene. This is implicitly taken into account by our HMM because it models the length distribution of genes, but it turns out that one can calculate a significance score, which works slightly better (se results) and has a more intuitive interpretation. The probability of finding high-scoring ORFs in a random sequence is highly length dependent; the number of ORFs decays exponentially with the length so there will be a lot more short ORFs than long ones. For a random sequence of given length (e.g. 1 Mb) the expected number of ORFs of length l' can be written as N(l') = exp (A - Bl'),     (1) where A and B are constants that can be found from linear regression of the log of the number of ORFs against the length. All lengths are measured in codons and we count both start and stop codon. For convenience we introduce the variable length l', which is the length of the ORF modelled by the looped codon submodels. A number of codons l0 are modelled explicitly in the beginning and end of the ORF (l0 = 4 in our model), so l' = l - l0. If the log-odds score is denoted β, we show in the appendix that β' = β - (n - 1) log(l' - n/2) is approximately normally distributed with a mean αμ + γμ l' and variance ασ + γσ l', which are both linear in l' (the α's and γ's are constants). Here n is the number of looped codon submodels (3 in our model). The coefficients of the linear mean and variance are estimated by linear regression on ORFs from random sequences. Then we define the standard score which is normally distributed with mean 0 and variance 1. For any given length l' and standard score Γ, one can now estimate the expected number C(l', Γ) of ORFs of the same length scoring higher than Γ in a fixed (long) random sequence using equation (1), C(l', Γ) = exp(A - Bl') [1 - Φ (Γ)],     (3) where Φ is the cumulative normal distribution. This number can be used directly to judge the significance of a gene predicted with length l and a standard score Γ. However, it would be preferable to know the total number of expected genes (of all lengths) predicted in a random sequence, rather than the expected number with a certain length. Therefore, instead of using C(l', Γ) to judge significance, we use the total number of expected predictions in a random sequence. Suppose a gene of length l1 is predicted with a standard score Γ1. Then the expected number of genes of that length predicted in a random sequence is C = C(, Γ1). Now we want to calculate the total number of genes of all lengths predicted in a random sequence at the threshold C. For any length above a certain lC, the expected number of ORFs will fall below C due to the exponential decay of the length distribution. Therefore the total number of predicted genes (regardless of length) is roughly lC C + the sum of predicted genes above lC (a sum of a geometric series). We end up with the following expression for the total number of predicted genes in a random sequence: The number 16 arises from the fact that the minimum ORF length considered is 20 codons corresponding to a variable length of 20 - 4 = 16. See Appendix for details. This is the expected number of genes predicted in a random sequence for a given value of C, and that number is the one we quote for each predicted gene. The constants A and B characterize the random sequence. R depends on the standard score Γ through C. This dependence is illustated in Figure 4. It clearly shows that a short ORF needs to have a much higher standard score than a long ORF in order to be significant (in the sense of having a low R-value). Figure 4 Relationship between R, Γ and variable length in codons l'. The numbers are taken from the E. coli runs described in Results and Discussion, but the qualitative behavior is independent of the genome We have chosen to normalize R to a random sequence of 1 Mb rather than a random sequence the length of the genome, because then significance can be compared across genomes. The precise recipe we use for calculation of statistical significance is: • For the genome in question, generate 40 Mb of random sequence using the 3rd order background model (estimated from the genome). Extract all ORFs and estimate the parameters A and B by linear regression of the log of the number of ORFs against the length and normalize to 1 Mb. To avoid distortion of regression lines due to noisy statistics of long ORFs, restrict the variable ORF length used to lie below 70 codons. This makes the range of variable length 16–70 codons, which corresponds to a total length range of 20–74 codons. • Score all the ORFs in the random sequence with the model and calculate β' for all ORFs. For each length, calculate the average and variance and estimate the parameters αμ, γμ, ασ, and γσ by linear regression, again using ORFs in the range 20–74 codons. • To calculate the significance of an ORF in the genome, first calculate the standard score Γ from equation 2, then C from equation 3, and finally the significance value R using equation 4. There are of course other possible choices of significance measure, but we believe that this is a simple and intuitively clear one and we have prefered it to the more traditional significance measure. By reporting the number R of expected false positives in one megabase of random sequence, it is easy for the end-user to estimate the number of false positives in a random sequence the length of the entire query genome – one simply has to multiply R with the size of the genome measured in megabases. Using other gene finders In order to benchmark EasyGene we compare it with some of the existing gene finders. GeneMark 2.4, Frame-by-Frame, GeneMark.hmm 2.1 and GeneMark.hmm/S all belong to the GeneMark suit of programs and are accessible via the web interfaces listed in the references. For GeneMark.hmm2.1, GeneMark.hmm/S and Frame-by-Frame the output is a coordinate listing (start and stop positions) of all predicted genes. GeneMark2.4 outputs a list of stop codons and corresponding high-scoring start codons. Each start/stop is listed with scores for coding potential and RBS. We collect all starts for a given stop and output the "Avg Prob" of the start with the highest RBS score. Whenever a threshold was needed for comparison purposes, we used 0.5 which is the default set on the web page. Glimmer2.02 and Orpheus2 were installed locally. We changed the minimum ORF length predicted by these gene finders to 60 bp which seems to be the minimum used by the other gene finders. Orpheus and Glimmer provide two kinds of output: a verbose coordinate list of starts, stops and ORF scores and a simpler, post-processed list of coordinates for ORFs regarded as genes. In order to test their ORF scoring we had to parse the scored output. We had some difficulties interpreting the scored Orpheus output since some ORFs were scored several times with identical results (several identical "Start chosen"). In cases where multiple copies were found, we simply chose one of them and used the corresponding "Coding potential" (with the recommended threshold of -1) for further analysis. For Glimmer2.02 the scored output was parsed simply by selecting the Gene Score attributed to every scored ORF and using the recommended threshold of 90. Finally, we used RBSfinder [25] for an alternative post-processing of Glimmer2.02 output. RBSfinder is designed to look for RBS sequences upstream of genes predicted by Glimmer2.02. If there are no RBS patterns in this region, RBSfinder searches for a start codon having a RBS pattern in the same reading frame upstream or downstream and relocates the start codon accordingly. The program may be iterated several times using revised predictions as new inputs. We found that running it twice was better than once but running it three times did not improve things further, so we chose to run it twice (with default options). Results and Discussion A number of tests were conducted in order to optimize the model architecture. We tested the number of codon models in series and found that three models yield a very good fit to the observed length distribution, see figure 5 for an example. Figure 5 Gene length distribution imposed by HMM architecture. The model length distribution given by a negative binomial (equation 8 with n = 3) compared to the length histogram of set A genes for H. pylori J99. The results of our experiments with the number of branches in the coding model and the order of the coding states are shown in figure 6 and 7 respectively, which show (cross-validated) performance curves (ROC curves) for varying numbers of branches and various orders. The performance curves are made by plotting the average true positive rates for a range of average false positive rates (the fraction of false positives made on average by the 10 different cross-validation models in 1 Mb of random sequence). Figure 6 Assessing the optimal number of HMM coding branches. Performance curves for 1,2,3 and 4 Markov branches of looped codon submodels for E. coli. The performance curves are made by the following procedure: First we sort the positive R-values in ascending order for each of the 10 subsets of set T (test sets). Then for each ascending R-value we calculate the fraction of genes in set T scoring below R (true positive rate) and the fraction of ORFs (with lengths greater than or equal to 20 codons) in one megabase double-stranded sequence scoring below R (false positive rate). The resulting 10 files with true and false positive rates are concatenated and 30 false positive cutoffs are selected (from 0 to 0.15 with steps of 0.005). The false positive entries in the 10 files which fall between these cutoffs are found and the corresponding true positive entries are averaged. Hence for each average false positive rate (halfway between two consecutive false positive cutoffs) we associate an average true positive rate and these tuples are then plotted. Figure 7 Assessing the optimal order of looped codon states. Performance curves for 3rd, 4th and 5th order Markov states of looped codon submodels for E. coli. For explanation of the construction of performance curves please confer the caption of figure 6. Figure 6 indicates that 3 branches is the best choice, while figure 7 suggests that the optimal order for these branches is 4. Note that the Y-axes of these figures have been zoomed in order to allow visual inspection of the performance differences. Note also that all architectures and orders in fact yield a relatively high true positive rate even at false positive rates below 0.02. Similar figures were also made for other organisms (not shown) and although the results were not always as clear as for E. coli, a choice of 3 branches of 4th order states as default models of the internal parts of genes works well for the organisms we tested. Similarly, figure 8 shows performance curves to compare log-odds and significance scores and the significance scoring is seen to be slightly better in that it allows detection of more true positives for a given false positive rate. The significance scoring has the additional advantage of being genome independent and has an intuitively appealing interpretation. Figure 8 Comparing significance and log-odds. Performance curves comparing significance and log-odds scores for E. coli. For explanation of the construction of performance curves please confer the caption of figure 6. Table 1 shows the average true and false positive rates for selected R-values in the case of E. coli models with three branches and fourth order branch states. This table gives an impression of the approximate R values corresponding to the graphs in figures 6, 7 and 8. Table 1 True and false positive rates for selected R-values. R-value TP rate FP rate 0.1 0.971 0 2 0.980 2.3e-6 10 0.984 3.0e-5 50 0.987 3.1e-4 150 0.991 8.7e-4 500 0.995 2.8e-3 10000 0.999 0.059 True and false positive rates averaged over 10 cross-validations for selected R-values in the case of the three branches, fourth order E. coli model. The FP rate is measured as the average fraction of ORFs (with lengths greater than or equal to 20 codons) in one megabase of double-stranded random sequence scoring lower than the given R-value. The TP rate is measured as the average fraction of ORFs in data set T scoring lower than the given R-value In order to test the validity of the approximations used in the derivation of the significance measure, we generated 1 Mb of random sequence from the Markov chain corresponding to the background state of the E. coli model. The plots in Figure 9 show the mean and variance of the length-corrected log-odds score, β', for each length. They are very close to being linear as we assumed. The length distribution of the ORFs in the same random sequence is also shown, which confirm the geometric length distribution. The predicted number of significant ORFs in 1 Mb random sequence is compared with the theoretical significance value R in Figure 10 and the agreement is seen to be rather good. We also compared the distribution of standard (3) scores for ORFs in the same random sequences to a normal distribution of unit variance and zero mean and the agreement also turned out to be good in this case as seen in Figure 11. We conclude that the assumptions and approximations used in the calculation of R are quite accurate. Figure 9 Statistical characteristics of random sequences. The top two panels show the mean and variance of log-odds scores versus variable ORF length in random sequences (E. coli model). Lowest subplot shows a logarithmic plot of the length distribution of random ORFs. The linear regression lines are shown in all three plots. Figure 10 Comparing predicted and found number of false positives. Empirical and theoretical number of false positives per Mb double-stranded random sequence according to the E. coli model. Figure 11 Probability density functions for the standard score. Empirical (dots) and theoretical (line) probability density functions for the standard scores (Γ) in random sequences (E. coli model). The lower plot is an enlargement of the distribution tail. Table 2 shows the percentage of genes found for eight different gene finders and some sets of high-confidence genes from E. coli as well as the number of genes found in the whole genome and in random sequences. The eight gene finders are: EasyGene, Glimmer2.02, Glimmer2.02 with RBSfinder post-processing, Orpheus, GeneMark2.4, GeneMark.hmm2.1/GeneMarkS hybrid, "pure" GeneMark.hmm2.1 and Frame-by-Frame. Table 2 Specificity, sensitivity and precision estimates for different gene finders in E. coli. Data set EasyGene Glim rbs-Glim Orpheus Gm24 GmS Gmhmm Frame A'-% found 98.4 98.9/98.9 98.9 98.0/95.3 91.5 97.2 98.1 97.0 A'-% exact 93.8 98.9/95.3 84.1 95.1/92.4 41.6 88.0 85.7 93.2 B'-% found 98.4 98.5/98.6 98.6 95.9/96.5 90.2 96.6 97.2 96.4 T-% found 98.1(98.0) 98.3/98.4 98.4 96.5/95.6 89.8 96.3 97.1 96.1 Genome 4145 6827/5756 5756 9333/7543 3552 4064 4230 4064 zero order 7 169/211 211 6761/5430 6 153 1459 0 first order 7 545/723 723 6836/4804 13 241 830 0 third order 1 2423/2694 2694 6582/4817 43 659 866 1 shadows 0 19/21 21 22/9 1 0 2 0 Upper part shows the percentage of genes found exactly (both 5' and 3' end) and partially (only 3' end exact) for different gene finders and sets of high confidence genes in E. coli. For Glimmer and Orpheus, the numbers before the "/" are based exclusively on their ORF scores and recommended threshold whereas the numbers after the "/" are based on their post-processing procedures. The number of genes predicted in the whole genome is also shown. This should be compared to the 4288 annotated genes in E. coli. The lower part of the table shows the number of false positives predicted in random sequences generated by Markov chains of order 0, 1 and 3 and the very last row shows the number of false predictions in the shadows of the high-confidence genes in data set A. All values listed for EasyGene are based on an R-value threshold of R = 2. Sequence set A' consist of the 1136 genes with high-confidence starts extracted from the E. coli genome as described in Methods. The "% exact" row indicates the percentage of predictions where both start and stop codon are correctly predicted whereas the "% found" row indicates that only the stop codon is correct. Note that all genes in set A have the most upstream start (they are Longest Possible Open Reading Frames – LPORFs) by construction, and hence performance on this set favours gene finders which are biased towards LPORFs (such as Glimmer). Set B' consists of 1690 high-confidence genes with uncertain starts extracted as described in Methods. Due to the uncertainty of start codon placement, one cannot evaluate the exact start prediction performance for this set. The same is true for set T which is a similarity reduced union of A' and B' (2042 sequences). The Genome row shows the number of genes predicted in the E. coli genome using default parameters and thresholds for the various gene finders. For Glimmer and Orpheus the minimum length of predicted genes was lowered to 60 in order to make their performance comparable to the others'. The next three rows show the number of false positives found in both strands of 1 Mb random sequences generated by zero, first and third order Markov chains estimated from the entire E. coli genome. The last row of the table shows the number of predictions wholly within the shadows of set A' – i.e. wholly within regions complementary to the genes in set A' where, ideally, no genes (or at least very few) should be predicted. The test sets overlap with the training sets for EasyGene. Therefore, for set T the 10-fold cross validation sensitivity is shown in parenthesis for EasyGene and it is seen to be reassuringly close to the non-crossvalidated sensitivity suggesting that EasyGene employs an appropriate model complexity and steers free of overfitting. Note also that the other gene finders have also been estimated from sets that overlap (or even contain) set T . For Orpheus and Glimmer we show two numbers N1/N2 for each entry corresponding to before and after post-processing. For Orpheus, N1 is the number of unique ORFs having a Coding Potential above the recommended minimum of -1 and N2 is the number of entries in the post-processed orfnuc file. The post-processing removes some overlaps but also appears to employ a less restrictive cutoff than the recommended -1. For Glimmer, N1 is the number of ORFs with Gene Scores greater than or equal to the recommended threshold of 90 and N2 is the number of entries in the post-processed list of putative genes. The post-processing elects ("votes") some ORFs as gene candidates despite a low score. On the other hand the post-procesing removes some same-strand overlaps in different reading frames so the N2 may be greater or less than N1 depending on the relative extent of overlaps and "voting". It is always difficult to asses the specificity of a gene finder, because it is difficult to find genomic regions that are certain to contain no genes. We have therefore assessed specificity in three different ways. First, by counting the number of predicted genes in a genome. If this number is much higher than the number of annotated genes, it is likely that there are many false positive predictions, i.e. poor specificity. Our second test is based on random sequence. Clearly, a high number of predicted genes in a random sequence of bases indicates a poor specificity. However, it is probably not possible to find an exact quantitative correspondence between predictions in random sequences and real genomes. Also, it is not clear what sort of random sequence to use for such a test. By 0'th order we mean a sequence with bases generated randomly and independently with the base frequencies of the genome. Bases are quite correlated in DNA sequences, so we have also tested on sequences that are generated by Markov chains of orders 1 and 3. These Markov chains are estimated from the genome, so the sequences will have the same distribution of dinucleotides and 4-nucleotides, respectively, as the genome. Finally, the third test is the number of genes predicted on the opposite strand of genes (shadows); these shadow regions should contain very few genes if any. All gene finders except EasyGene, Frame and GeneMark2.4 predict a rather large number of false positives in random sequences, but for GeneMark.hmm and GeneMarkS we do not see large over-prediction in the genome or in shadows. Evidently, Glimmer and Orpheus predict a lot more genes in the genome than the other gene finders, suggesting that these gene finders have very high false positive levels. This is supported by the high numbers of genes predicted in random sequences, and (to a much lesser extent) in shadows. Orpheus and Glimmer actually predict more genes pr. nucleotide in the third order random sequence than they do in the genome, suggesting that the coding potential calculated in these gene finders is far from optimal. The HMM used by Frame assumes a minimum gene length of 69 bases which could make its false positive level seem somewhat better (lower) than it is, but there was no convenient way to lower the minimum length so we simply left it. It should also be noted that Frame relies on pre-existing annotations for training and is therefore not a self-training gene finder like Glimmer, Orpheus, GeneMarkS and Easygene. The sensitivity of the gene finders is tested on sets of high-confidence genes. Glimmer has the highest sensitivity for all sets, but this is largely due to heavy over-prediction. One ought always to bear in mind that it is not difficult to achieve high sensitivity if high levels of false predictions are tolerated at the same time – sensitivity is 100% if all ORFs are predicted as genes! Although there are some very close competitors, EasyGene comes out as the second best in sensitivity for all sets. The exact prediction of start codons is tested on set A' and on an experimentally ver-ified set. As mentioned above, set A is biased, because all genes of this set are LPORFs. Glimmer always predicts the most upstream start and consequently achieves a high performance on this set. When Glimmer's output is post-processed by RBSfinder the performance drops considerably. The prediction of start codons was evaluated further on a set of 195 E. coli genes with experimentally verified starts [26] with results shown in table 3. Set LiC is the subset of 133 genes which coincide with the longest possible open reading frame (LPORF) while set LiD is the remaining 62 genes whose starts are downstream of the LPORF start. The goal is to find the starts of the challenging LiD set without loosing too many of the more trivial starts of set LiC. Table 3 shows that while most gene finders partially locate all genes in set LiC and LiD, there are large variations in their exact localization ability. Selecting for the highest combined performance on set LiC and LiD, one sees that EasyGene, GeneMarkS and Frame-by-Frame are best. Their performances also exceed that of [27] in which a cross-validated performance of 84.9% +/- 4% is reported on a subset of 184 genes out of the 195. In the low end we have Glimmer finding 0% of set LiD exact (1.6% with post-processing) and GeneMark2.4 finding 49.6% of set LiC exact. Using the RBSfinder post-processing [25] on the Glimmer predictions improves performance on set LiD to 75.8%, but at the cost of a substantial dip in set LiC performance to 88.7%. Table 3 Sensitivity and precision estimates for experimentally verified E. coli genes. Data set EasyGene Glim rbs-Glim Orpheus Gm24 GmS Gmhmm Frame LiC-% found 100 100/100 100 97.7/91.7 97.7 100 100 100 LiC-% exact 94.0 100/97.0 88.7 96.2/90.2 49.6 94.0 90.2 98.5 LiD-% found 100 100/100 100 96.8/98.4 100 100 100 100 LiD-% exact 96.8 0/1.6 75.8 51.5/51.6 67.7 95.2 80.6 87.0 Percentage of genes found exactly and partially in two subsets of the 195 experimentally verified genes published by [26]. All values listed for EasyGene are based on an R-value threshold of R = 2. Many gene finders are first developed for E. coli and then later adapted to other organisms. It is therefore important also to test gene finders on other organisms. Based on table 2 and 3 we chose EasyGene, GeneMarkS and Frame-by-Frame as the gene finders with the best overall performances and then conducted further comparisons between these for M. tuberculosis [28], H. pylori [29] and B. subtilis [30]. M. tuberculosis presents a challenge due to GC richness, H. pylori due to small genome size and B. subtilis was chosen on account of its reputation of being well annotated [11]. The results are presented in table 4 with the same rows as table 2. Table 4 Specificity, sensitivity and precision estimates for M. tuberculosis, H. pylori J99 and B. subtilis. M. tuberculosis Data set EasyGene GmS Frame A'-% found 96.7 97.2 96.0 A'-% exact 89.1 80.9 87.9 B'-% found 96.8 97.1 96.3 T-% found 96.9(96.6) 97.3 96.4 Genome 3749 3983 4341 zero order 0 - 8 first order 3 - 2 third order 2 - 12 shadows 1 0 1 H. pylori J99 Data set EasyGene GmS Frame A'-% found 99.2 99.2 99.2 A'-% exact 97.5 95.7 96.7 B'-% found 100 99.6 98.9 T-% found 99.7(98.8) 99.5 99.1 Genome 1491 1518 1479 zero order 2 1479 2 first order 1 336 2 third order 0 403 0 shadows 2 0 0 B. subtilis Data set EasyGene GmS Frame A'-% found 99.3 98.1 98.8 A'-% exacts 94.8 94.1 93.3 B'-% found 99.2 99.0 98.2 T-% found 99.3(99.2) 99.0 98.4 Genome 4083 4221 4006 zero order 1 675 0 first order 2 457 0 third order 1 813 2 shadows 0 0 0 Genes found exactly and partially for different gene finders and sets of high confidence genes in M. tuberculosis, H. pylori J99 and B. subtilis, where the number of annotated genes is 3918, 1491 and 4100 respectively. There are no pre-trained GeneMarkS models for M. tuberculosis, so it was not possible to obtain a false positive estimate for this organism. All values listed for EasyGene are based on an R-value threshold of R = 2. For M. tuberculosis GeneMarkS and EasyGene are comparable, although GeneMarkS seems to over-predict slightly (assuming that the 3918 annotated genes are close to being correct), and Easygene might under-predict. We believe that Frame predicts too many genes in this organism and at the same time it has lower sensitivity than the two others, suggesting a worse performance overall. This indicates that Frame is not very robust with respect to high GC content. The small dataset from H. pylori might give a slight over-fitting in EasyGene, where there is one percent difference between cross-validated results and non-crossvalidated. For this organism the three gene finders seem to have very similar performances. Finally, for B. subtilis EasyGene comes closest to the number of annotated genes and have higher sensitivity than the other methods. Conclusions The emerging overall picture is that the sensitivity of EasyGene tends to be comparable to GeneMarkS and higher than Frame. With regards to specificity, EasyGene and Frame tend to be comparable and both higher than GeneMarkS. Hence, EasyGene comes out with the best combined sensitivity/specificity performance. When it comes to exact starts, EasyGene also generally performs best. Glimmer and Orpheus have good sensitivities but at the cost of very low specificities in this comparison. We have lowered the ORF length cutoff from their default values in these methods to make the results comparable. This may be unfair, but since the challenge is to find the short genes, we believe that any gene finder should be able to score them. At present it is not possible to automatically find all genes in a prokaryotic genome. We believe the aim of a gene finding system is to help expert annotators as much as possible, and we consider the statistical significance of a gene an important help in classifying the predictions into almost certain genes and border-line genes needing more attention. Contrary to most other gene finders discussed here, it is up to the user which significance cut-off to use. EasyGene also predicts sub-optimal start codons if need be, so it will be easy to see if e.g. two alternative starts have almost the same score. A shortcoming of the significance value as calculated here is that long ORFs score well simply on account of their length, since very long ORFs occur rarely in random sequences. For this reason, EasyGene also provides a log-odds score in the output which may be held up against the R-value for ORFs longer than say 500 bp. Genes of this length ought to have high log-odds values. If they do not – i.e. if they score below 0 – then they are probably not real genes despite their length (sometimes very long non-coding ORFs occur in regions of repetitive DNA). If the amount of available genomic DNA is very small (as it may be in partially sequenced genomes) the 3 branches of 4th order coding models may have too many parameters to be reliably estimated. In such cases, one could reduce the parameter space simply by using fewer branches and/or lower orders. More generally, one could develop a method for automatic fine-tuning of HMM architecture for every new organism. Other lines of future research could focus on modelling of genes with errors and frame shifts. Finally, it may be noted that a prototype of EasyGene has already been used in the annotation of S. typhi [31]. Appendix: The length dependent score distribution The log-odds score distribution has several components. The probability of a sequence z containing and ORF given a model M may be written as where c1, . . . , cl denotes the codons in the ORF and flanks denotes all the rest of the sequence. P (l|M) is the length distribution of the HMM. In logarithmic form we have Similarly, the denominator of the log-odds score, log P (z|N ), may be rewritten in exactly the same way, so the entire log-odds score becomes where We will now look at the distribution of each of these terms. The null model consists of a state with a loop and three reverse codon states with a loop. For a long sequence one of the loops will usually dominate the probability, so the length distribution is well approximated by a geometric distribution P(l|N) k1 exp (-k1l) where k1 is a constant greater than zero. The length distribution from the looped codon states is a negative binomial [17], where p is loop probability and n is the number of looped codon submodels (in casu, n = 3). Some of the ORF (start and end) is modelled by non-looped states. The number of nucleotides modelled in the beginning and end of the ORF is denoted l0 and the variable length is denoted l', so the total ORF length is l = l0 + l'. Using that log   (n - 1) log(l' - n/2) - log(n-1)! for l' >>n, we obtain The next term, Q, in equation 6 is a sum of random variables since we are considering random sequences. The sum Q will therefore (according to the central limit theorem) converge to a normal distribution N (μl, σ), for large l. The submodels flanking the gene model are identical to the null model. Therefore p(flanks|M) and p(flanks|N) will almost cancel in the last term of equation 6, except from the contribution from the RBS model and the states after the stop codon. This contribution will be independent of the ORF length. Since it is again a sum of random terms it is well approximated by a normal distribution. Apart from the non-linear term from the negative binomial we have now shown that all terms in equation 6 are constant or scale linearly with l. Therefore is normally distributed for fixed length with an average and variance that are linear functions of l: average = αμ + γμl'     (11) The parameters of these linear functions can readily be estimated from random sequences by doing linear regression between the variable length and the mean and variance of log-odds scores as shown in the two upper plots of figure 9. (Note that one can switch from the variable length l' to the total length l in the formulas by changing the constants.) Finally it is convenient to introduce the standard score Γ, defined in equation 2, which is normally distributed with average 0 and variation 1. Since random ORF lengths are geometrically distributed (cf. equation 1), the expected number of ORFs of length l scoring more than Γ in a sequence is C(l', Γ) = exp (A - Bl') [1 - Φ(Γ)],     (13) where the exponential term is the expected number of ORFs of variable length l' and Φ is the cumulative normal distribution. Thus, for a given score we can calculate the number of expected predictions in a random sequence. For lengths l ≥ lC this expectation is always less than C (due to the exponential factor). So if we require that no more than C predictions be made at any length, the total number of predictions in a random genome is where ls is the shortest ORF length considered (we use 20). lC is given by Strictly speaking, we should take the smallest integer larger than , but using the real expression introduces only an insignificant error. Inserting into equation 14 finally yields equation 4, R = {(A - log C)/B - (ls - l0)}C + C/(1 - exp(-B)).     (16) Web sites used Frame-by-Frame: GeneMark.hmm 2.1: GeneMark.hmm 2.1 using GeneMarkS models: GeneMark 2.4:

Hernia-4-1-2231409

Impact of the mesh fixation technique on operation time in laparoscopic repair of ventral hernias

Background Fixation of the prosthesis is one of the critical components of laparoscopic repair of ventral and incisional hernia (LRVIH). The impact of the fixation technique used on operative time has never been analyzed. We compared the duration of the operation according to the fixation technique used in a series of 138 patients with primary umbilical hernia. Introduction Laparoscopic repair of ventral and incisional hernia (LRVIH) is gaining increasing popularity due to its low recurrence rate, short hospital stay, and low complication rate. Fixation of the prosthesis is one of the critical components of LRVIH. Currently, the two most popular methods of mesh fixation are the use of helical tackers either with or without transabdominal sutures (TAS). Overall it appears that the two laparoscopic mesh fixation techniques are similar in outcomes [1]. However, the impact of the applied fixation technique on operative time has never been analyzed. We compared the duration of the operation according to the fixation technique used in a series of patients with primary umbilical hernia (PUH) who underwent a straightforward repair by using completely standardized techniques. Patients and methods Data were collected from operative reports of all 167 patients who underwent laparoscopic repair of PUH up to January 2007. To make the data more homogenous, we excluded patients who required adhesiolysis (n = 13) or simultaneously underwent another procedure (n = 12), or with whom a minor complication occurred intraoperatively, i.e., bleeding or equipment problem (n = 4). The remaining 138 patients underwent straightforward repair of PUH by using a completely standardized technique, and they represented the study group used to compare operative times between the two mesh fixation techniques. Pneumoperitoneum was established by using a Veress needle. Three trocars (10, 5, and 5 mm) were inserted left laterally. A 1-mm-thick expanded polytetrafluoroethylene mesh (DualMesh®, WL Gore, Flagstaff, AZ) of 15 by 10 cm was used to overlap the hernia opening by at least 3–4 cm. The mesh was fixed using one of the following two techniques: either (1) with tackers (ProTack®, TycoUSS, Norwalk, CT) along the periphery of the mesh at intervals of 1–1.5 cm and eight TAS placed equidistant also along the periphery (n = 107, further called the TAS technique) or (2) with a double ring of tackers alone without the use of TAS (n = 31, further called the DC technique). With this technique, also known as a “double crown technique” [2], the outer ring of tackers is the same as in the TAS technique. The inner ring of tackers is placed around the hernia opening about 1 cm apart. Transabdominal sutures were pulled through the abdominal wall with a suture passer (Gore Suture Passer Instrument®, WLGore, Flagstaff, AZ). In the first 84 patients, the method of mesh fixation was based on the surgeon’s preference. For the last 54 patients, the mesh fixation technique performed was randomly chosen in conjunction with another study being done at our hospital that compares postoperative pain following these two methods of mesh fixation. Once the fixation was completed, the abdomen was desufflated and the trocar site of 10 mm and skin were closed. The time taken from the stab wound for insertion of a Veress needle to completion of skin closure was recorded to the nearest minute. Statistical analysis was done using the t test. Significance was set at a P < 0.05. Results Of the 138 patients in our study group, 106 were males and 36 females. Mean age was 55.0 ± 12.3 years. The TAS and DC groups were found to be well matched for sex and age. Mean operating time for the TAS technique was 50.6 min compared to 41.4 min for the DC technique. Mean difference in operating time was 9.2 min. This difference was significant (P = 0.002). Mean postoperative hospital stay was identical for both groups and equaled 1 day. Postoperative complications included seroma lasting longer than 6 weeks in seven patients and chronic pain at TAS sites in one patient. All seromas resolved without intervention. The patient with persisting pain underwent relaparoscopy and removal of all eight TAS that provided complete pain relief. There were no significant differences in morbidity between the two groups. No recurrences were detected during a mean follow-up of 26.4 ± 25.6 months. Discussion Besides prosthetic material that must sufficiently overlap the hernia defect, a reliable fixation of the prosthesis against the abdominal wall is a crucial component for success of LRVIH. Since the early years of LRVIH, there has been a strong belief among pioneers of this technique that reliable fixation of the mesh can only be achieved with the TAS technique [3, 4] and, in addition, that sutured mesh fixation is an imperative in LRVIH [5]. Another technique of fixation introduced later and consisting of a double crown of tackers only has been gaining increasing popularity due to a few specified advantages: technical simplicity, fewer incisions in the skin and possibly shorter operative time [2, 6]. One of the most interesting issues on LRVIH currently being debated is which of these two techniques is better. Prospective randomized studies comparing the two techniques are missing. Meta-analysis of published literature indicates that the two laparoscopic mesh fixation techniques are similar in main outcome parameters such as recurrence and complication rate [1]. In the same analysis, the use of tackers alone resulted in a slightly shorter operative time than when the TAS technique was used. However, in most of the studies that were included in this meta-analysis, fixation with tackers was performed by applying only a single row of tackers or technical details of the tacker fixation technique were not explained. Reliable data comparing TAS technique and DC technique are definitely missing. We decided to compare the two techniques in a maximally homogenous model of the procedure: same site of hernia, same logistics of the operation, same prosthetic material, same fixation device, and same operation technique until the moment that the outer ring of tackers is completed. The only difference between the two techniques was the completion of the mesh fixation. In one technique eight TAS had to be inserted and in the other technique usually 6–8 tackers of the inner crown. Once the fixation had been completed, whatever the method was, the rest of the procedure was again identical. Obviously, the difference in operative times between the two operative techniques can be entirely accounted to the difference in time needed for insertion of eight TAS compared to the time needed for application of an inner crown of tackers. Since tacking of the inner crown takes definitely not more than 1 min in laparoscopic repair of PUH, insertion of eight TAS required at least 8 min. This strongly indicates that insertion of every single TAS prolongs LRVIH by approximately 1 min. A laparoscopic repair of PUH is definitely the least complex procedure among all LRVIHs. Insertion of TAS is probably easier than in other LRIVHs due to a central location of the hernia, a general absence of adhesions, maximal space between the distended abdominal wall and the bowel underneath, and an excellent view. It may be anticipated that insertion of TAS during more complex laparoscopic repairs of incisional hernias at less suitable sites, in the presence of adhesions and proximity of the bowel can be much more challenging and as a consequence will require more time than during repair of a PUH. Our results indicate that LRVIH by using the DC technique indeed requires less operative time than when the TAS technique is used. As long as no significant differences between the two fixation techniques are demonstrated on issues of recurrence, complications, and postoperative pain, the time difference we have measured might be an argument in favor of the DC technique, especially when mesh fixation would require a large number of TAS.

Ann_Surg_Oncol-4-1-2266786

Liver Transplantation for Hepatocellular Carcinoma

Background Orthotopic liver transplantation (OLT) is the best available option for early hepatocellular carcinoma (HCC), although its application is limited by stringent selection criteria, costs, and deceased donor graft shortage, particularly in Asia, where living donor liver transplant (LDLT) has been developed. Hepatocellular carcinoma (HCC) is the third most common cause of cancer-related death worldwide.1 Orthotopic liver transplantation (OLT) is the best therapeutic option for early, unresectable HCC, although it is limited by graft shortage and the need for appropriate patient selection. In the late 1980s, the results after OLT for HCC were disappointing, with high early recurrence and 5-year survival rates ranging between 18 and 40%.2 This discouraging experience and shortage of deceased donor grafts compelled the transplant community to establish stringent selection criteria to predict posttransplant survival of HCC patients. Recognizing that patients with small, incidentally found tumors had survival rates after liver transplantation equivalent to those after transplantation for benign disease, Mazzaferro et al. in Milan established criteria for OLT in a landmark study published in 1996.3 They showed that a subgroup of patients with radiologic evidence of a single tumor ≤5 cm in diameter, or two to three tumors ≤3 cm in diameter had 5-year and recurrence-free survival rates of 75 and 83%, respectively. The Milan criteria were subsequently adopted by the United Network for Organ Sharing (UNOS) staging system for allocating organs for OLT in the United States. OLT is limited by a shortage of deceased donor liver grafts, particularly in Asia, where the rate of deceased donors is negligible. To overcome this shortage, living donor liver transplantation (LDLT) has been developed with favorable preliminary results. Other areas of study include expanding the criteria for liver transplantation to include larger tumors, as shown by the group from the University of California, San Francisco (UCSF) and molecular profiling of HCC to improve prognostication and patient selection. MILAN CRITERIA FOR LIVER TRANSPLANTATION IN HCC In the past 10 years, results of OLT have improved steadily because of careful patient selection pioneered by the introduction of the Conventional Milan Criteria (CMC) in 1996.3 The aim of these criteria was to achieve a good outcome in patients who fulfilled the criteria and avoid a poor prognosis in patients who exceeded them. This aim was achieved by the Milan group, who showed that the 10-year overall survival surpassed 70% in 300 liver transplants for HCC that fulfilled the CMC. Such good results have been confirmed worldwide.4,5 The elements of the CMC, namely size and number of tumors, have been shown in multivariate analysis to be the only independent variables predicting patient survival and tumor recurrence, with other biological prognostic factors playing a role only within the “size and number” limits (Table 1). TABLE 1.Prognostic factors affecting survival of 250 patients transplanted for HCC according to Conventional Milan Criteria at the National Cancer Institute of Milan10-year survival: multivariate analysisVariableHRCI (95%)P valueOverall patient survivalConventional Milan criteria (in vs out)3.11.35–6.93.007Tumor-free survivalConventional Milan criteria (in vs out)5.51.39–21.27.01Microsatellites (yes vs no)3.61.5–8.71.004Microvascular invasion (yes vs no)3.41.36–8.76.009Tumor grading (G3 vs G1-2)3.41.04–11.14.04 The successful outcome of OLT based on the CMC has led to more patients with HCC being routed to transplantation and an increasing number of proposals for expansion of the CMC.6,7 None of these expanded criteria are, however, supported by a robust statistical sample size or prospective validation. In order to investigate proposals to expand the CMC, an unprecedented multicenter study has been conducted in 24 European centers, collecting data from 466 patients transplanted for HCC whose tumors exceeded the CMC at posttransplant pathologic assessment (http://www.hcc-olt-metroticket.org).5,7 The results have been plotted in a tumor size-and-number Cartesian contour plot showing the 5-year survival probability according to the size and number of HCC nodules detected in the explanted liver. Based on a preliminary analysis, a “HCC forecast chart” has been developed, which can predict 5-year posttransplant survival rates on the basis of morphological tumor characteristics (Fig. 1). It is conceivable that similar chart models obtained from large sample sizes could replace the CMC for OLT in patients with HCC. Unlike the CMC or expanded proposals, which are based on tumor morphology, these chart models can incorporate important variables for OLT in HCC, including priority score for HCC and dropout rate based on donor availability at different centers. Preliminary ad interim analysis shows a significant shift of the isograms of the proposed chart depending on the presence of vascular invasion in explant specimens, confirming the need for reliable biological markers of important tumor characteristics, such as vascular invasion. FIG. 1.HCC forecast chart showing probability of 5-year survival based on tumor size and number in explanted liver. MILAN VERSUS EXPANDED CRITERIA FOR LIVER TRANSPLANTATION Following the 1996 publication by the group from Milan, using a restrictive set of criteria for OLT in patients with HCC, excellent 5-year posttransplant patient survival of at least 70% has been reported from many centers.4,5 The growing experience and success of OLT for HCC have fueled controversies related to expansion of the Milan criteria for OLT, since many studies have suggested that tumor stage beyond the Milan criteria does not necessarily predict worse survival after OLT.8,9 Among the proposed expanded criteria, the UCSF criteria (single tumor nodule up to 6.5 cm; or three or fewer tumors, the largest of which is ≤4.5 cm with the sum of the tumor diameters ≤8 cm) reflect a modest expansion of tumor size limits beyond the Milan criteria.6 At least three transplant experiences from Europe and the United States, however, underlined the limitations of applicability of the UCSF criteria in the pretransplant setting, considering that most of the patients adhering to the UCSF were also within the Milan criteria.10–12 Even though the UCSF criteria have been independently validated in several studies, the overlapping population of patients adhering to the UCSF but not the Milan criteria is often negligible and estimated to be <10% of the total transplanted population.12,13 Such a limitation was evidenced by a multicenter study from France, in which 39 of 461 patients (8.7%) had explanted tumors beyond the Milan but within the UCSF criteria.10 Although the 5-year survival of 67% for patients within the UCSF criteria was equivalent to that of the 183 patients meeting the Milan criteria (and significantly better than the 34% 5-year survival rate among 238 patients exceeding both criteria), the 44 patients meeting UCSF but exceeding Milan criteria at pretransplant staging had a 5-year survival rate of only 48%, compared with 60% observed in the 272 patients within Milan criteria and with 37% in the 121 patients beyond both criteria. The limitations of pretransplant imaging studies, exemplified by tumor understaging in 20% of patients, have been a major concern for liberalizing the existing criteria for OLT.5 The UCSF group applied their proposed criteria according to preoperative imaging in 138 patients over a 5-year period; the 1- and 5-year recurrence-free probabilities were 95 and 91%, and the respective probabilities for recurrence-free survival were 91 and 80%.14 The 106 patients with HCC meeting the Milan criteria (T1/2) had 1- and 5-year recurrence-free probabilities of 96 and 90%, respectively, compared with 93% at both 1 and 5 years for patients with HCC exceeding Milan but meeting UCSF criteria (T3A). Understaging by preoperative imaging was observed in 21% of T1/2 and 28% of T3A tumors. When explant tumor stage exceeded the UCSF criteria, the 1- and 5-year recurrence-free probabilities were 79 and 61%, compared with 98 and 97% within the UCSF criteria. Patients meeting T3A criteria did not have a significantly higher incidence of poorly differentiated tumor grade or vascular invasion. In the current era of increasing demand and unrelenting organ shortage, the foundation of the debate regarding expansion of the Milan criteria for HCC may ultimately rest on what the transplant community would consider an acceptable survival after OLT for HCC. Some groups have proposed a 50% 5-year patient survival to be the minimum acceptable cutoff.5 This mark may have been surpassed by the UCSF group, who have applied expanded criteria to benefit an additional 10% of patients with HCC with respect to posttransplant survival and tumor recurrence. LIVING DONOR LIVER TRANSPLANTATION FOR HCC: EASTERN EXPERIENCE Liver transplantation offers the best chance of cure for early unresectable HCC. However, its role has been limited by the shortage of deceased donor liver grafts, which is particularly severe in Asia, where the deceased donor organ rates are fewer than 5 donors per million population, compared with 10–35 donors per million population in Western countries.15 In most Asian countries, HCC is the most common cancer and the most frequent indication for OLT, aggravating the unmet demand for liver grafts. The development of living donor liver transplantation (LDLT), especially adult-to-adult right lobe liver transplantation, has allowed more patients with HCC to benefit from OLT.16 LDLT can theoretically provide an unlimited source of liver grafts and eliminate the uncertainty of prolonged waiting times and the risk of dropout due to tumor progression. However, LDLT is a novel treatment for HCC with unresolved issues regarding indications and results. Recent studies on LDLT for HCC suggest favorable long-term survival results.17,18 However, it remains unclear whether the outcome after LDLT for HCC is equivalent to that of deceased donor liver transplantation (DDLT). A study from a Korean group reported similar 3-year survival rates after LDLT and DDLT (91.4 vs 89.9%) in patients with tumors fulfilling the Milan criteria, after excluding perioperative mortality.13 However, others have found significantly higher rates of tumor recurrence after LDLT compared with DDLT for HCC.19 This may be related to selection bias, as LDLT eliminates the waiting period for grafts. In DDLT the waiting period provides time for a natural selection process in which patients with biologically more aggressive tumors drop out due to tumor progression. In addition, the ischemic-reperfusion injury associated with small-for-size grafts in LDLT and angiogenesis associated with liver regeneration may theoretically promote growth of tumors in the transplanted liver after LDLT, although the actual clinical impact of such biological processes remains unclear.20,21 Liver transplantation is conventionally offered to Child-Pugh class C patients with unresectable early HCC. Recently, there have been heated debates on whether liver transplantation should be used as first-line therapy for Child-Pugh class A patients with early HCC.22 While some previous studies showed that liver transplantation for early HCC may achieve better survival compared with resection, this may be partly related to selection bias in favor of transplantation because patients with more aggressive tumors drop off the waiting list for DDLT.19 Recent studies suggest that for patients with preserved liver function and early stage HCC, hepatic resection can achieve a 5-year survival rate of 70%, comparable to that after OLT.13,23 The availability of grafts from dedicated live donors has been considered one of the main arguments favoring LDLT as primary therapy for patients with early HCC and preserved liver function. However, even in Asian countries where LDLT is commonly performed, up to half of the patients with early HCC may not have suitable living donors for various reasons, including ABO-incompatibility, hepatitis serology, and patient refusal to accept living donation.18 Furthermore, the risks of donor hepatectomy, with morbidity and mortality rates of 14–21% and 0.25–1%, respectively, should be carefully balanced against the benefit of LDLT.19 Risking the life of a donor for HCC patients who have an alternative option of hepatic resection, which achieves long-term survival equivalent to LDLT, is ethically not acceptable to many Asian surgeons. Most Asian centers still consider resection as first-line treatment for HCC patients with preserved liver function and reserve LDLT as an option for salvage transplantation in patients with recurrent tumors.13,23,24 Another matter of debate in LDLT involves expanding the indications beyond the CMC or UCSF criteria. LDLT is currently being performed in some Asian centers for patients with HCC beyond the Milan criteria, with results that are expectedly worse than those for patients within the Milan criteria.13,17 Some transplant surgeons argue that despite the poorer results, LDLT for advanced HCC may be justified, since donors voluntarily accept the risks of donor hepatectomy to dedicate a graft for HCC patients, who may otherwise have no effective treatment options. However, others argue that the medical profession should not relegate the issue to individual donor autonomy.25 With the lack of clear data showing benefits of LDLT for advanced HCC, the medical community should take a conservative moral stand and limit the use of LDLT for HCC that meets the same criteria as DDLT. LIVING DONOR LIVER TRANSPLANTATION FOR HCC: WESTERN EXPERIENCE Over the past 10 years, LDLT has developed as an alternative to DDLT because of the scarcity of deceased donor livers. Patient survival after LDLT is similar to that after DDLT, despite data from the UNOS database showing lower post-LDLT graft survival, possibly due to the learning curve for LDLT and lower graft-to-recipient body weight ratio.26 In the United States, enthusiasm for LDLT peaked in 2001, when more than 500 LDLTs were performed; the aftermath of a widely-publicized donor death in 2002 led to a nationwide retrenchment, with no more than 325 procedures per year since then. Nevertheless, LDLT has an important role in the treatment of HCC. While previously there was concern over possible tumor stimulation due to regeneration after LDLT, it appears that the type of graft (living vs deceased donor) has little, if any, impact on post-OLT tumor progression. Typically, LDLT results in a liver graft that is smaller than the expected liver volume of the recipient. The outcome of a relatively small donor organ depends not only on graft size but also on the recipient’s preoperative degree of liver failure and portal hypertension. Compared with patients awaiting transplant for end-stage cirrhosis, candidates with HCC generally have better-preserved liver function and less portal hypertension, and are thus better able to tolerate a small graft. Ready availability is the most important advantage of LDLT for HCC. Posttransplant tumor recurrence is tied to pretransplant tumor stage; progression of tumor while awaiting transplant can only worsen the prognosis. This fact is reflected in the United States allocation scheme for deceased donor livers, which accords priority to patients with HCC meeting the Milan criteria. Since adoption of the Model of End-Stage Liver Disease organ allocation policy in 2002, candidates with T2 HCC (1 nodule between 2 and 5 cm; or 2–3 nodules, all ≤3 cm) have enjoyed priority such that in many regions of the United States, waiting times are short, obviating the need for LDLT.27 In some regions, however, longer waiting times and attendant higher dropout rates support the use of LDLT in cases of HCC within the Milan criteria. The Milan criteria were adopted because they identify a subgroup of candidates with HCC for whom transplant results are similar to those in patients transplanted for end-stage cirrhosis without HCC. It is widely recognized, however, that many patients with HCC beyond the Milan criteria can be cured by OLT.8,9 In the United States, adoption of expanded priority criteria for DDLT, as proposed by the UCSF group, is currently under consideration. LDLT can be undertaken in candidates who do not meet criteria for waiting list priority. Broadening criteria based on tumor size and number will inevitably lead to more recurrence; nevertheless, Roayaie et al. demonstrated 55% freedom from recurrence at 5 years in patients transplanted with HCC between 5 and 7 cm and no macroscopic vascular invasion.28 Based on data such as these, many centers offer LDLT to patients with HCC meeting expanded criteria that are estimated to yield a 5-year survival of approximately 50%.5 SELECTION IN HCC: BIOLOGY OR MORPHOLOGY The distinction between biology and morphology in simple terms can be described as behavior versus appearance. Ultimately, the behavior of HCC is the final deciding factor on patient outcome. The ability to predict the biology of HCC is desperately needed for patient selection in OLT. The current staging criteria for HCC takes into account tumor morphology but not tumor biology, with transplantable stages stratified according to tumor size and number. For treatment options other than liver transplantation (i.e., liver resection, chemotherapy, and cytoablative therapies), tumor morphology as described by the pathological TNM staging system is reasonably adequate. For liver transplantation, however, gross and microscopic morphology are inadequate for predicting outcomes because of the scarcity of the donor organs that must be judiciously allocated. The issue would be less significant if resources were unlimited. Ideally, the TNM or morphology-based staging system would be able to stratify patients into homogeneous groups with the same outcomes. However, the current staging systems are not precise enough to segregate patients with HCC and cirrhosis into such homogeneous groups. As almost all patients with gross invasion of the hepatic venous system, positive lymph nodes, or metastatic disease experience recurrence and are therefore not candidates for transplantation, the N and M components of the TNM classification can immediately be eliminated from the organ allocation scheme. This leaves only the T component upon which we can base our biological prediction, namely tumor size and number. In our efforts to refine our current staging systems, morphological or chemical factors have been studied, including microscopic vascular invasion, encapsulation, plasma albumin mRNA, and serum alpha-fetoprotein.5 To date, however, none of these have improved upon the current systems. Part of the problem has been radiologic techniques are not yet sufficient to perfectly define tumor number and size and will never be able to determine microscopic vascular invasion. Moreover, many new staging systems require pretreatment biopsy, which poses a small but real risk of tumor seeding. To understand the biological behavior and identify genes associated with survival after OLT, Marsh and colleagues at the University of Pittsburgh performed microdissection on explanted tissue and studied DNA mutations near 9 tumor-associated gene loci to create an index of cumulative mutational damage, termed the fractional allelic imbalance (FAI).29,30 They found that FAI and vascular invasion were the strongest independent predictors of tumor-free survival. Thus, incorporation of gene mutational data allows desegregation of HCC patients from imprecise, morphology-based staging systems and allows improved prognostication. CONCLUSIONS While patients with early resectable HCC and preserved hepatic function should undergo surgical resection, in those with unresectable disease due to underlying liver dysfunction, orthotopic liver transplantation (OLT) offers the best chance for cure. In the past 10 years, results of OLT have steadily improved because of careful patient selection pioneered by the introduction of the Milan criteria in 1996. Supported by studies showing that many patients with tumor stage beyond the Milan criteria can be cured by OLT, a number of expanded criteria have been proposed. While expanding the criteria for OLT allows more patients to be eligible for transplantation, arguments against expanding the criteria include the increased risk of vascular invasion and tumor recurrence at higher stages of HCC. The principal limitation of OLT for HCC is the shortage of deceased donor living grafts, especially in Asia. The development of LDLT has allowed more patients to benefit from OLT with favorable preliminary survival results. Given the shortage of organs available for OLT and lack of predictive power of currently used staging systems, improved prognostic tools are needed to predict outcomes after OLT. Molecular markers of cancer progression may add significant discriminatory power to the current staging systems and may improve organ allocation schemes for patients with HCC.

Pediatr_Radiol-4-1-2292499

Sonographic findings in bacterial meningitis in neonates and young infants

Cranial sonography plays an important role in the initial evaluation of infants with suspected bacterial meningitis and in monitoring for complications of the disease. Echogenic widening of the brain sulci, meningeal thickening and hyperemia suggest the diagnosis in an at-risk population. Sonography can identify the presence of extra-axial fluid collections, and color Doppler sonography can be very helpful in differentiating benign enlargement of subarachnoid spaces from subdural effusions. Intraventricular debris and stranding, and an irregular and echogenic ependyma are highly suggestive findings associated with ventriculitis. Sonography can play an important role in the detection of postinfectious hydrocephalus, in the determination of the level of obstruction, and in the evaluation of intracranial compliance. Focal or diffuse parenchymal involvement can represent parenchymal involvement by cerebritis, infarction, secondary hemorrhage or early abscess. Introduction Bacterial meningitis is an uncommon but serious condition in the newborn and infant child. The incidence of bacterial meningitis has shown a dramatic decline in the last 25 years in the United States [1]. But despite a number of recent advances in neonatal intensive care, rapid bacteriological assays, and potent antibiotics, neonatal bacterial meningitis remains an important cause of neurological disability worldwide [2–4]. Early diagnosis of the primary disease and its complications is essential for the prevention of neurodevelopmental sequelae. Cranial sonography is often the first imaging modality used in the evaluation of critically ill infants with suspected meningitis. Thus, familiarity with the pathophysiology and sonographic appearance of bacterial meningitis is important for the radiologist and neonatologist taking care of these patients. In this article we review and update the US features of bacterial meningitis and its complications with emphasis on the relationship between imaging findings and underlying pathological changes. Pathophysiology of meningitis Bacterial meningitis usually occurs as a result of bacteremia and sepsis, with initial seeding of the central nervous system via the choroid plexus. The infection then spreads into the cerebrospinal fluid (CSF) and can cause inflammation of the ventricular system (ventriculitis). Inflammation of the meninges typically follows [2, 5]. A key feature of arachnoiditis is the presence of vasculitis, predominantly in small and medium-size veins that traverse the subarachnoid space [5]. Meningeal inflammation extends to the walls of the bridging and cortical veins to cause thrombophlebitis and vascular occlusion, which is associated with cortical infarcts [5, 6]. A number of bacterial virulence factors and inflammatory cytokines contribute to the inflammatory process, resulting in diffuse cerebral edema, increased intracranial pressure, and alterations of cerebral blood flow [7–9]. Lipopolysaccharides found in the bacterial cell wall can lead to a disruption in the blood–brain barrier and invasion of microorganisms into deeper cerebral structures [10, 11]. Arachnoiditis and ventriculitis might also contribute to a more generalized encephalopathy characterized by neuronal loss, gliosis and periventricular leukomalacia [2]. Later pathological findings include diffuse cerebral cortical and white matter atrophy, hydrocephalus, which can also be seen in the early phase, multicystic encephalomalacia, porencephaly and ventricular septations [2, 3, 5, 12, 13]. In the acute phase of meningitis, marked vasodilatation of cortical vessels is often present as the result of potent vasogenic effects of bacterial toxins. In later stages of the disease, there is a global reduction in cerebral blood flow attributed to impairment of cerebral autoregulation and brain edema [14]. In the newborn infant, streptococcus group B, Escherichia coli, and Listeria monocytogenes are the most commonly isolated organisms in CSF. Although other gram-negative enteric organisms such as Klebsiella species and Citrobacter species are less common, they can have far more devastating consequences [15]. Sonographic features of meningitis Sonographic abnormalities are present in approximately 65% of infants with acute bacterial meningitis [3, 16–22]. However, the frequency of imaging abnormalities in patients with a clinical presentation complicated by persistent seizures, abnormal neurological findings and a deterioration of CSF examination within the first 48 h can be as high as 100% [17]. The high incidence of positive sonograms in infants with suspected complications resulted in the recommendation by some authors to use sonography only when there is a clinical suspicion of a complication [3]. Han et al. [3] found that no infant without a clinical indication of complications of meningitis had clinically significant findings on sonography. However, other authors suggest a sonographic examination should be used on every infant with or without evidence of complications on the basis of reports of a high incidence of positive findings in infants without any clinical indication of complications [17, 23, 24]. Rosenberg et al. [23] suggest an initial examination at the time of diagnosis of bacterial meningitis and a follow-up study within 1 week if the initial study demonstrates any parenchymal or ventricular abnormality. We recommend cranial sonography as a baseline study in every infant who has an adequate size fontanel if the diagnosis of bacterial meningitis is suspected clinically. Because the major advantage of sonography is its ability to be safely repeated, a second study should be performed if any clinical deterioration occurs, such as increasing head circumference, occurrence of new neurological findings, and lack of response to therapy. However, in patients with complicated bacterial meningitis whose clinical situation is sufficiently stable to leave the nursery, MRI should be the next study of choice. In children, MRI is widely accepted for its safety and high soft-tissue resolution. It is superior to sonography in showing the existence and extension of the complications, especially in the posterior fossa. It has the further advantage of not using ionizing radiation. It is essential to use state-of-the-art equipment with optimized settings including linear, curved and vector probes with variable megahertz in order to perform a complete evaluation of the meninges because of the often subtle sonographic changes of bacterial meningitis [25, 26]. Meningeal and extra-axial findings In healthy individuals, the pia-arachnoidal membrane is seen as an echogenic line on the surface of the brain. Normal thickness of the membrane measured from the surface of a frontal gyrus (single layer) and within a sulcus (double layer) should not exceed 1.3 mm and 2 mm, respectively [27]. Echogenic widening of brain sulci, or meningeal thickening, is the most common and earliest sonographic sign of meningitis, seen in 26–83% of affected patients (Fig. 1) [3, 16–20, 28]. Prominent cortical vessels can be seen within the pia-arachnoid on color Doppler sonography (Fig. 2). These findings reflect the intense inflammatory exudate that accumulates in the depths of the fissures and sulci, especially around the pial and subarachnoid vessels [5]. Despite their common occurrence, signs of meningeal inflammation are temporary and unrelated to any significant clinical or neurological outcome [3, 16, 20]. Fig. 1Thickened meninges. a Coronal sonogram through frontal lobes in an 11-week-old infant with group streptococcus group B meningitis demonstrates diffuse echogenic thickening of the leptomeninges (arrows) and prominent extra-axial fluid spaces (asterisk). b, c Coronal sonogram on a different infant with group streptococcus group B meningitis (b) shows marked thickening of the suprasellar cisterns (arrows), compared to a normal infant (c)Fig. 2Vasodilatation of pial vessels. Angled sagittal midline color Doppler sonogram in a 3-week-old infant with streptococcus group B meningitis shows prominent vascularity along the surface of the brain parenchyma and a patent superior sagittal sinus (arrowheads) Accumulation of extra-axial fluid might be present in 8–33% of infants with bacterial meningitis. Most commonly, these represent sterile, reactive subdural effusions that have no prognostic significance [3, 16, 19, 20, 28]. On sonography, they appear as hypoechoic concave fluid spaces, sometimes containing mobile echogenic debris (Fig. 3). Empyema is a very rare complication of meningitis, present in up to 1% of affected patients [3, 5, 17, 24]. Although it is often difficult to differentiate a sterile effusion from an early empyema [28], an enlarging complex extra-axial fluid space is a concerning finding. Such spaces can become quite large and cause a mass effect on the adjacent brain parenchyma [29] (Fig. 4). Clinical findings including persistent fever, new focal neurological signs or seizures are not sufficient for suspicion of subdural empyema because they are not specific for the diagnosis of empyema [30]. Hence, a combination of imaging modalities and clinical evaluation is necessary for the early diagnosis of subdural empyema. The final diagnosis is based on the analysis of the subdural effusion [31]. Fig. 3Subdural effusion. a Coronal midline sonogram in a 3-week-old infant with streptococcus group B meningitis shows echogenic debris within subdural fluid. b Axial diffusion-weighted MR image shows markedly hyperintense signal in the extra-axial fluid consistent with restricted diffusion caused by meningeal edema or pusFig. 4Subdural empyema. a Angled sagittal sonogram in a 1-week-old infant with E. coli meningitis shows thickened septations in the subdural space (arrows). b Repeated sonogram at 2 weeks of age shows a complex heterogeneous collection consistent with focal empyema (arrows). Needle aspiration confirmed the presence of pus The distinction between subarachnoid and subdural location of extra-axial fluid is important in the management of infants with meningitis. Isolated subarachnoid fluid spaces up to 3.3 mm are considered incidental findings of no clinical importance in meningitic infants, whereas neurosurgical evacuation is the therapy of choice for infected subdural effusions (empyema) [25, 32]. Early detection with frequent sonographic monitoring of subdural effusions might lessen the neurological sequelae in infantile and neonatal purulent meningitis [33]. Color Doppler sonography can be very helpful in differentiating benign enlargement of subarachnoid spaces from subdural effusions [34, 35]. When subarachnoid fluid is present cortical vessels on the brain surface are surrounded by fluid, whereas fluid in the subdural space compresses the cortical vessels along the surface of the brain (Fig. 5). Fig. 5Subdural fluid. Midline coronal color Doppler sonogram in an 11-week-old infant with group streptococcus group B meningitis shows cortical vessels along the surface of the left hemisphere (blue vessels). Vasculature is located in the pia-arachnoid space and is not surrounded by fluid. This finding strongly suggests that fluid is in the subdural space. Note the superior sagittal sinus (asterisk) Intraventricular findings The most common sonographic signs of ventriculitis include an irregular and echogenic ependyma, the presence of intraventricular debris and stranding, often associated with ventricular dilatation. Debris in the ventricles can be caused by any infecting organism; however, it is most frequently seen with E. coli meningitis [3, 36]. The amount and coarseness of intraventricular debris can vary considerably (Fig. 6) and reflects the extent of ventriculitis. Sonographic resolution of intraventricular debris reflects the efficacy of antibiotic treatment [37]. Inflammation within the ventricles can cause intraventricular adhesions and formation of septae as a more chronic complication of meningitis in about 10% of patients [21, 28, 38]. These adhesions can cause compartmentalization in the ventricles, leading to intraventricular cyst formation (Fig. 7). The identification of the ventricular septations is very important in planning for appropriate shunt placement. Sonography is the study of choice for the demonstration of these septations, often demonstrating them better than CT [38]. Fig. 6Ventriculitis. a–c Coronal (a), left sagittal (b), and mastoid sonograms (c) obtained in a 12-day-old premature infant with E. coli meningitis show extensive septations and echogenic debris clots filling the lateral, third and fourth ventricles. Note the echogenic thickening of edematous ependyma (arrows). d Repeat coronal sonogram at 19 days of age shows a decrease in size and heterogeneity of the intraventricular debris. Note progression of ventricular dilatationFig. 7Cyst formation. Coronal image obtained in a 5-week-old infant with E. coli meningitis shows a midline cyst displacing the third ventricle (arrow) superiorly and to the right Although histopathological studies have shown exudative inflammation of the choroid plexus (plexitis) and of the ependymal lining (ventriculitis) in up to 93% of those autopsied [5, 13], only those with more severe involvement exhibit abnormal findings on cranial sonography [3, 16, 20, 28]. By US examination, the choroid plexus can appear more echogenic and its contours irregular and poorly defined [16, 28, 38, 39]. The ependyma can have a thick, irregular and echogenic appearance, as well [20, 28, 37–39]. Parenchymal findings Abnormal parenchymal echogenicity Areas of abnormal brain echogenicity have been reported in 12% to 65% of infants with bacterial meningitis [3, 20, 28]. Lesions can be focal or diffuse and can represent parenchymal involvement by cerebritis, infarction, secondary hemorrhage or early abscess [3, 17, 40]. The presence and size of parenchymal lesions are associated with significant neurological sequelae and are indicators of a poor prognosis [3]. One of the most devastating complications of bacterial meningitis is abscess formation. It has been reported with a frequency of between 1% and 18% in newborns with meningitis [2, 3, 16, 21, 28]. Mortality rates are significant (15–75%), and up to 66% of survivors develop neurological sequelae [41, 42]. Most cerebral abscesses are caused by a member of the Enterobacteriaceae family (Citrobacter and Enterobacter species) [43, 44]. Citrobacter koseri (formerly C. diversus) is the main causative agent leading to abscess formation in the brain [43, 45]. Up to 77% of infected patients develop an intracerebral abscess. The major pathological feature of Citrobacter meningitis is vasculitis followed by infarction, with necrosis and liquefaction of large portions of the white matter of the hemispheres [45, 46]. Hemorrhagic necrosis and liquefaction can also occur [43, 46]. Surgical drainage of these lesions is not recommended to avoid further damage to surrounding brain parenchyma [46]. The sonographic appearance of abscess varies with the stage of the infection. During the initial stage of cerebritis, sonography might only show a poorly marginated area of increased echogenicity with increased vascularity using color or power Doppler [2] (Fig. 8). As the abscess matures the lesion becomes a well-circumscribed, complex solid mass with highly echogenic walls [3, 21]. After progression with/or without antibiotic therapy the lesions progress to cavitation with marked peripheral hyperemia [16, 21, 47, 48]. Sonographic differentiation between hemorrhagic necrosis and abscess is not always possible, and percutaneous sampling of the intracavitary fluid might be necessary. Sonographic guidance can be useful for puncture, aspiration, and irrigation of the abscess cavity [49, 50]. Fig. 8Liquefactive cerebritis. a, b Coronal (a) and sagittal right (b) sonograms through the frontal lobes in a full-term infant with Citrobacter meningitis show extensive echogenic cerebritis involving the right frontal lobe and basal ganglia. c Repeated sonogram 5 days later shows early liquefaction of the frontal lobe (arrow). d, e Coronal sonogram obtained after 7 days (d) demonstrates complete liquefaction of involved brain and surrounding hyperemia on color Doppler sonography (e). Needle aspiration revealed necrotic tissue with no organism Brain edema In the acute stage of meningitis, both intra- and extracellular edema can occur. As a result, the lateral ventricles as well as the other cisternal structures can be compressed or obliterated [5]. The range of ventricular size is quite wide, from mild compression to complete sonographic obliteration, the latter being a poor prognostic factor [28, 51]. On sonography, there might be a diffuse or heterogeneous increase in the echogenicity of the brain with effacement of the sulci and gyri [20, 23, 28, 50, 52] (Fig. 9). Increased intracranial pressure is associated with increased pulsatility of arterial flow, reflected in elevated resistive index (RI) values on pulsed Doppler sonography. Other causes of elevated pulsatility of flow include venous thrombosis (Fig. 10), large extra-axial fluid collections, and postinfectious hydrocephalus. Retrograde blood flow during diastole is an indication of markedly altered intracranial compliance [53–57]. When equivalent antegrade and retrograde blood flow is present during numerous cardiac cycles, this might be an indication of nonviable brain blood flow [35, 58]. Fig. 9Brain edema. Coronal sonogram in a 7-day-old infant with group B streptococcus meningitis shows generalized edema with compression of lateral ventricles and sulci and diffuse increased heterogeneity of the cerebral hemispheresFig. 10Sagittal sinus thrombosis, nonviable blood flow. a Coronal color Doppler image an 18-day-old infant with group B streptococcus meningitis shows flow limited to internal carotid arteries at the base of the brain. b On the magnified coronal color Doppler image no flow is identified in the superficial sagittal sinus (arrows), and the thrombosis of the sinus was also confirmed on sagittal scans using high-sensitivity Doppler settings (not shown). c Duplex Doppler waveform of the anterior cerebral artery without fontanel compression shows reversal of flow during the entire diastolic phase Diffuse cerebral atrophy, multicystic encephalopathy, and porencephaly might be seen as the end-stage of complications of bacterial meningitis. Hydrocephalus Ventricular dilatation is present in 14–65% of infants with bacterial meningitis and can occur in either the acute or the chronic phase of the disease [16–21, 28]. The level of obstruction to CSF flow is usually outside the ventricular system but can occur within the narrowest portions of the ventricles as a result of ependymitis and secondary adhesion. Common locations for intraventricular obstruction include the aqueduct of Sylvius and the exit foramina of the fourth ventricle [16]. Sonography can play an important role in the detection of hydrocephalus, in the determination of the level of obstruction, and in the evaluation of intracranial compliance. In aqueduct stenosis the lateral and third ventricles are disproportionately dilated, the aqueduct is thickened and echogenic and the fourth ventricle is of normal size (Fig. 11). Detection of CSF flow through the aqueduct of Sylvius is possible by imaging the aqueduct using the mastoid fontanel view. Using power Doppler sonography, the anterior fontanel is compressed and quickly released. Particulate debris within the CSF allows power Doppler signal to be detected when CSF flows in a retrograde fashion through a patent aqueduct into the third ventricle [59]. Fig. 11Postinfectious hydrocephalus. a Midline sagittal sonogram in an infant with aqueductal stenosis following group B streptococcus meningitis shows dilatation of the lateral and third ventricles; the fourth ventricle is normal. b Mastoid fontanel view of another infant with group B streptococcus meningitis shows markedly dilated lateral, third, and fourth ventricles secondary to obstruction of the fourth ventricle outlet foramina. Note the echogenic particle in the third ventricle, which could be debris or a clot Pulsed Doppler interrogation of the anterior cerebral artery before and after compression of the anterior fontanel can be a useful adjunct in the serial monitoring of infants with postinfectious hydrocephalus [53, 54, 57]. Increasing response to fontanel compression as manifested by increasing changes in arterial RI can be an indication for shunt placement in infants with rapidly progressive hydrocephalus. Sonographic evaluation of the thoracolumbar spine in infants with meningitis might reveal the presence of echogenic debris within the spinal subarachnoid space. This finding has been associated with a higher risk of progressive hydrocephalus [60]. Conclusion Cranial sonography continues to be an excellent front-line imaging modality in the initial diagnosis and monitoring of infants with acute bacterial meningitis and its potential complications. Its utility is enhanced by the use of alternative scanning approaches and Doppler techniques for identification of ventriculitis, extra-axial fluid collections, and hemodynamic alterations. Spinal sonography might also play a role in assessing the risk of progressive hydrocephalus.

Vision_Res-2-1-2386851

Contrast discrimination: Second responses reveal the relationship between the mean and variance of visual signals

To explain the relationship between first- and second-response accuracies in a detection experiment, Swets, Tanner, and Birdsall [Swets, J., Tanner, W. P., Jr., & Birdsall, T. G. (1961). Decision processes in perception. Psychological Review, 68, 301–340] proposed that the variance of visual signals increased with their means. However, both a low threshold and intrinsic uncertainty produce similar relationships. I measured the relationship between first- and second-response accuracies for suprathreshold contrast discrimination, which is thought to be unaffected by sensory thresholds and intrinsic uncertainty. The results are consistent with a slowly increasing variance. 1 Introduction First applied to psychophysical data by Tanner and Swets (1954), signal-detection theory (SDT) posits that all stimuli elicit some sensation. However, due to noise, sensations experienced in the absence of stimulus can sometimes be more intense than sensations actually elicited by a stimulus. Crucial evidence for these faint hallucinations comes from Swets, Tanner, and Birdsall’s (1961) two-response, four-alternative forced-choice (2R4AFC) detection experiment, in which observers reported both their first and second choices for the temporal position of a visual target.1 SDT predicts that second-guessing should be better than chance, and this is what Swets et al. found. 1.1 Signal-detection theory According to SDT (Green & Swets, 1966), each stimulus X, gives rise to a Gaussian probability-density function (PDF)of sensory intensity x. In its simplest form, all PDF’s have the same standard deviation, i.e., σX = σ,  ∀X. 1.2 Increasing variance Simple SDT proved incapable of explaining Swets et al.’s (1961) 2R4AFC detection experiment. Instead, they proposed that sensation variance increased with sensation mean:Increases in the “sigma-to-mean ratio”2 r = dσX/dμX, produce decreases in both first- and second-response accuracies, but the second-response accuracies to decrease faster. Swets et al. tried several values for this ratio and obtained as good fit to their data when r = 0.25. 1.3 Intrinsic uncertainty Elsewhere (Solomon, 2007), Swets et al.’s (1961) 2R4AFC data have been successfully fit with an alternative model, incorporating intrinsic uncertainty. Intrinsic-uncertainty models posit that perceived intensity depends on the maximum activity in several independent sensory mechanisms, only one of which is actually sensitive to the stimulus. Given a sufficient number M, of these mechanisms, the variance of their maximum activity will decrease as the intensity of the stimulus decreases (see Pelli, 1985, for a graphical demonstration of this). Thus, to some degree, intrinsic uncertainty mimics Swets et al.’s proposal of increasing variance. With suprathreshold stimuli, the maximum activity always occurs in the appropriately tuned mechanism, and the others have no influence on perception. 1.4 Low-threshold theory At odds with SDT is the idea of a sensory threshold, which weak stimuli must exceed to be detected. In a detection task, stimuli that do not exceed the threshold can be selected only when no other stimulus exceeds it, and the observer is forced to make a choice. Swets et al. (1961) developed this idea into a “low-threshold” hybrid of signal-detection and threshold theories. Unlike other models with this name, Swets et al. claimed theirs could fit the 2R4AFC results. Elsewhere (Solomon, 2007), I have corroborated this claim, and shown that the fit is not quite as good as those produced by models including either intrinsic uncertainty or increasing variance. 1.5 This study Neither intrinsic uncertainty (Pelli, 1985; Tanner, 1961) nor Swets et al.’s (1961) low-threshold theory requires sensation variance to increase with sensation mean. These theories are somewhat special because they are thought to affect the visibility of only very faint stimuli. Increasing variance, on the other hand, has implications for suprathreshold contrast discrimination. For this reason, I decided to conduct a 2R4AFC contrast-discrimination experiment. The goal was to obtain an estimate of the sigma-to-mean ratio, which would not be contaminated by intrinsic uncertainty or a low threshold. 2 Methods There were five observers: the author (JAS), another psychophysicist who understood the purposes of the experiment (MJM), two experienced psychophysical observers who were naïve to the purposes of this experiment (FV and MT) and one further observer who had no previous laboratory experience (NN). As described below, NN produced a very high proportion of “finger errors.” This suggested to us a general unreliability, and no further analyses were performed on his data. The Psychophysica (Watson & Solomon, 1997a) software used in these experiments is available at http://vision.arc.nasa.gov/mathematica/psychophysica.html. The 23.5-cd/m2 display (a Sony GDM F-520 CRT) was viewed in a dark room from 1.15 m. Luminances of vertically adjacent pixels were effectively independent, and could obtain any value between 1.06 and 46 cd/m2. Stimuli were horizontal, cosine-phase Gabor patterns whose wavelength and spatial spread were λ = 0.25° and σ = 0.18°, respectively. Stimuli were flashed simultaneously, in four positions, each marked by four dark spots. The centers of these positions formed a 5.6° × 5.6° square centered on fixation (see Fig. 1). On each trial, three stimuli appeared with a pedestal contrast, which varied between blocks of 90 trials each. The contrast of the fourth stimulus was somewhat greater. After each 0.18-s stimulus exposure, observers gave two responses. The first response indicated which of the four positions the observer thought most likely to have contained the high-contrast target. The second responses from JAS and MJM indicated their second choices for the target position. Following their second responses, JAS and MJM received auditory feedback indicating which—if either—of their responses was correct. The naïve observers were not told that three of the four stimuli would have the same contrast. They were instructed merely to indicate their choices for the positions containing the two highest contrasts, in order. This encouraged them to fully consider their second responses, even when they felt confident about their first. The naïve observers received no feedback. Although I was primarily concerned with suprathreshold contrast discrimination, I was also eager to replicate Swets et al.’s (1961) findings at detection threshold. Since I was therefore committed to measuring both the left and right ends of the threshold-vs.-contrast function (Nachmias & Sainsbury, 1974; Fig. 2), I decided to devote a few extra trials to get the middle as well. (Note: due to limited availability, FV performed only the critical conditions, i.e., those with supra-threshold pedestals.) Table 1 shows the number of trials each observer performed with each pedestal contrast. In it, and in the discussion below, I use the conventional decibel scale of contrast energy: if m is the maximum available contrast, then an x dB stimulus is one that has a contrast of m10x/20. Prior to each trial, the quest procedure (Watson & Pelli, 1983) estimated the performance threshold ct, i.e., the contrast increment required for 62%-accurate first responses. This is halfway between chance performance (25%) and a hypothetical ceiling of 99%. As a means of more accurately estimating the ceiling, or equivalently, the frequency of finger errors, a −10 dB increment was used on one-ninth of the trials. For JAS, the target was given an increment of either ct − 2 dB or ct + 2 dB, with equal probability, on the remaining trials. For the other observers, target increments were either ct − 2 dB or ct . This modification allowed better sampling of their psychometric functions. Finally, to further encourage the naïve observers to fully consider their second responses, one of the three alternatives to each of the “obvious” (−10 dB) targets was fixed at −16 dB. 3 Results 3.1 Psychometric functions For each observer and each pedestal, first-response accuracies were maximum-likelihood3 fit with a modified Gaussian distribution.In the preceding expression, c is the increment (in dB), Ψ1 is the probability of a correct first response and f is the PDF defined in Eq. (1). Threshold ct, and σ were free parameters, but the frequency of finger errors δ, was not allowed to vary with pedestal contrast. These psychometric fits were obtained for purely descriptive purposes. Unlike some of the fits described below, these were not driven by any particular model of performance. Best-fitting values for δ were 0.006, 0.002, 0.048, 0.018 and 0.095 for JAS, MJM, FV, MT and NN, respectively. When debriefed, NN reported a tendency to respond before the end of a trial. 3.2 Threshold-vs.-contrast functions (first response only) Fig. 2 shows how threshold varies with pedestal contrast. JAS’s and MJM’s thresholds were similar, and formed the classic “dipper” shaped function. MT did not suffer as much masking. That is, his thresholds with high-contrast pedestals were lower than the JAS’s and MJM’s. However, his detection threshold—obtained with pedestals having zero contrast (or −∞ dB)—was similar: between −24 and −23 dB. FV’s thresholds with high-contrast pedestals fall within the range spanned by the other observers’, thus we can be reasonably confident that these pedestals exceeded her detection threshold, as they did for the other observers. 3.3 Second-vs.-first-response-accuracy functions: Detection Fig. 3 shows how first- and second-response accuracies co-varied when, as in Swets et al.’s (1961) experiment, there was no pedestal. Appendix A contains a full description of the raw data. Several features of Fig. 3 deserve a detailed description. The axes differ from those used by Swets et al. (1961). For the horizontal axes, instead of signal strength, which may not be a linear function of contrast, I prefer first-response accuracy Ψ1. For the vertical axes, Swets et al. used second-response accuracy, divided by the proportion of first-response errors. However, both first- and second-response accuracies are subject to measurement error. When one uncertain statistic is divided by another, the confidence intervals for the quotient are necessarily very large. To avoid this problem, I get rid of the denominator and plot simply second-response accuracy Ψ2. One argument against this way of plotting the results is that a large portion of the graph will be wasted because Ψ2 ⩽ 1 − Ψ1. A dotted line has been added to each graph to indicate this upper limit for second-response accuracy. Similarly, Ψ2 ⩽ Ψ1. Of course, given a finite number of trials, we may find that the frequency of correct second responses P2 exceeds the frequency of correct first responses P1, but only a perverse observer would have a greater probability of being correct in his or her second response. Therefore I have added another dotted line to each graph to indicate this other upper limit for second-response accuracy. Each point in Fig. 3 represents data collected with a unique target contrast. Several of these points reflect only a few responses made at the beginning of the experiment, before the adaptive staircase had converged. We can thus have little confidence in the likelihood of a correct first or second response with these targets. To convey this confidence (or lack thereof), I have plotted 95% confidence intervals, both horizontally and vertically, about each point. These intervals are based on binomial probabilities, calculated from the range defined by the limits described in the preceding paragraph (see Appendix B for details). 3.4 Modeling finger errors SDT can be modified to accommodate finger errors. Let ψ1 denote the first-response accuracy without errors. For those trials containing a first-response finger error, the probability of a correct first response is (1 − ψ1)/3. Thus, if the finger-error rate is δ, the overall probability of a correct first response is (1 − δ)ψ1 + δ(1 − ψ1)/3. To derive the formula for second-response accuracy, it helps to understand that first-response finger-errors will be incorrect with probability 1 − [(1 − ψ1)/3] = (2 + ψ1)/3. Without loss of generality, we may assume that observers correct some proportion ε of first-response finger errors with their second response. Thus on these trials, the probability of a correct second response is ψ1. When first-response finger-errors are not explicitly corrected, I will assume that the second response is completely random. On these latter trials, the second response will be correct with probability [(2 + ψ1)/3]/3 = (2 + ψ1)/9. Thus, on those trials in which a first-response finger error occurred, the second-response accuracy should beSecond-response accuracy overall will be (1 − δ)ψ2 + δ [εψ1 + (1 − ε)(2 + ψ1)/9], where ψ2 would have been the second-response accuracy, had there been no first-response finger errors. To estimate ε, trials containing an “obvious” (−10 dB) increment and an incorrect first response were examined. (Because the highest pedestals had the potential for masking even these large increments, they were excluded from this analysis.) On these trials, we may assert that ψ1 = 1, and we can then solve Eq. (4) for ε. Solutions were 0.70, 0, 0.58 and 0 for JAS, MJM, FV and MT, respectively. 3.5 Maximum-likelihood fits These values of ε were assumed when calculating the curves in Figs. 3–6. In Fig. 3, the solid black curves represent the prediction of simple SDT, that is, when r = 0 in Eq. (2). The dashed lines represent the prediction of high-threshold theory, which ascribes all errors to unlucky guesses rather than faint hallucinations. (See Swets et al., 1961 or Solomon, 2007, for derivations of these predictions.) Maximum-likelihood fits were also obtained for three further modifications of SDT. They were: (i) increasing variance with power-law transduction, (ii) intrinsic uncertainty and (iii) low-threshold theory. Details of these three models can be found in Appendix C, and receiver-operating characteristics for all three models applied to a yes/no-detection task appear in Nachmias (1972). The fits appear as solid green, blue and red curves, respectively, in Fig. 3. Goodness-of-fit is indicated by the generalized (log) likelihood-ratios in Table 2. Specifically, these values reflect the maximum log likelihoods, minus the conventional upper-bound on log likelihood described in Footnote 3. Note that unlike some generalized likelihood-ratios (Mood, Graybill, & Boes, 1974), these cannot be expected to follow the chi-square distribution because there are so many conditions (i.e., specific increment contrasts) containing only one or two trials (Wichmann & Hill, 2001). Only for JAS, low-threshold theory produced, by far, the best fits of the three SDT models. In fact, the best-fitting “low” threshold for JAS was effectively a high threshold, never exceeded by zero-contrast pedestals. That is why JAS’s red curve in Fig. 3 is visually indistinguishable from the dashed black line. Since JAS’s results are consistent with a high threshold, they are inconsistent with the findings and conclusions of Swets et al. (1961). For the other two observers, the sigma-to-mean ratios of the best-fitting increasing-variance models were between 0.26 and 0.28; similar to Swets et al.’s (1961) estimate of 0.25. Thus, these results can be considered a successful replication of Swets et al.’s. At the end of this section, I speculate on why JAS’s results differ. 3.6 Transducer-independent estimates of r At detection threshold, as described above, the raw data were simply fit with a multiplicative-noise/power-law-transducer version of SDT. However, with suprathreshold pedestals, we cannot be certain what shape the transducer takes. Increasing-variance models (e.g., Kontsevich, Chen, & Tyler, 2002) use a simple power-law transducer, but constant-noise models (e.g., Legge & Foley’s, 1980), use a transducer that is initially expansive, then compressive, as the pedestal increases. The compressive non-linearity is required to produce masking; i.e., threshold elevation from high-contrast pedestals. We do not really have to worry about the form of transducer, because SDT’s predictions for the relationship between first- and second-response accuracies are independent of signal transduction (Solomon, 2007; Swets et al., 1961). We merely need to quantify how these predictions change with the sigma-to-mean ratio, and find the values most consistent with the data. This examination of a transducer-independent facet of contrast-discrimination data is complementary to attempts at modeling contrast-discrimination data without putting any constraint on the form of the transducer (Katkov, Tsodyks, & Sagi, 2006a; Katkov, Tsodyks, & Sagi, 2006b; Klein, 2006). Transducer-independent estimates of the sigma-to-mean ratio r were obtained by maximizing the likelihood of observing second-response accuracies P2, given the first-response accuracies P1. A complete description of this process appears in Appendix D. The best fits were 0.56 for JAS, 0.32 for MJM and 0.31 for MT. These values are illustrated in Fig. 4. All of these fits are comparable to the maximum-likelihood fits described above. 3.7 Binning accuracy Fig. 4 is less cluttered than Fig. 3. It has fewer data points and no horizontal confidence intervals. Nonetheless, the same data appear in both figures. For legibility, I have decided to combine data from increments producing similar first-response accuracies, as determined by the psychometric functions described above. I have adopted the relatively arbitrary decision to use 5 dB bin-widths. There is one visible consequence of this manipulation: a rightward shift for one data point in JAS’s panel. However, without binning, the suprathreshold data presented below would be impossible to read. I have also decided to forgo plotting data from increments producing both floor (i.e., 0.25) and ceiling (i.e., 1 − δ, see Eq. (3)) accuracies. Such data are worthless at discriminating between candidate models. Finally, also for the sake of legibility, I have also decided to cull data points representing fewer than 10 trials. Of course, these latter data are not completely worthless; they have not been excluded from any model fits. A complete description of the binned data appears in Appendix A. 3.8 Second-vs.-first-response-accuracy functions: Suprathreshold discrimination Fig. 5 shows how first- and second-response accuracies co-varied for the four highest pedestals. (The binned data are tabulated in Appendix A.) Plotting conventions have been inherited from Fig. 4. It should be apparent that the second responses with these suprathreshold pedestals tend to be more accurate than the second responses at detection threshold. Using the same procedure as was described for the detection data (above and in Appendix D), values for the sigma-to-mean ratio r were found that maximized the likelihood of observing second-response accuracies P2, given the first-response accuracies P1. The best fits were 0.11 for JAS, 0.16 for MJM, 0.21 for FV and 0.09 for MT. These values are illustrated in Fig. 5. These values of r are smaller than those required to fit the detection data. FV’s relatively high value may have something to do with her relatively high finger-error rate. No detection data from FV are available for comparison. Pedestal-by-pedestal estimates of r appear in Fig. 6. Most of these estimates remain near or below the value of 0.25, selected by Swets et al. (1961), when neither the effect of a low threshold nor that of intrinsic uncertainty is considered. The results from JAS are different; best-fitting values of r start at 0.56, and decrease to 0 as pedestal intensity increases. JAS’s small-pedestal estimates of r are strangely high. As noted above, they are not incompatible with the notion that visual noise never exceeds the threshold of visibility in the absence of a stimulus. Previous attempts to replicate Swets et al.’s (1961) 2R4AFC results have also met with mixed success. Despite their chronological precedence, Kincaid and Hamilton’s (1959) results have been described as both successful and unsuccessful replications of Swets et al.’s (Green & Swets, 1966 and Blackwell, 1963, respectively). As yet, I have not been able to track down a copy of Kincaid and Hamilton’s publication. One further attempt to replicate Swets et al. was described by Eijkman and Vendrik (1963). They reported that second responses for light detection were greater than chance in just one of three observers. An alternative interpretation for JAS’s high-threshold-like performance is that he simply ignored sensory information and selected his second responses more-or-less randomly. This explanation would be easier to swallow if all his estimates of r were high. However, with suprathreshold pedestals, his r is no higher than that of the other observers. Expectation may have caused JAS to change his strategy with pedestal intensity, but I have no definitive answer. 4 Discussion The most equitable summary of these results is that they are consistent with a performance-limiting source of noise, which increases slightly with suprathreshold contrast. Best estimates for the rate varied from 0.09 to 0.21. I wondered whether such small sigma-to-mean ratios would be sufficient to model the high contrast-discrimination thresholds obtained with high-contrast pedestals, or whether a saturating transducer function for stimulus contrast would also be necessary. Previous attempts (Kontsevich et al., 2002; Solomon, 2007) to fit contrast-discrimination data without compressive transduction have not focused on the minimum necessary sigma-to-mean ratio, but that ratio can be inferred from the published parameter values. 4.1 Fitting contrast discrimination In those previous attempts, the standard deviation of visual signals was allowed to increase as a decelerating power function of the mean. Specifically,(Kontsevich et al. considered only suprathreshold contrast and thus could set σ0 = 0; Solomon used σ0 = 1.) Therefore, the sigma-to-mean ratios decreased as the means increased. From the best-fitting parameter values, I have used a contrast of 100% to infer the minimum sigma-to-mean ratios required to explain contrast discrimination without compressive transduction. The smallest of these ratios was 0.13 (obtained using the parameter values fit to observer SV with “sustained” stimuli in Kontsevich et al., 2002). Thus it seems that the sigma-to-mean ratios estimated in the present study may in fact be able to produce appreciable masking. Four different models were maximum-likelihood fit to all of JAS’s, MJM’s and MT’s 2R4AFC data. Details of all four models appear in Solomon (2007). One of these models was a four-parameter, non-linear transducer model with constant Gaussian noise (i.e., where r = 0). This transducer is initially expansive, then compressive, as the input increases. Foley (1994) obtained good fits to 2AFC contrast-discrimination thresholds with this model, but we already know this model will over-estimate second-response accuracies, particularly in a (zero-pedestal) detection experiment (Solomon, 2007; Swets et al., 1961). In the other three models that were fit to JAS’s, MJM’s and MT’s data, the variance of visual signals was allowed to increase with the mean. Two of these models have already proven capable of producing acceptable fits to some of Foley’s (1994) 2AFC thresholds (see Fig. 13 of Solomon, 2007). In one of these, a power-law transducer is responsible for the ‘dip’ in the contrast-discrimination function. In the other, intrinsic uncertainty produces the dip. The remaining increasing-variance model explored in this paper uses a low threshold to produce the dip. Fit details appear in Table 3. Initially, four parameters were allowed to vary freely in each fit. For JAS and MJM the best-fitting model was the one with increasing variance model and intrinsic uncertainty. In general, the fits improved as uncertainty M, increased. For JAS and MJM, the fits of the increasing-variance/intrinsic-uncertainty model remained superior when M was fixed at a value of 10,000. When the relationship between signal mean and standard deviation was forced to be linear (i.e., q = 1 in Eq. (5)), the fits of the increasing-variance/intrinsic-uncertainty model remained superior. They even remained superior when M was fixed at a value of 1000. Thus, all of JAS and MJM’s data can be satisfyingly summarized by a relatively simple model, combining increasing variance with intrinsic uncertainty. Within the context of this model the best-fitting values of r (regardless of constraints on M or q) were 0.16 for JAS and 0.14 for MJM. These values are nearly identical to the transducer-independent estimates (0.14 for JAS and 0.16 for MJM), described above. When MT’s data were fit with a 3-parameter increasing-variance/intrinsic uncertainty model, the best-fitting values for the sigma-to-mean ratio and intrinsic uncertainty were r = 0.11 and M = 440, respectively. However, his data are best-fit by the constant-variance, 4-parameter, nonlinear-transducer model (Foley, 1994). Thus, in two out of three cases, the contrast-discrimination data can be satisfyingly summarized by a 3-parameter model of intrinsic uncertainty and increasing variance. Compressive transduction is not required. 4.2 Sensory thresholds for contrast discrimination Although “sensory threshold” means different things to different people (Swets et al., 1961), it is usually understood to be some sort of barrier weak stimuli must overcome to be perceived (Swets, 1961). However, it is no less valid to apply the concept to the task of contrast discrimination, even with large pedestals. That is, forced-choice errors may occur simply because all the alternatives appear identical and the observer simply guesses incorrectly. Inspection of Fig. 5 should be sufficient to rule out any “high” sensory threshold for contrast discrimination. The data shown there are even less similar to the (dashed) high-threshold prediction than Swets et al.’s (1961) detection data (not shown). However, some proportion of correct second responses may indeed have been just lucky guesses, which is to say my data cannot rule out a “low threshold” for “suprathreshold” contrast discrimination. 4.3 Other models The logic of this study hinges on the assumption that unmasked, suprathreshold contrast discrimination can be modeled with a single sensory mechanism. Although this assumption is popular, it is not completely uncontroversial. Before I describe other possible models, I should first stress the importance of the word “unmasked.” In a highly influential paper, Foley (1994) argued that the mechanism responsible for contrast discrimination was not immune to the activity in differently tuned mechanisms. By manipulating the spatial phase, orientation and temporal frequency of masking stimuli, Foley and Boynton (1994) were able to probe the interactions between mechanisms responsible for contrast discrimination. However, when no masks are present, most models (including all of Foley’s) consider contrast discrimination to be mediated by a single mechanism or channel; the one best tuned to the target. There are three notable exceptions. Teo and Heeger (1994) and Yu, Klein, and Levi (2004) have developed models with greater physiological plausibility, in which individual mechanisms have very limited dynamic ranges. Elsewhere (Watson & Solomon, 1997b) I have argued this type of model is well approximated by the more popular, single-mechanism model for contrast discrimination. The two types of model can be considered equivalent when performance-limiting noise is added after the outputs of multiple mechanisms are combined. The third exception was recently proposed by Henning and Wichmann (2007) to account for their finding that the low-contrast “dip” of threshold-vs.-contrast functions (e.g., Fig. 2) disappears in the presence of a notched-noise mask. This result suggests the dip is due to off-frequency looking. That interpretation may be correct, but I am obligated to note their results are also consistent with intrinsic uncertainty, which would attribute a less-pronounced dip to uncertainty reduction. Indeed, Blackwell (1998) argued that noise, both within the detector’s pass band and outside it, could reduce intrinsic uncertainty and facilitate detection. She also provided psychophysical evidence for this facilitation.

Histochem_Cell_Biol-4-1-2228381

Protein quality control: the who’s who, the where’s and therapeutic escapes

In cells the quality of newly synthesized proteins is monitored in regard to proper folding and correct assembly in the early secretory pathway, the cytosol and the nucleoplasm. Proteins recognized as non-native in the ER will be removed and degraded by a process termed ERAD. ERAD of aberrant proteins is accompanied by various changes of cellular organelles and results in protein folding diseases. This review focuses on how the immunocytochemical labeling and electron microscopic analyses have helped to disclose the in situ subcellular distribution pattern of some of the key machinery proteins of the cellular protein quality control, the organelle changes due to the presence of misfolded proteins, and the efficiency of synthetic chaperones to rescue disease-causing trafficking defects of aberrant proteins. Introduction Folding and assembly of proteins and their function depend on each other. Like in industrial production lines, in cells the quality of newly synthesized proteins is monitored in regard to proper folding and correct assembly in the early secretory pathway, the cytosol and the nucleoplasm (Bennett et al. 2005; Bukau et al. 2006; Dobson 2003; Ellgaard and Helenius 2003; Park et al. 2007; Ravid et al. 2006; Roth 2002; Sitia and Braakman 2003; Zhang and Kaufman 2006). Protein quality control is a basic cellular phenomenon through which aberrant proteins become eliminated. Aberrant proteins can occur as waste products at a certain rate during de novo synthesis, or are caused by cellular stress, or due to disease-causing mutations (Aridor and Hannan 2002; Gregersen et al. 2006; Kim and Arvan 1998; Kopito 2000; Lukacs et al. 1994; Petäjä-Repo et al. 2000; Schubert et al. 2000; Turner and Varshavsky 2000; Ward and Kopito 1994). Once recognized as non-native or incompletely assembled, those proteins will be removed and degraded by a process generally termed ERAD, for ER-associated degradation (Hirsch et al. 2004; Meusser et al. 2005). For secretory and membrane proteins, the molecular machinery involved in the recognition, retention and dislocation of aberrant proteins has been identified to a certain detail (Carvalho et al. 2006; Denic et al. 2006; Hirsch et al. 2003; Ismail and Ng 2006; Katiyar et al. 2005; Li et al. 2006; Lilley and Ploegh 2004; Schuberth and Buchberger 2005; Tsai et al. 2002; Ye et al. 2003, 2004). The ensuing final step consists of polyubiquitination of aberrant proteins condemning them for degradation by proteasomes (Eisele et al. 2006; Hochstrasser 1996; Jarosch et al. 2002; McCracken and Brodsky 2005; Wolf and Hilt 2004; Zwickl et al. 2002). In addition to various chaperones aiding proteins to achieve their proper conformation, various machinery proteins are involved in the recognition and retention of aberrant proteins. For glycoproteins, the importance of specific oligosaccharidic structures generated initially by trimming glucosidase II and UDP-glucose:glycoprotein glucosyltransferase and later on by ER-mannosidase I has been recognized (Helenius and Aebi 2004; Parodi 2000; Roth 2002; Roth et al. 2002). Glycoproteins bearing monoglucosylated oligosaccharides will be bound by calnexin or calreticulin. If the aberrant glycoproteins are considered, binding to calnexin or calreticulin will protect them temporarily from degradation. The complete deglucosylation by glucosidase II will result in their exit from the calnexin/calreticulin cycle. Subsequent trimming of mannose residue(s) of the oligosaccharide B branch by ER-mannosidase I opens the gate to dislocation and degradation of aberrant proteins. The link between the calnexin/calreticulin cycle and the dislocation process is apparently provided by two lectin-like proteins: EDEM1 (yeast ortholog Htm1p/Mnl1p) (Hosokawa et al. 2001; Jakob et al. 2001; Kanehara et al. 2007; Nakatsukasa et al. 2001; Oda et al. 2003) and Yos9p (mammalian orthologues OS-9 and XTP3-B) (Bhamidipati et al. 2005; Buschhorn et al. 2004; Gauss et al. 2006; Kanehara et al. 2007; Kostova and Wolf 2005; Szathmary et al. 2005). Depending on the type of protein and the location of the lesion, different ERAD dislocation pathways have been identified (Carvalho et al. 2006; Denic et al. 2006; Ismail and Ng 2006; Schuberth and Buchberger 2005). Aberrant luminal proteins and membrane proteins with a defect in their luminal domain undergo the ERAD-L pathway, which is defined by the E3 ubiquitin ligase Hrd1p complex. The Hrd1p complex consists of several proteins including Hrd3p, an E2 complex (Ubc7p and its membrane-anchoring factor Cue1p), the Cdc48p complex (AAA-ATPase Cdc48p or p97, the Ufd1 and Npl4 cofactors, and the Ubx2p membrane anchor), Der1p, Yos9p, Kar2p (BiP) and Usa1p. It should be noted that the actual function of some of those proteins in the complex remains to be established. Aberrant membrane proteins with lesions in their cytosolic domain enter the ERAD-C pathway organized by the E3 ubiquitin ligase Doa10p complex. This complex is comparably simple and consists in addition to Doa10p only of the E2 complex and the Cdc48p complex. The ERAD-M pathway is followed by membrane proteins with a lesion in their transmembrane domain and involves only Hrd1p and Hrd3p. These dislocation pathways were established for yeast cells but most probably will apply to higher eukaryotes as well because of the evolutionary conservation of the ERAD pathways. The Doa10p complex also operates in the polyubiquitination of aberrant cytosolic and nuclear proteins, in addition to the ERAD-C pathway (Neuber et al. 2005; Ravid et al. 2006; Swanson et al. 2001). This review will focus on how immunocytochemical labeling and electron microscopic analysis have helped to disclose the in situ subcellular distribution pattern of some of the key machinery proteins of the protein quality control, the organelle changes due to the presence of misfolded proteins, and the efficiency of synthetic chaperones to rescue disease-causing trafficking defects of aberrant proteins. Machinery proteins of the protein quality control reside beyond the ER For the quality control of glycoprotein folding, glucosidase II (Gls II) and UDP-glucose:glycoprotein glucosyltransferase (GT) in connection with the calnexin/calreticulin cycle are of eminent importance (Helenius and Aebi 2004; Parodi 2000; Roth 2002). Gls II is a luminal glycoprotein, which exists in two isoforms (Pelletier et al. 2000; Ziak et al. 2001) and does not contain known ER retention signals of the C-terminal KDEL type, nor any hydrophobic region characteristic of transmembrane proteins (Flura et al. 1997; Trombetta et al. 1996). As depicted in Fig. 1a, Gls II acts second to glucosidase I by removing the two inner α1,3-linked glucose residues (Brada and Dubach 1984; Burns and Touster 1982). The presence of three or two glucose residues on oligosaccharides can be considered to represent a trimming glyco-code whereas one glucose residue represents a trimming as well as folding glyco-code (Fig. 1b) (Jakob et al. 1998b). The involvement of Gls II and of mono-glucosylated oligosaccharides generated by the enzyme in the protein quality control is well documented (Hammond et al. 1994; Hebert et al. 1995; Jakob et al. 1998a, b). By confocal immunofluorescence, Gls II not unexpectedly exhibited a pattern typically observed for the ER as shown in Fig. 2b (Roth et al. 2003; Zuber et al. 2001). By high-resolution immunoelectron microscopy, ER localization of Gls II could be definitely established (Lucocq et al. 1986; Zuber et al. 2000, 2001). In addition to the rough ER including the nuclear envelope and the transitional ER, the smooth ER was also positive for Gls II. However, with the superior resolution of electron microscopic immunogold labeling, Gls II was additionally found in tubulovesicular clusters between transitional ER and the cis Golgi apparatus. They represent pre-Golgi intermediates involved in antero- and retrograde transport of cargo (Appenzeller-Herzog and Hauri 2006; Bannykh and Balch 1997, 1998; Hammond and Glick 2000; Palade 1975; Saraste et al. 1987; Schweizer et al. 1988). Fig. 1Schematic presentation of the oligosaccharide trimming pathway by glucosidase I (Gls I), glucosidase II (Gls II), UDP-glucose:glycoprotein glucosyltransferase (GT) and ER-mannosidase I (ER-Man I)Fig. 2Double confocal immunofluorescence for UDP-glucose:glycoprotein glucosyltransferase (a) and glucosidase II (b) demonstrates co-distribution (c) of the two protein quality control machinery proteins in cultured clone 9 hepatocytes. Immunogold localization of UDP-glucose:glycoprotein glucosyltransferase in an ultrathin frozen section of clone 9 hepatocytes reveals immunoreactivity in the rough ER (ER) including nuclear envelope and a pre-Golgi intermediate (pGI). The cisternal stack of the Golgi apparatus (GA) is not labeled Mono-glucosylated oligosaccharides of glycoproteins are targeted by the calnexin/calreticulin cycle and after being deglucosylated by Gls II, will be targeted by GT if not correctly folded (Fig. 1b). GT apparently senses exposed patches of charged amino acids and reglucosylates the aberrant glycoproteins, which is followed by their re-entry in the calnexin/calreticulin cycle (Parodi et al. 1983; Sousa and Parodi 1995; Trombetta and Parodi 2003). When the subcellular distribution of GT was studied by confocal immunofluorescence, its labeling pattern (Fig. 2a) was alike that observed for Gls II (Fig. 2b; Zuber et al. 2001). By immunoelectron microscopy, GT was detectable in the rough ER including the nuclear envelope and the transitional ER as well as the smooth ER (Fig. 2d). Unlike Gls II, for which the labeling intensity over rough and smooth ER was equal, labeling intensity for GT over smooth ER was only 11% that of the rough ER. GT immunolabeling was also discovered in the pre-Golgi intermediates (Fig. 2d). Notably, the pre-Golgi intermediate immunolabeling for GT was approximately twice that of rough ER (Zuber et al. 2001). Double immunogold labeling for GT combined with the pre-Golgi intermediate marker ERGIC-53 and the COPII component sec23p (Hughes and Stephens 2008) proved the identity of the GT-labeled structures. Interestingly, specific immunogold labeling for calreticulin was also observed in the pre-Golgi intermediates (Zuber et al. 2000, 2001). Together, these results provided new insight into the in situ subcellular organization of some key elements of the protein quality control machinery. Gls II, GT and calreticulin were not only present in the rough ER, as expected, but also in the smooth ER and unequivocally present in pre-Golgi intermediates. This pattern was found in different rat cell lines and tissues as well as Drosophila tissue and cell lines. The presence of three functionally closely associated proteins is a strong evidence for the involvement of pre-Golgi intermediates in protein quality control. Of course, immunolocalization provides no direct evidence for the functionality of the detected protein at a certain location. However, there is no reason to assume that Gls II, GT and calreticulin would be only functional in the ER. Studies in yeast have provided strong evidence that multiple, sequentially acting quality control checkpoints exist along the secretory pathway extending as far as to the Golgi apparatus (Arvan et al. 2002; Caldwell et al. 2001; Sayeed and Ng 2005; Taxis et al. 2002; Vashist et al. 2001; Vashist and Ng 2004; Younger and Chen 2006). In mammalian and insect cells, protein quality control is apparently not restricted to the ER, and the pre-Golgi intermediates appear to be involved in this fundamental cellular process as well. As will be discussed later, pre-Golgi intermediates represent not only a quality control checkpoint, but are also sites of accumulation of aberrant proteins. The ERAD factor EDEM1 defines a novel vesicular ER exit pathway As mentioned in the “Introduction”, an impressive body of molecular and functional data exists concerning the macromolecular assemblies involved in the various ERAD pathways. The current well-founded basic conception of ERAD in yeast and higher eukaryotes is that the aberrant proteins after being removed from folding cycles are dislocated to the cytosol and eventually degraded by the 26S proteasome, and that this occurs in the ER. It is not clear whether this is a randomly occurring event or a more structured affair. Recent studies on EDEM1 in mammalian cells have provided preliminary evidence for a high level of subcellular organization. The discovery that the Man8 B isomer oligosaccharide was actively involved in ERAD-L (Fig. 1c) in yeast (Jakob et al. 1998a) and mammalian cells (Liu et al. 1999) paved the way to the identification of a lectin-like protein with sequence similarity to class I α1,2-mannosidases in yeast -Htm1p/Mnl1p- and mammalian cells -EDEM1- (Hosokawa et al. 2001; Jakob et al. 2001; Nakatsukasa et al. 2001). EDEM1 of mammalian cells is a soluble glycoprotein (Olivari et al. 2005; Zuber et al. 2007), which is regulated by the unfolded protein response (Hosokawa et al. 2001) and seems to connect the calnexin/calreticulin cycle to the dislocation process (Molinari et al. 2003; Oda et al. 2003). EDEM1 appears to exist in complex with the dislocation proteins Derlin-2 and -3, and the AAA ATPase p97 (Oda et al. 2006). It is not fully understood how EDEM1 interacts with aberrant proteins. However, there is evidence for interaction with ER-mannosidase I-trimmed oligosaccharides such as depicted in Fig. 1c (Hosokawa et al. 2003). Overexpression of EDEM1 has been shown to prevent formation of dimers of misfolded Null Hong Kong variant of alpha1-antitrypsin (Hosokawa et al. 2006). Notably, EDEM1 and ER-mannosidase I do not exist in complexes, which can be immunoprecipitated (Hosokawa et al. 2003). Recently, the subcellular distribution of endogenous EDEM1 in various mammalian cell types was established with a specific anti-peptide antibody (Zuber et al. 2007). Unexpectedly, its immunofluorescence pattern did not correlate with that of calnexin and other ER marker proteins. Rather, an unusual pattern of well distributed punctate structures along with some localized finger-like structures was revealed (Fig. 3a–c). The distribution patterns of endogenous EDEM1 and that of overexpressed tagged EDEM1 were dramatically different: instead of a punctate, non-ER pattern, a typical reticular ER pattern plus punctate staining was observed (Zuber et al. 2007). This striking difference in subcellular distribution between endogenous EDEM1 and overexpressed tagged EDEM1 was confirmed by Optiprep density gradients. Endogenous EDEM1 was restricted to the densest fractions, whereas tagged EDEM1 showed the same broad distribution as observed for calnexin, sec61β, and Derlin-1 and -2 (Zuber et al. 2007). In this context, it needs to be emphasized that previous biochemical analyses of EDEM1 interaction with quality control machinery proteins and ERAD substrates were performed with cells transiently overexpressing tagged EDEM1 (Hosokawa et al. 2003; Molinari et al. 2003; Oda et al. 2003). The nature of the EDEM1 immunofluorescence pattern was clarified by immunogold labeling and serial section analysis (Fig. 3d–h). It revealed the presence of EDEM1-reactive buds along rough ER cisternae which apparently gave raise to ∼150 nm vesicles. These buds and vesicles were devoid of a COPII coat, formed outside the canonical ER exit sites of the transitional ER and were not found in the tubulovesicular clusters of pre-Golgi intermediates (Fig. 3i). Occasionally, EDEM1 luminal immunolabeling in limited parts of distended ER cisternae was observed, which accounted for approximately 11% of the immunogold labeling for GT. Double confocal immunofluorescence for endogenous EDEM1 in rat hepatoma clone 9 cells stably expressing the Null Hong Kong variant alpha 1-antitrypsin showed co-distribution of the two proteins (Zuber et al. 2007). Together, these data revealed the existence of a vesicular transport pathway out of the rough ER through which the ERAD factor EDEM1 and an ERAD substrate became sequestered from the early secretory pathway. Through this pathway potentially harmful aberrant luminal proteins can be removed. These findings also indicate that the Gls II and GT containing pre-Golgi intermediates appear to be not involved in the dislocation of an ERAD-L substrate. Fig. 3Double confocal immunofluorescence for endogenous EDEM1 (a) and calnexin (b) reveals different distribution patterns for the two proteins (c) in human HepG2 cells. Detection of endogenous EDEM1 by immunogold labeling of ultrathin frozen sections (e) or pre-embedding immunoperoxidase labeling (d, f–h) reveals sparse labeling in the lumen of ER cisternae (arrowheads in e and f) and intense labeling over ER buds and vesicles pinching-off the ER (from Zuber et al. 2007). The subcellular distribution pattern of endogenous EDEM1 is schematically depicted in (i). In contrast to COPII-coated buds, which are formed at the transitional ER (TE) and give raise to COPII-coated vesicles present in pre-Goli intermediates (pre GI), EDEM1-positive buds occur outside the transitional ER and EDEM1-positive COPII-unreactive vesicles form clusters in the cytoplasm Endomannosidase assigns glucose trimming function to the Golgi apparatus It is generally assumed the glucose trimming occurs exclusively by Gls I and II and, therefore, is limited to the ER and pre-Golgi intermediates. However, under conditions of inhibition of trimming glucosidases, formation of mature oligosaccharides has been observed. This apparent paradox could be explained by the existence of an alternate glucose-trimming pathway by neutral endo-alpha-mannosidase (Lubas and Spiro 1987, 1988; Moore and Spiro 1990, 1992; Spiro 2000). Endomannosidase is currently the only known endoglycosidase. In contrast to the trimming Gls I and II, it cleaves internally between the glucose-substituted mannose and the remaining oligosaccharide (Fig. 4a). Its substrate specificity (Fig. 4a) is basically that of Gls I and II (Moore and Spiro 1990, 1992; Rabouille and Spiro 1992). However, unlike Gls I and II, ER-mannosidase I trimmed mono-glucosylated oligosaccharides are a substrate of endomannosidase. The resulting Man8–5 GlcNAc isomer A is the specific product of endomannosidase. It should be noted that this oligosaccharide is no more a substrate for reglucosylation by GT. Biochemically, activity for endomannosidase was found to be enriched in Golgi membranes (Lubas and Spiro 1987). By immunofluorescence (Dong et al. 2000; Zuber et al. 2000), endomannosidase exhibited a crescent-shaped perinuclear staining and fine punctate staining throughout the cytoplasm which partially overlapped with immunofluorescence for Gls Golgi mannosidase II (Fig. 4b–d). High-resolution immunoelectron microscopy demonstrated endomannosidase in the peripheral and Golgi-associated pre-Golgi intermediates as well as cis and medial cisternae of the Golgi apparatus (Fig. 4e) (Zuber et al. 2000). Trans cisternae of the Golgi apparatus and the trans Golgi network were unreactive. Quantification revealed ∼85% of the immunogold labeling for endomannosidase in the Golgi apparatus and ∼15% in pre-Golgi intermediates. Although, both endomannosidase and Gls II could be detected in pre-Golgi intermediates by double immunogold labeling, they labeled different elements of the vesiculotubular clusters (Fig. 4e). Thus, endomannosidase and Gls I and II exhibited non-overlapping subcellular distributions (Roth et al. 2003; Zuber et al. 2000). Functionally, the presence of endomannosidase in the ER would interfere with the action of glucosyltransferase by preventing the reglucosylation of misfolded glycoproteins. Together, these findings demonstrating a predominantly Golgi apparatus localization of endomannosidase strongly indicated that glucose trimming of N-linked oligosaccharides is not limited to the ER. Fig. 4The various oligosaccharidic substrates of endomannosidase are depicted in (a). Like trimming glucosidases, endomannosidase trimms Gls1–3Man9GlcNAc2, and unlike trimming glucosidase II, monoglucosylated oligosaccharides with mannose-trimmed B and C branches. Double confocal immunofluorescence for endomannosidase (b) and Golgi mannosidase II (c) reveals co-distribution of the two enzymes (d) in clone 9 hepatocytes. Double immunogold labeling for endomannosidase (small gold particles, arrowheads) and glucosidase II (large gold particles, arrows) reveals endomannosidase localization in cis and middle Golgi apparatus cisternae (g), whereas glucosidase II is observd in rough ER including nuclear envelope. Non-overlapping immunogold labeling for both enzymes exists in pre-Golgi intermediates. N: nucleus, PM: plasma membrane. Micrographs b–e are from Zuber et al. (2000) Since glucose trimming is indispensable for the synthesis of mature oligosaccharide side chains, deglucosylation by endomannosidase in the Golgi apparatus ensures that this important process is not blocked. Further biochemical and morphological analyses demonstrated that Golgi apparatus localized endomannosidase-processed oligosaccharides of alpha 1-antitrypsin irrespective of their folding state (Torossi et al. 2006). From the literature, it is well known that disease-causing misfolded glycoproteins to a certain extent might escape the protein quality control and become secreted (Cox 2001; Desnick et al. 2001). As a case in point, in humans suffering from alpha1-antitrypsin deficiency, the Z-variant of alpha1-antitrypsin not only becomes partially secreted, but also is active as serine protease inhibitor (Cabral et al. 2002; Teckman and Perlmutter 1996). As experimentally shown for the Z-variant of alpha1-antitrypsin (Torossi et al. 2006), endomannosidase provided a back-up mechanism for its de-glucosylation en route through the Golgi apparatus. Processing of its oligosaccharides to mature ones is apparently important for their proper trafficking and correct functioning. Organelle changes due to intracellular accumulation of misfolded proteins As a general rule, misfolded proteins become targeted by the protein quality control and following polyubiquitination will be degraded by proteasomes (Eisele et al. 2006; Hochstrasser 1996; Jarosch et al. 2002; McCracken and Brodsky 2005; Wolf and Hilt 2004; Zwickl et al. 2002). Depending on various factors such as the efficiency of the ubiquitin–proteasome system, the intracellular amounts of misfolded glycoproteins and their biophysical properties as well as interactions with other proteins, a whole spectrum of organelle changes can be observed in protein folding diseases. For some protein folding diseases, no significant structural aberrations of the early secretory pathways could be observed. An example is Fabry’s disease, an inherited deficiency of lysosomal alpha-galactosidase A (alpha-Gal A), which causes progressive lysosomal glycosphingolipid accumulation (mainly globotriosylceramide Gb3) (Desnick et al. 2001). Disease-causing mutant alpha-Gal A could be shown by immunofluorescence to be retained in the ER where it existed in complexes with the chaperone BiP (Yam et al. 2005, 2006). From this, we concluded that recognition and ER-retention of the mutant alpha-Gal A by the protein quality control machinery constituted the mechanism leading to lysosomal deficiency in alpha-Gal A. Electron microscopic analysis of cultured fibroblast from Fabry patients harboring different mutations did not reveal significant changes of the morphology of the ER and the pre-Golgi intermediates. As expected, the fibroblasts contained numerous large lysosomes with characteristic multilamellar inclusions. Thus, the intracellularly retained mutant alpha-Gal A apparently became dislocated and was efficiently degraded by the ubiquitin–proteasome system. A similar situation was observed for a polytope membrane protein, aquaporin-2, whose folding mutants can cause renal diabetes insipidus (Canfield et al. 1997; Morello and Bichet 2001). The T126M mutant aquaporin-2 was found to be retained in the ER and efficiently degraded by proteasomes without causing ER dilatation (Hirano et al. 2003). ER retention and rapid proteasomal degradation are also hallmarks of the pulmonary form of alpha 1-antitrypsin deficiency (Lomas and Parfrey 2004; Sifers et al. 1988). However, other types of protein folding diseases have been shown to result in the distention of the ER cisternae. One example is the congenital hypothyroid goiter in which the mutant thyroglobulin is misfolded (Kim et al. 1996; Kim and Arvan 1998; Kim et al. 1998, 2000; Medeiros Neto et al. 1996). In disorders of procollagen biosynthesis, distended ER cisternae were also observed (Bogaert et al. 1992). Other examples are represented by LDL receptor class 2 mutants (Lehrman et al. 1987; Pathak et al. 1988). There are protein folding diseases associated with both distended ER cisternae and enlarged pre-Golgi intermediates. A misssense mutation of the insulin 2 gene (Cys96Tyr) in Akita mice disrupting one of the two interchain disulfide bonds is associated with intracellular accumulation of misfolded proinsulin (Wang et al. 1999). This resulted in a significant increase of the volume density of dilated ER profiles and of the pre-Golgi intermediates (Fig. 5a, b) (Fan et al. 2007; Zuber et al. 2004). For the latter, a significant increase of the tubular elements was observed. Although the mutant proinsulin was degraded through proteasomes (Wang et al. 1999), its accumulation in the early secretory pathway caused an activation of the unfolded protein response and induced apoptosis (Oyadomari et al. 2002a; Oyadomari et al. 2002b). Other mutant proteins such as the cystic fibrosis (Kopito 1999; Riordan 1999) causing delta F508 variant of the chloride channel (Gilbert et al. 1998) and misfolded major histocompatibility complex class I protein (Hsu et al. 1991; Raposo et al. 1995), have been shown to accumulate in the expanded pre-Golgi intermediates. Fig. 5Details of an insulin-producing pancreatic beta cell from Akita mice. The presence of misfolded proinsulin results in the local distention of rough ER cisternae (RER*). Arrows point to regions of transition of non-distended ER (RER) to distended ER (RER*). In addition, the pre-Golgi intermediates (pGI) are greatly enlarged. G: Golgi apparatus, TE: transitional ER. In (B), the organelle changes in terms of differences of their volume density (Vv) and differences in proinsulin distribution pattern (LI) are schematically shown (from (Zuber et al. (2004) Certain other misfolded proteins are accompanied by the formation of insoluble aggregates in the lumen of the ER, which physically precludes dislocation to the cytosol and exposure to proteasomes. The stress-induced so-called intracisternal granules in the pancreas of starved guinea pigs (Palade 1956), which are composed of aggregated proenzymes (Fig. 6a) (Geuze and Slot 1980; Pavelka and Roth 2005), form a classical example. For the liver-disease-causing alpha 1-antitrypsin Z variant, about 15% of the non-secreted mutant protein is polymerogenic and thus forms insoluble aggregates in the ER lumen, which cannot be degraded (Lomas et al. 1992, 2004) The Glu342Lys substitution of the Z-variant results in a spontaneous loop-sheet polymerization of the protein. In contrast to the above-mentioned protein folding diseases, which all have in common a loss-of-function pathogenesis, the Z-variant-caused alpha 1-antitrypsin deficiency seems to involve a pathologic gain-of-function pathogenesis (Hidvegi et al. 2005). The ER inclusions described above belong to the category of Russell bodies. Russell bodies represent subregions of the rough ER in which insoluble proteins accumulate (Fig. 6b) (Alanen et al. 1985; Kopito and Sitia 2000; Valetti et al. 1991). They are typically found in cells synthesizing mutant immunoglobulins (Alanen et al. 1985; Kopito and Sitia 2000; Mattioli et al. 2006; Valetti et al. 1991) but also in cells synthesizing, for instance, mutant myocilin. Mutations of the myocilin gene are associated with primary open-angle glaucoma (Tamm 2002). Mutant myocilins are secretion-incompetent and have been shown biochemically to form intracellular detergent-insoluble complexes (Gobeil et al. 2004; Jacobson et al. 2001; Sohn et al. 2002). In cultured cells transfected to express both mutant and wild-type myocilin, heteromeric, detergent–insoluble protein complexes were formed which were segregated into typical Russell bodies (Yam et al. 2007c). Thus, myocilin-caused open-angle glaucoma represents a protein folding disease. Its pathogenesis involves a pathological gain-of-function mechanism because of the interaction and complex formation of mutant with wild-type myocilin (Gobeil et al. 2004; Joe et al. 2003; Sohn et al. 2002; Yam et al. 2007c). As a consequence, unfolded protein response factors and pro-apoptotic factors were up-regulated and cells underwent apoptosis (Yam et al. 2007c) as detected by the appearance of lobulated nuclei and the TUNEL assay (Taatjes et al. 2008). Fig. 6a Intracisternal granules (asterisks) in the rough ER of exocrine rat pancreatic cells induced by puromycin treatment. These granules correspond to mini Russell bodies and are composed of aggregated proenzymes. b Russell bodies (RB) induced by heat shock in CHO cells. They represent distended parts of rough ER cisternae filled with protein aggregates. Note the structurally normal appearing rough ER cisternae in their neighborhood All the mutant proteins discussed above are luminal or membrane proteins. What happens to aberrant cytosolic and nuclear proteins? Same like ER proteins, one extreme situation is that they become efficiently degraded by cytosolic and nuclear proteasomes subsequent to polyubiquitination (Schubert et al. 2000; Turner and Varshavsky 2000). Thus, cytosolic and nuclear quality control in normal cells suppresses the formation of aggregates of aberrant proteins by degrading them. The other extreme is represented by the formation of cytosolic and nuclear inclusion bodies due to inefficient degradation of aberrant proteins by the ubiquitin–proteasome system. Cytosolic, non-membrane bounded inclusion bodies are generally called aggresomes (Corboy et al. 2005; Kopito and Sitia 2000). They consist of pericentriolar protein aggregates surrounded by a cage of intermediate (vimentin) filaments that are the most consistent component of aggresomes in addition to ubiquitin, proteasomes and molecular chaperones. Aggresomes can be induced experimentally by forced overexpression of aggregation-prone mutant proteins or by experimentally inhibiting proteasomes (Fig. 7) (Anton et al. 1999; Fan et al. 2007; Johnston et al. 1998; Wigley et al. 1999). On the other hand, it has been shown that protein aggregates can directly impair the function of the ubiquitin–proteasome system (Bence et al. 2001). The formation of aggresomes is a multi-step process, which depends on the intact microtubules. Aggresomes are formed by the coalescence of small protein aggregates transported from the cell’s periphery along microtubules to centrioles (Garcia-Mata et al. 1999; Johnston et al. 1998; Kawaguchi et al. 2003; Vidair et al. 1966; Wigley et al. 1999; Wojcik et al. 1996). In the nucleus, the inclusion bodies can be found in association with the promyelocytic leukemia oncogenic domains (Anton et al. 1999). Fig. 7Formation of pericentriolar aggresomes following proteasome inhibition by lactacystin in CHO cells stably expressing misfolded proinsulin. Irregularly shaped, electron dense flocculent material is present in the cytoplasm and surrounded by intermediate filaments (a). At higher magnification, the spatial relationship between the protein aggregates and a centriole can be seen (b). Micrographs from Fan et al. (2007) Inclusion bodies have been observed in association with a number of chronic neurodegenerative diseases such as Parkinson’s disease, Huntington’s disease, Alzheimer’s disease and amyotrophic lateral sclerosis (Johnston et al. 2000; Kabashi and Durham 2006; Rubinsztein 2006; Selkoe 2003; Shults 2006; Soto 2003). Inclusion bodies named Lewy bodies are a morphological hallmark of Parkinson’s disease and other neurodegenerative disorders (McNaught et al. 2002b; Olanow et al. 2004; Shults 2006). Lewy bodies in the dopaminergic neurons resemble aggresomes and represent spherical bodies commonly composed of a core of granular material and peripheral radiating filaments. They contain a variety of proteins such as alpha-synuclein, the alpha-synuclein-binding protein synphilin-1 torsin A, neurofilaments, ubiquitin, proteasomal subunits and various heat shock proteins as well as ubiquitin-activating enzyme, ubiquitin-conjugating enzyme, ubiquitin ligase enzymes and proteasome activators. Furthermore, they contain centrosome-related gamma-tubulin and pericentrin. Thus, it has been proposed that the formation of Lewy bodies represents an aggresome-like response in dopaminergic neurons (McNaught et al. 2002c). Considering the observed impairment of the ubiquitin–proteasome system in patients with Parkinson’s disease (McNaught et al., 2001, 2002a, 2006), it is assumed that Lewy bodies, by segregating increasing levels of aberrant and potentially cytotoxic proteins, might protect the neurons (Olanow et al. 2004). It should be stressed, however, that the mechanism leading to selective neuronal death in Parkinson’s disease is not fully understood and the role Lewy bodies might be playing needs to be studied further studies (Lindholm et al. 2006). Synthetic chaperones for treatment of protein folding disease The various protein folding diseases mentioned above can be classified based on the pathogenetic mechanism. Efficient proteasomal degradation of the misfolded protein is characteristic of the loss-of-function pathogenesis. This is the case in protein folding diseases such as cystic fibrosis, the lung form of alpha 1-antitrypsin deficiency, aquaporin 2-caused renal diabetes insipidus, Gaucher’s disease and Fabry’s disease. Here, the missing function of the degraded protein alone can be the cause of the clinical symptoms, or secondary effects due to substrate accumulation like in lysosomal storage diseases. Intracellular accumulation due to inefficient proteasomal degradation of misfolded proteins is representative of a pathological gain-of-function mechanism, which is combined with a loss of function. Intracellular accumulation of misfolded proteins associated or not with protein aggregation can result in the activation of the unfolded protein response leading to ER stress and apoptosis. A pathological gain-of-function mechanism can be also the cause of a dominant clinical course when the wild-type protein in complexes with the mutant protein is retained inside the cells. Examples for pathological gain-of-function pathogenesis-associated protein folding diseases are myocilin-caused open-angle glaucoma, familial hypophyseal diabetes insipidus, Parkinson’s disease and Huntington’s disease. Many attempts have been made to at least partially correct the protein misfolding in order to overcome their trafficking defect and to alleviate ER stress. Among other approaches, small molecule synthetic chaperones have been used in order to shift the folding equilibrium of mutant proteins towards a more native state (Arakawa et al. 2006; Chaudhuri and Paul 2006; Cohen and Kelly 2003; Papp and Csermely 2006; Perlmutter 2002). Chemical chaperones include osmotically active substances such as DMSO, glycerol, polyols or deuterated water, and other compounds such as 4-phenylbutyric acid (Burrows et al. 2000; Lim et al. 2004; Liu et al. 2004; Pedemonte et al. 2005; Rubenstein and Zeitlin 2000; Tamarappoo and Verkman 1998; Tveten et al. 2007; Welch and Brown 1996). Other substances such as enzyme inhibitors (Fan et al. 1999; Matsuda et al. 2003; Sawkar et al. 2002) and receptor ligands or antagonists {Petäjä-Repo, 2002 #16211;Egan, 2002 #12283} have been shown to function as pharmacological chaperones. Here, we have chosen two examples from our work to demonstrate how immunocytochemistry and microscopy in combination with biochemical analyses can be applied to demonstrate the functionality of a chemical and a pharmacological chaperone in rescuing the consequence of disease-causing protein misfolding. It has been mentioned above that open-angle glaucoma-causing mutant myocilin forms insoluble protein aggregates in the ER lumen (Russel bodies), which result in ER stress and apoptotic cell death (Yam et al. 2007b). Among the other tested chemical chaperones, treatment with sodium 4-phenylbutyrate significantly reduced the amount of intracellular detergent–insoluble myocilin aggregates and thereby the number of Russel bodies in the cells (Fig. 8a–c), diminished mutant myocilin interaction with calreticulin and restored the secretion of mutant myocilin. As a consequence, the ER stress was released and most interesting, the apoptosis rate was reduced close to levels observed in control cells expressing wild-type myocilin (Fig. 8d). Thus, sodium 4-phenylbutyrate exerts a beneficial effect by protecting the cells from the deleterious effects of mutant myocilin. Since sodium 4-phenylbutyrate is a tissue and cell-permeable molecule, it holds the potential for topical administration in the treatment of myocilin-caused primary open-angle glaucoma. Fig. 8Confocal double fluorescence of HEK 293 cells stably expressing GFP-wt and FLAG-wt myocilin reveals an ER and fine punctate pattern (a). HEK 293 cells coexpressing GFP-mutant myocilin and FLAG-wt myocilin exhibit distinct cytoplasmic aggregates, which correspond to Russell bodies (b). Treatment with the chemical chaperone sodium 4-phenylbutyrate results in reduction of the percentage of cells with myocilin-containing Russell bodies (c) and a drastic reduction of the apoptosis rate (d). Empty columns in d show values for untreated cells and filled columns for sodium 4-phenylbutyrate treated cells. From Yam et al. (2007a) The second example concerns Fabry’s disease, a lysosomal storage disorder caused by a deficiency of alpha-Gal A activity in lysosomes that results in the accumulation of glycosphingolipid globotriosylceramide (Gb3). The lysosomal trafficking of mutant alpha Gal A is impaired because the enzyme is retained in the ER by the protein quality control (Yam et al. 2005). Others had demonstrated that the activity of mutant alpha-Gal A in vitro at neutral pH could be stabilized with the competitive enzyme inhibitor 1-deoxygalactonorijimycin (DGJ) (Fan et al. 1999). Treatment of cells expressing mutant alpha-Gal A with a non-inhibitory dose of DGJ enhanced the intracellular enzyme activity (Yam et al. 2005, 2006). In addition, we could demonstrate by immunofluorescence and quantitative immunogold labeling that the mutant enzyme was redistributed from the ER to lysosomes and that this trafficking was mannose 6-phosphate-dependent. The DGJ treatment resulted in release of mutant alpha-Gal A from the chaperone BiP and in its conversion in the mature lysosomal form. Double confocal immunofluorescence and immunogold labeling demonstrated that the lysosomal Gb3 storage was cleared and that the size of the lysosomes became normalized (Yam et al. 2005, 2006). Together, this demonstrated that DGJ exhibited a chaperone-like effect and induced the trafficking of ER-retained mutant alpha Gal A to lysosomes where the enzyme was catalytically active. Therefore, the pharmacological chaperone DGJ potentially offers a convenient and cost-efficient therapeutic alternative to enzyme replacement therapy.

Pflugers_Arch-3-1-1950587

Transmural variations in gene expression of stretch-modulated proteins in the rat left ventricle

The properties of left ventricular cardiac myocytes vary transmurally. This may be related to the gradients of stress and strain experienced in vivo across the ventricular wall. We tested the hypothesis that within the rat left ventricle there are transmural differences in the expression of genes for proteins that are involved in mechanosensitive pathways and in associated physiological responses. Real time reverse transcription polymerase chain reaction was used to measure messenger RNA (mRNA) levels of selected targets in sub-epicardial (EPI) and sub-endocardial (ENDO) myocardium. Carbon fibres were attached to single myocytes to stretch them and to record contractility. We observed that the slow positive inotropic response to stretch was not different between EPI and ENDO myocytes and consistent with this, that the mRNA expression of two proteins implicated in the slow response, non-specific cationic mechanosensitive channels (TRPC-1) and Na/H exchanger, were not different. However, mRNA levels of other targets, e.g. the mechanosensitive K+ channel TREK-1, Brain Natriuretic Peptide and Endothelin-1 receptor B, were significantly greater in ENDO than EPI. No targets had significantly greater mRNA levels in EPI than ENDO. On the basis of these findings, we suggest that the response of the ventricle to stretch will depend upon both the regional differences in stimuli and the relative expression of the mechanosensitive targets and that generally, stretch sensitivity is predicted to be greater in ENDO. Introduction It is known that mechanical stimulation of the myocardium (stress and strain) can lead to both acute and chronic changes in the electrical and mechanical activity of the heart and contribute to the development of cardiac hypertrophy [7, 11]. It is acknowledged that the ventricular myocardium is not homogenous and that there are regional differences in the basic properties of ventricular myocytes, e.g. [1, 10]. In the whole heart, transmural gradients in ventricular wall stress and strain exist [6], which may influence the expression of genes for proteins that are sensitive to mechanical stimuli. This may contribute to regional variations in physiological characteristics and thereby to transmural differences in the response to stimuli such as stretch. Therefore, regional variations in wall stress and strain, in conjunction with regional differences in gene expression, may regulate the response of the heart. To date, relatively little is known about these possibilities, although it has been shown that the messenger RNA (mRNA) expression and current density of the mechanosensitive twin-pore K+ channel TREK-1 is greater in the rat sub-endocardium (ENDO) than the sub-epicardium (EPI) [12, 8, 6]. The purpose of this study was to test the hypothesis that the mRNA expression levels of several proteins, which are thought to modulate the mechanical and electrical response to stretch, vary across the rat left ventricular wall. Additionally, we wished to test whether the slow contractile response to axial stretch [2], which is thought to be modulated by some of the chosen targets, differs between myocytes from different regions of the left ventricular free wall, thus potentially linking gene expression to function. Materials and methods All animal experimentation was carried out in accordance with the Animals (Scientific Procedures) Act 1986 and the local Institutional ethical review panel. Isolation of RNA Ten Sprague–Dawley rats (194–240 g) were killed and the hearts removed. The left ventricular free wall was excised and snap frozen in liquid nitrogen between the large flat ends of a pair of tongs. The thickness of the left ventricular wall was measured with pre-cooled callipers, and 25-μm sections, equivalent to one third of the wall thickness, were collected on a cryostat from the EPI and ENDO surfaces, leaving a distinct mid-myocardial region. All samples were stored at −80°C until RNA extraction. Total RNA extraction was performed following the Qiagen mini-kit protocol for striated muscle, with the exception that an additional volume of RLT was added to the Proteinase K solution before binding of the RNA onto the mini-columns. Integrity and concentration of the isolated RNA were assessed by electrophoresis through 0.6% formaldehyde/1% agarose gels with ethidium bromide and the concentration of RNA adjusted to 1 g/l. Complementary DNA (cDNA) was prepared from 4 μg of total RNA with random priming using Superscript III first-strand synthesis system (Invitrogen, Life Technologies, Rockville, MD) and diluted 1:10 in TE (10 mM Tris-HCl [pH 7.5], 1 mM ethylenediamine tetraacetic acid) before use in real-time polymerase chain reaction (PCR). Real-time RT-PCR Real-time reverse transcription PCR (RT-PCR) was performed using TaqMan low-density arrays (Micro Fluidic Cards, Applied Biosystems, Foster City, CA). Each card consisted of 384 wells, preloaded with pre-designed fluorogenic TaqMan probes and primers, configured to allow detection of 48 transcripts for 7 experimental and 1 calibrator samples. Our calibrator sample consisted of a mixture of cDNA from EPI and ENDO regions. Each of the eight sample lanes in a card was loaded with 100 μl of a 1:1 mixture of cDNA (equivalent to 77 ng input RNA) and TaqMan Universial PCR master mix (Applied Biosystems). PCR was done according to the recommended protocol (50°C for 2 min and 94.5°C for 10 min, followed by 40 cycles at 97°C for 30 s and 59.7°C for 1 min) on an ABI RISM 7900HT Sequence detection system (Applied Biosystems). Data were collected with instrument spectral compensations by the Applied Biosystems SDS 2.2 software and analysed using the threshold cycle relative quantification method. Relative transcript expression was normalised to the housekeeper gene GAPDH. Mechanosensitive targets mRNA targets chosen for this investigation were: TREK-1 (a two-pore-domain K+-selective channel) and TRPC-1 (transient receptor potential canonical, a cationic non-selective mechanosensitive channel [MSC]); the Na/H exchanger, thought to be central to the slow increase in force that occurs upon axial stretch; endothelin-1 and its receptor types A and B and the angiotensin II receptor A, known to regulate the hypertrophic response to mechanical stimulation and possibly the slow inotropic response; caveolin-3 a critical component of caveolae, which are potential mechanosensors; caveolin-1, a major component of caveolae in non-cardiac muscle and recently identified in cardiac myocytes; brain natruiretic peptide (BNP) and the α and β sub-types of myosin heavy chain (MHC), the expression of both BNP and β-MHC are known to be increased by stretch [11]. Isolation of left ventricular myocytes The isolation of single rat left ventricular myocytes and the measurement of the rapid and slow responses to axial stretch by attached carbon fibres was performed as previously described [2]. In brief, excised hearts were perfused on a Langendorff apparatus with a collagenase/protease containing physiological solution. EPI and ENDO strips, each one third of the wall thickness, were manually cut from the left ventricular free wall with scissors leaving a distinct and intact mid-myocardial region. Further incubation with enzyme solution resulted in the isolation of separate populations of EPI and ENDO myocytes. Myocytes were attached to carbon fibres to record auxotonic force development in response to external stimulation and to axially stretch myocytes. Force was calculated from fibre motion, and stretch was calculated from the increase in sarcomere length. The rapid response to stretch was measured within 20 s of an increase in sacromere length, whereas the slow response was calculated as the further increase in force 5 min after the stretch. These experiments were carried out at a stimulation frequency of 1 Hz at a temperature of 22–24°C (to improve the experimental success rate) using a bicarbonate-based buffer containing (in mM): NaCl 118.5; NaHCO3 14.5; KCl 4.2; KH2PO4 1.2; MgSO4 1.2; glucose 11.1; CaCl2 1, equilibrated with 95% O2/5% CO2 (pH 7.4). Statistics Data were analysed using paired t tests to compare relative mRNA expression from the EPI vs ENDO region of each heart and unpaired t tests to compare the magnitude of the rapid and slow inotropic responses. Statistical significance was assumed when P < 0.05. Results There was some consistency in the pattern of regional expression of mechanosensitive genes (Table 1). Although the expression levels of some targets were unchanged, others were significantly greater in ENDO but never so in EPI. This observation was not due to a systematic error of the technique as levels of mRNA of targets known to be greater in EPI, e.g. the potassium channel component, Kv4.2, were significantly greater in EPI than ENDO (results not presented). Indeed, if α-MHC is excluded (being chosen as a counterpoint to β-MHC), 9 of the 11 targets had a larger mean mRNA expression in ENDO than EPI; this trend is statistically significant (P < 0.05, sum of ranks test). Table 1Comparison of mRNA expression of mechanosensitive proteins from the EPI and ENDO region of rat left ventricleAssay ID referenceCommon nameEPIENDOPaired t testMean ± SEMMean ± SEMSlow response and/or hypertrophic Agtr1a-Rn00578456_m1Angiotensin II receptor 1A1.64 ± 0.161.30 ± 0.22NS Edn1-Rn00561129_m1Endothelin-11.17 ± 0.141.58 ± 0.24NS Ednra-Rn00561137_m1Endothelin-1 receptor type A1.19 ± 0.141.38 ± 0.12NS Ednrb-Rn00569139_m1Endothelin-1 receptor type B1.27 ± 0.171.93 ± 0.220.001 ENDO > EPI Nppb-Rn00580641_m1Brain natriuretic peptide1.42 ± 0.152.77 ± 0.420.002 ENDO > EPI Slc9a1-Rn00561924_m1Sodium–hydrogen exchanger1.33 ± 0.091.51 ± 0.17NSMyofilaments Myh6-Rn00568304_m1α-myosin heavy chain1.56 ± 0.121.37 ± 0.17NS Myh7-Rn00568328_m1β-myosin heavy chain1.20 ± 0.232.19 ± 0.360.006 ENDO > EPIMechanosensitive channels Kcnk2-Rn00597042_m1Kcnk2 (TREK-1)0.84 ± 0.041.04 ± 0.040.001 ENDO > EPI Trpc1-Rn00585625_m1Transient receptor potential canonical-10.72 ± 0.040.72 ± 0.05NSMechanotransducers (caveolae) Cav-Rn00755834_m1Caveolin-10.90 ± 0.030.96 ± 0.030.015 ENDO > EPI Cav3-Rn0055343_m1Caveolin-30.81 ± 0.030.85 ± 0.05NSData shown are mean ± SEM, expression is relative to GAPDH. The probability of paired t tests is shown in the table. P > 0.05 (n = 10 pairs).NS Not significantly different When single myocytes were stretched from their resting sarcomere length (ENDO 1.83 ± 0.09 μm; EPI 1.82 ± 0.07 μm) by an equivalent amount (ENDO 7.8 ± 0.9 %; EPI 7.5 ± 0.5%, n = 11 ENDO, 18 EPI myocytes) as illustrated in Fig. 1a, we observed that the magnitude of neither the rapid nor slow inotropic response was dependent upon regional origin (Fig. 1b). Fig. 1The inotropic response of left ventricular myocytes to axial stretch. a Representative trace showing the typical biphasic inotropic response of a myocyte to a stretch of 8% from a resting sacromere length (SL) of 1.85 μm. The trace shows changes in active force, resting force has been subtracted with a sample and hold device. The rapid increase in force is seen immediately upon an increase in SL, the slow response develops over the following minutes. b Mean data showing the rapid and slow response to stretch of ≅8% from a resting SL of ≅1.82 μm in sub-epicardial (EPI) and sub-endocardial (ENDO) left ventricular myocytes. The responses from EPI and ENDO cells were not significantly different from each other. P > 0.05, n = 18 EPI , n = 11 ENDO Discussion This investigation of transcripts for proteins sensitive to stretch, or involved in mechanosensitive pathways, has allowed us to detect a number of small but nevertheless significant differences in mRNA levels between ENDO and EPI samples. Rapid and slow inotropic responses to axial stretch Following equivalent amounts of axial stretch, the rapid inotropic response was not significantly different in EPI and ENDO rat myocytes. This observation is consistent with findings in untrained intact myocytes [10]. However, a greater stretch-induced increase in myofilament Ca2+ sensitivity (δpCa50) has been reported in untrained skinned ENDO myocytes than EPI myocytes when larger stretches (from sarcomere lengths of 1.9 to 2.3 μm) are applied [3]. We observed that the slow response did not vary with myocyte regional origin. This is an important and novel observation because the multicellular cardiac preparations typically used to study the slow response (trabeculae and papillary muscles) are ENDO in nature. Additionally, previous whole heart studies recorded the aggregate response from the whole left ventricle while our own previous single cell experiments did not distinguish between myocyte origin. Consistent with the functional data, we observed that the mRNA expression levels for proteins implicated in the development of the slow inotropic response in rat (non-specific cationic MSCs, possibly TRPC-1) and the Na/H exchanger, see [2], are not regionally different. It should be noted however, that although TRPC-1 has been proposed as a non-specific cationic MSC [9], such a role has not yet been established in the myocardium. MSCs and mechanotransducers Although levels of TRPC-1 mRNA were not regionally different, we found that TREK-1 mRNA expression was greater in ENDO than EPI. This latter finding agrees with the observations of Tan et al. [12] and Kelly et al. [6], although the regional difference in our study was smaller than previously reported. We would therefore hypothesise that an equivalent mechanical stimulus that activated TRPC-1 and TREK-1 would produce a larger electrical response in ENDO tissue (see [6]). Whether increased TREK-1 in ENDO would stabilise ENDO membrane potential during the diastolic period and be potentially anti-arrhythmic or reduce the transmural gradient in action potential duration and be potentially pro-arrhythmic [1] remains to be resolved. Although we found no evidence for a regionally different expression of caveolin-3, an essential component of cardiac caveolae, the role of caveolae may also be influenced by regional expression of the receptors and other signalling molecules that they aggregate. The observation that mRNA for caveolin-1 is greater in ENDO is intriguing as this isoform has only recently been identified in cardiac myocytes, and its role in myocyte caveolae is unknown. Indicators of cardiac hypertrophy The ATII/ET-1 signalling pathways are known to play a role in the hypertrophic response to mechanical stimulation and, in some species, the slow inotropic response, see [4] for review. We only observed significantly greater levels of mRNA for the ET-1B receptor whereas it is the ET-1 A receptor that Cingolani et al. report to be important for the slow response. We also saw increased mRNA levels of BNP and of β-MHC in ENDO. Previous studies have variously reported increased [5] or similar [3] levels of β-MHC protein in normal rat ENDO vs EPI myocardium. Both BNP and β-MHC are indicators of hypertrophy provoked by mechanical stimuli. It is therefore interesting that wall stress and, in some studies, wall strain, is thought to be greater in ENDO [6]. Assumptions and limitations In this study, we have made the common, implicit assumption that there is a link between levels of mRNA expression and protein activity. This assumption can only be validated by measuring in situ protein activity, a huge task given the diverse nature of our targets. However, in measuring the slow inotropic response to stretch and the mRNA expression of the targets thought to underlie this effect, we have attempted to address this problem. We chose the rat for consistency with our previous work on regional differences and on the slow response to stretch [10, 2] and because of the ready availability of mRNA sequences for this species. Given that regional variations exist in other species [1], it seems reasonable to expect that our observations on mechanosensitive targets are relevant to other mammalian species, although investigation of other species is required to confirm this. Conclusions The level of mRNA expression of some proteins that are modulated by mechanical stimuli vary across the rat left ventricular wall. In vivo, the response to mechanical stimulation is likely to be complex. Targets (e.g. in the present study caveolin-3 and TRPC-1) whose mRNA expression are not regionally different might still be differentially regulated by transmural variation in the mechanical stimuli, whereas other permutations of increased/decreased mRNA expression and protein activity with increased/decreased stimuli are possible. Based upon our present observations and those from the literature reporting increased stretch sensitivity of various responses in ENDO, e.g. [5, 10, 3, 6], it seems that in general, ENDO tissue is likely to be more influenced by, and be responsive to, mechanical stimulation than EPI tissue.

Histochem_Cell_Biol-4-1-2413110

The desmosome and pemphigus

Desmosomes are patch-like intercellular adhering junctions (“maculae adherentes”), which, in concert with the related adherens junctions, provide the mechanical strength to intercellular adhesion. Therefore, it is not surprising that desmosomes are abundant in tissues subjected to significant mechanical stress such as stratified epithelia and myocardium. Desmosomal adhesion is based on the Ca2+-dependent, homo- and heterophilic transinteraction of cadherin-type adhesion molecules. Desmosomal cadherins are anchored to the intermediate filament cytoskeleton by adaptor proteins of the armadillo and plakin families. Desmosomes are dynamic structures subjected to regulation and are therefore targets of signalling pathways, which control their molecular composition and adhesive properties. Moreover, evidence is emerging that desmosomal components themselves take part in outside-in signalling under physiologic and pathologic conditions. Disturbed desmosomal adhesion contributes to the pathogenesis of a number of diseases such as pemphigus, which is caused by autoantibodies against desmosomal cadherins. Beside pemphigus, desmosome-associated diseases are caused by other mechanisms such as genetic defects or bacterial toxins. Because most of these diseases affect the skin, desmosomes are interesting not only for cell biologists who are inspired by their complex structure and molecular composition, but also for clinical physicians who are confronted with patients suffering from severe blistering skin diseases such as pemphigus. To develop disease-specific therapeutic approaches, more insights into the molecular composition and regulation of desmosomes are required. Introduction Desmosomes are intercellular adhering junctions serving to attach neighbouring cells to each other. They are most numerous in tissues subjected to significant mechanical stress such as the stratified squamous epithelia of the skin (Bizzozero 1864) and of mucous membranes (Farquhar and Palade 1963) as well as the myocardium (Fawcett and Selby 1958). Moreover, desmosomes are found in simple epithelia and in non-epithelial cells such as the meningeal cells of the arachnoidea (Gusek 1962) and the follicular dendritic cells of lymph follicles (Swartzendruber 1965). Desmosomes were discovered as cell contacts in the middle of the nineteenth century (Calkins and Setzer 2007). By the means of light microscopy, desmosomes were first described in the epidermis by the Italian pathologist Bizzozero in (1864). In his histology text book, the anatomist Josef Schaffer from Vienna introduced the term “desmosome”, by combining the greek words “desmos” (bond) and “soma” (body) although, to that time, he, like most others in the field believed that desmosomes were cytoplasm-filled intercellular bridges (Schaffer 1920). It took almost another century until Keith Porter, using electron microscopy, was able to confirm the basic observation of Bizzozero that desmosomes rather are contacts between adjacent cells and to allow the first description on desmosome ultrastructure (Porter 1956). With these new technical advances at hand, several studies were performed in the following years on the distribution and organization of desmosomes in various tissues. In addition, starting in the 1970s, biochemical approaches and molecular cloning techniques were applied to identify the desmosomal components and to characterize their interactions (Drochmans et al. 1978; Moll et al. 1986; Moll and Franke 1982; Schwarz et al. 1990; Skerrow and Matoltsy 1974). Significant insights into the regulation of desmosomal adhesion also came from the field of dermatology since it was demonstrated that autoantibodies in patients suffering from the autoimmune blistering skin diseases pemphigus vulgaris (PV), and pemphigus foliaceus (PF), are directed to Ca2+-sensitive cell surface proteins within desmosomes (Eyre and Stanley 1987, 1988; Jones et al. 1986b; Karpati et al. 1993), which were identified as the desmosomal cadherins desmoglein 1 (Dsg 1) and Dsg 3 (Amagai et al. 1991; Koulu et al. 1984). The term “pemphigus” comes from the greek word “pemphix” (blister) and is being used in dermatology since 1791 (Schmidt et al. 2000), long before it was found that pemphigus is associated with autoantibodies against keratinocyte surface antigens (Beutner and Jordon 1964) and that these antibodies are sufficient to cause acantholysis, i.e. loss of cell–cell adhesion, in human skin in vivo and in vitro (Anhalt et al. 1982; Schiltz and Michel 1976). The final break-through was the finding that autoantibodies against the extracellular domains of Dsg 3 and Dsg 1 in PV and in PF are pathogenic (Amagai et al. 1995, 1994a, 1992). Therefore, autoantibodies from pemphigus patients have been used to characterize the mechanisms involved in the regulation of desmosomal adhesion. Except from pemphigus, other diseases in which desmosomal adhesion is altered by mutations or bacterial toxins helped to elucidate the functional role of the different desmosomal components. During the last past several years, a number of comprehensive reviews have been published on both desmosome structure and function (Dusek et al. 2007b; Garrod et al. 2002; Getsios et al. 2004b; Green and Simpson 2007; Holthofer et al. 2007; Kitajima 2002; Kottke et al. 2006; Muller et al. 2008a; Yin and Green 2004) and/or on the mechanisms involved in pemphigus pathogenesis (Amagai 2003; Hashimoto 2003; Lanza et al. 2006; Payne et al. 2004; Sharma et al. 2007; Sitaru and Zillikens 2005; Stanley and Amagai 2006), which indicates that the perspective of the existing model of the desmosome and its role in pemphigus pathogenesis are constantly reshaped. Moreover, because even textbook knowledge such as on the molecular composition of myocardial intercalated discs needs revision (Borrmann et al. 2006; Franke et al. 2006), it becomes obvious that after almost 150 years of desmosome research, our knowledge is still far from complete. This article focuses on the mechanisms regulating desmosomal adhesion, which are compromised in diseases such as pemphigus. The ultrastructure and composition of desmosomes The first detailed analysis of desmosome ultrastructure was provided by Odland (1958). Desmosomes are discoid junctions with a diameter of about 0.2–0.5 μm and are composed of two electron-dense plaques in each of the two cells which are separated by an intercellular cleft of 24–30 nm (Figs. 1, 2) (Farquhar and Palade 1963; Odland 1958). Within the plaques, an outer dense plaque can be separated from a less dense inner plaque, the latter of which is linked to loops of intermediate filament bundles (Kelly 1966). Desmosomes contain members of at least three protein families. Desmosomal cadherins form the intercellular adhesive interface, whereas armadillo and plakin family proteins built up the plaques. It is believed that the cytoplasmic tail of Dsgs and Dscs interact with plakoglobin which in turn binds to desmoplakin (Fig. 1). Desmoplakin finally is anchored to the intermediate filament cytoskeleton (Green and Simpson 2007). These interactions seem to be stabilized laterally by plakophilin (Hatzfeld 2007). Fig. 1Molecular model of the desmosome. The desmosomal cadherins desmoglein and desmocollin undergo homophilic and heterophilic binding via interaction with the amino-terminal extracellular (EC) 1 domain of partner molecules on the same (cis) as well as on the neighbouring cell (trans). The cytoplasmic domains are largely embedded in the outer dense plaque (ODP) where they are associated with plakoglobin and plakophilin. In the inner dense plaque (IDP), desmoplakin links these adaptor molecules to the intermediate filament cytoskeletonFig. 2Ultrastructure of the desmosome. The electron micrograph of a keratinocyte desmosome shows the desmosomal plaque with inserting cytokeratin intermediate filaments as well as some fuzzy material within the extracellular space likely reflecting the extracellular domains of desmosomal cadherins Desmosomes and desmosome-like junctions Adhering junctions are divided into two main forms: (1) desmosomes, which serve as anchoring structures for intermediate filaments to desmosomal cadherins, and (2) adherens junctions, which contain cell-type specific adhesion molecules from the cadherin super-family that are linked to the actin cytoskeleton. Both, desmosomes and adherens junctions can be found as constituents of more elaborated cell contact complexes. Moreover, chimeric cell contacts exist which share features of both adherens junctions and desmosomes. Junctional complexes Polarized epithelial cells display junctional complexes located at the uppermost section of the baso-lateral membrane. In apico-basal direction, the complex is composed of the zonula occludens (tight junction), the zonula adherens and a desmosome (macula adherens) (Farquhar and Palade 1963). Accompanied by a line of separated desmosomes, the zonula occludens and the zonula adherens span the entire cell by forming continuous junction belts. These junctional complexes are regarded as hallmarks of polarized epithelial cells but differ in terms of size and ultrastructure in cell type-specific manner. Area composita The intercalated discs of the myocardium also consist of three types of cell junctions, i.e. adherens junctions, desmosomes and gap junctions (Fig. 3). Although “transitional forms” between adherens junctions and desmosomes were described in the very beginning (Fawcett and Selby 1958), most morphological studies regarded the intercalated discs to be composed of separated desmosomes and adherens junctions, the former linked to the desmin type intermediate filament cytoskeleton and the latter to actin filaments of the myofibrills (Fig. 3) (Fawcett and McNutt 1969; McNutt and Fawcett 1969; Shimada et al. 2004). This view seemed to be supported by immuno-localization studies, which showed that desmoplakin as a desmosomal marker and the myocardial adherens junction protein N-cadherin displayed mutually exclusive spatial distribution patterns (Angst et al. 1997; Gutstein et al. 2003). However, a recent comprehensive set of studies unequivocally demonstrated that the desmosomal components desmoglein 2 (Dsg 2), desmocollin 2 (Dsc 2), desmoplakin and plakophilin-2 as well as the adherens junction components N-cadherin, cadherin-11, α-catenin and β-catenin, afadin, vinculin and ZO-1 are present in all parts of intercalated discs irrespective of whether their ultrastructure resembles more closely typical desmosomes or adherens junctions (Borrmann et al. 2006; Franke et al. 2006). It seems that the two types of junctions coalesce within the first-year postpartum and that plakophilin 2 is of special importance for junction integrity (Pieperhoff and Franke 2007; Pieperhoff et al. 2008). Therefore, the intercalated discs were now reclassified as “area composita”, a mixed type of adhering junctions. Fig. 3Ultrastructure of the area composita of a myocardial intercalated disc. The electron micrograph shows an intercalated disc containing a gap junction (GJ) in its longitudinal section as well as an adhering junction with an extensive electron-dense plaque in the section perpendicular to the cellular axis. Note that insertion of actin filaments, which is typical for adherens junctions, is present in some parts of the junction (asterisk) but not in others (hash key). Based on the recent finding that all parts of these adhering junctions contain the same set of desmosomal components, they are now defined as area composita Complexus adhaerentes and meningeal junctions Lymphatic endothelial cells in certain lymphatic vessels and in the sinus of lymph nodes form a kind of chimeric cell contact, which contains components from desmosomes (desmoplakin, plakoglobin), adherens junctions (VE-cadherin, α-catenin, β-catenin, afadin) as well as from tight junctions (claudin-5 and ZO-1) (Hammerling et al. 2006; Schmelz and Franke 1993; Schmelz et al. 1994). The ultrastructure of complexus adhaerentes shares features of both adherens junctions and desmosomes. Complexus adhearentes form continuous belt-like junctions similar to adherens junctions of vascular endothelial cells, whereas their junctional plaques are more similar to desmosomes. It was proposed that plakoglobin in these contacts is responsible for recruitment of desmoplakin (Kowalczyk et al. 1998). Recently, in meningeoma cells, a new type of adhering junctions was discovered in which adherens junctions also contained the desmosomal plaque protein plakophilin 2 (Akat et al. 2008). The desmosomal components Desmosomal cadherins The desmosomal members of the cadherin superfamily, desmogleins (Dsg 1–4) and desmocollins (Dsc 1–3), are single-pass transmembrane glycoproteins, which mediate adhesion in Ca2+-dependent manner (Buxton et al. 1993; Garrod et al. 2002; Getsios et al. 2004b; Nollet et al. 2000). Desmosomal adhesion molecules have been first isolated from desmosomal intercellular regions (Gorbsky and Steinberg 1981). Using specific antibodies to localize proteins at the cell surface and to inhibit desmosome formation, Dsc 1 (130 kDa) and Dsc 2 (115 kDa) were shown to be directly involved in cell–cell adhesion (Cowin et al. 1984) and later on were identified to be cadherin family proteins (Collins et al. 1991; Koch et al. 1991b). Similarly, Dsg 1 (165 kDa), Dsg 2 (116 kDa), Dsg 3 (130 kDa) and Dsc 3 (110 kDa) were characterized (Amagai et al. 1991; Arnemann et al. 1991; Jones et al. 1986a; King et al. 1995; Koch et al. 1991a, 1990; Schafer et al. 1994; Schmelz et al. 1986a, b). More recently, Dsg 4 (108 kDa) was found to be the principal Dsg expressed in hair follicles (Kljuic et al. 2003; Whittock and Bower 2003). The genes encoding desmosomal cadherins, which share an amino acid identity of approximately 30–50%, both with each other and with classical cadherins, are all located on chromosome 18 in humans (Cowley et al. 1997). Mouse models revealed that Dsg 2 and Dsc 3 are the most important desmosomal cadherin members because deficiency caused embryonic lethality (Table 1) (Den et al. 2006; Eshkind et al. 2002). Because lethality induced by Dsc 3 deficiency occurred before mature desmosmes were formed, and because Dsg 2 was observed to be also localized outside of desmosomes in embryonal stem cells, non-desmosmal functions of Dsc 3 and Dsg 2 seemed to be responsible in this context. In contrast, ablation of cadherins with more restrictive expression patterns such as Dsc 1 led to localized superficial epidermal acantholysis or mucosal and deep epidermal splitting in traumatized skin accompanied by hair loss in the case of Dsg 3 (Chidgey et al. 2001; Koch et al. 1998). Dsg 4 mutations are followed primarily by defective hair formation (Kljuic et al. 2003). Table 1Desmosome-associated diseases in humans and in transgenic mouse modelsMouse model Desmosomal componentHuman diseaseInactivationGenetic alteration–Dsg 1Striate palmoplantar keratoderma (SPPK) (Rickman et al. 1999)Embryonic lethalityDefective stem cell proliferation(Eshkind et al. 2002)Dsg 2Arrhythmogenic right ventricular cardiomyopathy (ARVC) (Pilichou et al. 2006)Suprabasal epidermal acantholysisOral erosionsHair loss(Koch et al. 1998)Dsg 3–Impaired hair keratinisationHyperproliferation (Kljuic et al. 2003)Dsg 4Hypotrichosis (Kljuic et al. 2003)Epidermal granular layer acantholysisImpaired barrier functionHyperproliferation(Chidgey et al. 2001)Dsc 1––Dsc 2Arrhythmogenic right ventricular cardiomyopathy (ARVC)(Syrris et al. 2006; Heuser et al. 2006)Embryonic lethality(Den et al. 2006)Dsc 3–Embryonic lethalityMyocardial fragilityEpidermal subcorneal acantholysis(Bierkamp et al. 1996, 1999; Ruiz et al. 1996)PlakoglobinNaxos disease:Arrhythmogenic right ventricular cardiomyopathy (ARVC)Palmoplantar keratodermaWoolly hair(McKoy et al. 2000)Embryonic lethalityDefects in heart, skin, blood vesselsNeuroepithelium(Gallicano et al. 1998, 2001)DesmoplakinArrhythmogenic right ventricular cardiomyopathy (ARVC)(Rampazzo et al. 2002)Striate palmoplantar keratoderma (SPPK)(Armstrong et al. 1999; Whittock et al. 1999)Carvajal syndrome:Dilated left ventricular cardiomyopathyStriate palmoplantar keratoderma (SPPK)Woolly hair(Norgett et al. 2000)Lethal acantholytic epidermolysis bullosa:-suprabasal epidermal blistering-universal alopecia-nail loss(Jonkman et al. 2005)–Plakophilin 1Ectodermal dysplasia and skin fragility syndrome:Skin blistering around mouth, on soles and palms(McGrath 1997)Embryonic lethalityHeart defects(Grossmann et al. 2004)Plakophilin 2Arrhythmogenic right ventricular cardiomyopathy (ARVC)(Gerull et al. 2004)Bacterial toxinsSuperficial epidermal acantholysis(Melish and Glasgow 1970; Amagai et al. 2000a, b)Dsg 1Staphylococcus scalded skin syndrome:Superficial skin splittingFever, erythemaSkin tenderness(first described by Ritter von Rittershain 1878)AutoantibodiesSuperficial epidermal acantholysis(Amagai et al. 1995)Dsg 1Pemphigus foliaceus (PF)Superficial epidermal acantholysisPemphigus vulgaris (PV, mucocutaneous)Suprabasal epidermal acantholysisMucosal erosions (mouth, larynx, nasal cavity, vagina)(Lever 1953; Koulu et al. 1984; Amagai et al. 1991)Suprabasal epidermal acantholysisMucosal erosions(Anhalt et al. 1982; Amagai et al. 1992; Mahoney et al. 1999)Dsg 3Pemphigus vulgaris (PV, mucosal dominant)Suprabasal epidermal acantholysis(Lever 1953; Amagai et al. 1991) The structure of desmosomal cadherins Desmosomal cadherins are type I integral membrane proteins. The amino-terminal extracellular domain of desmosomal cadherins consists of four cadherin repeats (EC1–4) of about 110 amino acids followed by a less related membrane-proximal domain (EC 5) (Dusek et al. 2007b). Based on the crystal structure of classical cadherins (Boggon et al. 2002; Overduin et al. 1995; Shapiro et al. 1995), the EC 1–4 domains are thought to be connected via flexible linkers which are rigidified by binding of up to three Ca2+ ions each (Pertz et al. 1999). In the cytoplasmic domain, a juxtamembranuous anchor (IA) region is located which, at least in the case of Dsc 1a, contains a desmoplakin-binding element (Troyanovsky et al. 1994b) and maybe be involved in binding and trafficking of p120catenin similar to its role in E-cadherin (Miranda et al. 2003). The following cadherin-typical sequence (ICS) is required for binding to plakoglobin (Mathur et al. 1994; Troyanovsky et al. 1994a). For desmocollins, in which a long and short “a” and “b” isoform is generated by alternative splicing (Collins et al. 1991), it has been shown that the ICS domain is lacking in the b isoform which therefore is unable to bind plakoglobin but instead associates with plakophilin 3 (Bonne et al. 2003; Troyanovsky et al. 1993). The functions of the C-terminal proline-rich linker (L), the Dsg-specific repeated unit domains (RUDs) and the desmoglein terminal domain (DTD), which are all only present in desmogleins, are not clear (Dusek et al. 2007b). The Ca2+-dependency of desmosomal cadherin-mediated binding It is well established that binding of desmosomal and classical cadherins is strictly Ca2+-dependent (Chitaev and Troyanovsky 1997; Heupel et al. 2007; Pertz et al. 1999; Waschke et al. 2007). For Dsg 1, Ca2+-dependency of homophilic binding has been characterized in more detail. It was found that the Ca2+ concentration for half-maximal binding activity of Dsg 1 is 0.8 mM Ca2+ and that binding is highly cooperative with the Hill coefficient being ≥5 (Waschke et al. 2007). This indicates that Dsg 1 binding is strong only at extracellular Ca2+concentrations higher than 0.8 mM. Although the exact extracellular Ca2+concentration within the epidermis is unknown, it has been shown that a gradient exists with low Ca2+concentrations in the basal layers and high concentrations in the superficial epidermis (Elias et al. 2002; Menon and Elias 1991). Therefore, if homophilic binding of Dsg 1 occurs in vivo, it has to be considered that it may contribute to effective intercellular adhesion only in the superficial epidermis. Transinteraction mechanisms of desmosmal cadherins At present, it is still a matter of debate how desmosmal cadherins interact with each other in vivo. However, several lines of evidence indicate that the N-terminal EC 1 domain is important. Similar to classical cadherins, desmosomal cadherins contain a cell adhesion recognition (CAR) site containing a central alanine residue (Blaschuk et al. 1990). However, instead of the conserved tripeptide HAV sequence of classical cadherins, the sequence is YAT for Dsc 1 and RAL for Dsg 1, respectively (Tselepis et al. 1998). Peptides derived from these sequences were able to block homophilic adhesion mediated by Dsg 1 and Dsc 1 and to inhibit desmosomal adhesion in epithelial cells when the peptides for Dsg 1 and Dsc 1 were applied together, indicating that the CAR site in the EC 1 domain is critical for maintenance of desmosmal adhesion (Runswick et al. 2001; Tselepis et al. 1998). This hypothesis is supported by studies in which mechanisms underlying the loss of keratinocyte cohesion in pemphigus were investigated. AK23, a monoclonal Dsg 3 antibody from a PV mouse model directed against the predicted binding motif of the EC 1, has been shown to be pathogenic in vivo whereas antibodies targeting other parts of the Dsg 3 extracellular domain were not (Tsunoda et al. 2003). Together with the recent finding that AK 23 similar to Dsg 3 antibodies from PV patients, which are known to be also primarily directed against the N-terminal part of EC 1 (Sekiguchi et al. 2001), is able to directly interfere with Dsg 3 binding (Heupel et al. 2007), these data demonstrate that interaction of EC 1 is crucial for Dsg 3 binding. In addition to these functional studies, morphologic studies sought to address the mode of desmosomal cadherin interaction within desmosomes by using electron tomography imaging of epidermal tissue (Al-Amoudi et al. 2007; He et al. 2003). Based on predictions from the C-cadherin crystal structure, He and colleagues reported that in mouse epidermis several desmosomal cadherins form knots in which cadherins display stochastic arrangement. In these knots, desmosomal cadherins seemed to interact via their EC 1 domains with both molecules on the same cell in cis as well as with molecules from opposing cells in trans (He et al. 2003). Al-Amoudi and co-workers refined the technique by employing cryo-electron microscopy in human epidermis. They confirmed cis- and trans-interactions of the EC 1 domains, possibly via insertion of the tryptophane 2 into the hydrophobic pocket of the CAR site (Al-Amoudi et al. 2007). However, they found that cis-interactions of the EC 4–5 regions may also occur and that desmosomal cadherins rather show periodically zipper-like arrangements similar to classical cadherins (Boggon et al. 2002). Homophilic and heterophilic binding of desmosomal cadherins In contrast to classical cadherins from adherens junctions which primarily bind in homophilic manner, data indicate that desmosomal cadherins undergo both homophilic and heterophilic transinteraction. Using EC 1–2 fragments of Dsg 2 and Dsc 2, it was shown that homophilic binding occurs in vitro (Syed et al. 2002). Similarly, homophilic binding of Dsg 3 was found (Amagai et al. 1994b). When recombinant proteins consisting of the complete extracellular domain were used for atomic force microscopy (AFM) measurements, it was demonstrated that the unbinding force of single homophilically transinteracting molecules was about 37–68 pN (depending on the retrace velocity 300–6,000 nm/s) for Dsg 1 with a lifetime of 0.17 s and about 50 pN for Dsg 3 (Heupel et al. 2007; Waschke et al. 2005, 2007), which is in the same range as observed for other types of cadherins characterized by the same method such as VE-cadherin, N-cadherin or LI-cadherin (Baumgartner et al. 2003, 2000; Wendeler et al. 2007). These data indicate that the molecular binding properties of homophilic adhesion of desmosomal cadherins may be comparable to other cadherins. Heterophilic binding of Dsg 2 to Dsc 1 or Dsc 2 was also found on the molecular level (Chitaev and Troyanovsky 1997; Syed et al. 2002) but no interaction of Dsg 1 with Dsg 3 (Heupel et al. 2007). Aggregation assays of transfected cells indicated that in cells, heterophilic binding of Dsgs and Dscs might be of even greater importance than homophilic binding to induce strong intercellular adhesion (Kowalczyk et al. 1996; Marcozzi et al. 1998; Runswick et al. 2001) and that adhesion is strictly dependent on the ratio of the respective Dsgs and Dscs (Getsios et al. 2004a). This view is supported by the recent finding that a conditional Dsc 3-deficiency in mice induced a severe pemphigus-like phenotype with epidermal blistering (Chen et al. 2007). Because antibodies in typical pemphigus are usually not directed against Dsc 3 but against Dsg 1 and Dsg 3, it has to be considered that heterophilic binding of these three molecules is important for epidermal cohesion in vivo. Armadillo family proteins From the Armadillo family, plakoglobin and plakophilins 1–3 are important components of desmosomes. Plakoglobin Plakoglobin (82 kDa), also termed γ-catenin, is the only essential desmosomal component which is also found in typical adherens junctions (Cowin et al. 1986; Franke et al. 1989, 1983). The gene encoding for plakoglobin was mapped to chromosome 17 (Aberle et al. 1995). Plakoglobin binds to the cytoplasmic cadherin-typical sequence of Dsgs and Dscs via its first three armadillo repeat domains (Chitaev et al. 1998). Because the same binding site is required for interaction of plakoglobin to α-catenin, the latter is excluded from desmosomes. Similarly, although the armadillo repeat domain of β-catenin can also bind to Dsg 2, its flanking domains inhibit this interaction, which may explain the absence of β-catenin from desmosomes (Troyanovsky et al. 1996; Wahl et al. 1996). Plakoglobin has been demonstrated to interact with other desmosomal plaque components such as desmoplakin, plakophilins and also with cytokeratin filaments (Bonne et al. 2003; Chen et al. 2002; Kowalczyk et al. 1997; Smith and Fuchs 1998). The importance of this linker function can be concluded from studies in which inactivation of plakoglobin led to embryonic lethality due to mechanical fragility of the myocardium and, when mice are viable, to subcorneal skin blistering indicating that plakoglobin is essential for desmosomal stability (Table 1) (Bierkamp et al. 1996; Ruiz et al. 1996). Besides its function as a desmosomal adaptor protein, plakoglobin seems to be involved in nuclear signalling. It has been shown that plakoglobin, comparable to β-catenin in the canonical wnt signalling pathway, confers transcriptional activity together with TCF-4/LEF transcription factors (Maeda et al. 2004). This mechanism seems to interfere with β-catenin-mediated transcription (Hu et al. 2003). Because plakoglobin like β-catenin is a target of glycogen synthase kinase-3 β, which drives proteosomal degradation of both proteins (Kodama et al. 1999; Muller et al. 2008a; Williamson et al. 2006), a complex pattern of direct and indirect transcriptional regulation seems likely. A target gene of Lef-1/plakoglobin signalling is c-Myc, the expression of which was inhibited in keratinocytes but enhanced in transformed rat kidney epithelial cells (Kolligs et al. 2000; Kolly et al. 2007; Williamson et al. 2006). This indicates that the role of plakoglobin transcriptional regulation is strictly cell type-dependent. In keratinoytes, c-Myc repression by plakoglobin is required to stop proliferation and to allow terminal differentiation (Williamson et al. 2006). Another potential target gene is the anti-apoptotic molecule Bcl–XL, which was found to be upregulated in plakoglobin-deficient cells leading to reduced apoptosis and thus might also be repressed by plakoglobin (Dusek et al. 2007a). Taken together, plakoglobin serves as functional linker between intercellular adhesion and regulation of the cell cycle. This might also be important for cancer progression because many tumors are characterized by loss of plakoglobin expression. Plakophilins Plakophilin 1 (80 kDa) was first identified as an “accessory” plaque protein because, in contrast to plakoglobin and desmoplakin, it was found in cells from certain stratified and complex epithelia only (Franke et al. 1983; Hatzfeld et al. 1994; Heid et al. 1994). Afterwards, it became clear that plakophilin 2 (100 kDa) is ubiquitously present in all desmosomes and also plakophilin 3 (87 kDa) is present in most simple and stratified epithelia (Mertens et al. 1999; Schmidt et al. 1999). The genes encoding plakophilin 1, 2 and 3 are located on chromosomes 1, 12, and 11, respectively (Bonne et al. 1998). Plakophilin 1 and 2 exist in two splice variants with a shorter “a” and a longer “b” form (Mertens et al. 1996; Schmidt et al. 1997). In addition to their localization in desmosomes, plakophilins 1 and 2 are also found in the karyoplasm in a variety of cells and plakophilin 1 b is exclusively located in the nucleus. Plakophilin 2 deficiency leads to embryonic death due to heart defects indicating that the presence of at least one member of the plakophilin family is required (Table 1) (Grossmann et al. 2004). Under these conditions, cytokeratin filaments were retracted from cell borders and desmoplakin was localized in the cytoplasm rather than at desmosomes in cardiomyocytes, demonstrating the relevance of plakophilin 2 to desmosplakin recruitment. Because cardiomyocytes in contrast to epithelial cells express only plakophilin 2 but not plakophilin 1 and 3, defects were present in the heart only, whereas epithelia were not affected. Plakophilins can directly interact with all other desmosomal components including cytokeratins via the aminoterminal head domain (Bonne et al. 2003; Hatzfeld 2007; Hatzfeld et al. 2000). Plakophilin 1 recruits desmosomal components to the cell membrane, increases size and number of desmosomes and therefore seems to be a scaffolding protein, which induces desmosome assembly (Hatzfeld et al. 2000; Kowalczyk et al. 1999; Wahl 2005). On the other hand, because plakophilin 1 is located in the dense inner desmosomal plaque whereas cytokeratin filaments only loop into the outer plaque, it is believed that plakophilin 1 enhances desmoplakin lateral interactions but does not directly associate with cytokeratin filaments in vivo (Hatzfeld 2007; North et al. 1999). It has been demonstrated that desmosome formation is mediated by the aminoterminal domain, whereas recruitment of plakophilin 1 itself to the plasma membrane is dependent on the carboxyterminal region (Sobolik-Delmaire et al. 2006). In addition to its function in the regulation of desmosome assembly, plakophilins may also regulate signalling mechanisms, both at cell borders as well as in the nucleus. Plakophilin 1 associates with actin filaments at cell borders and has been reported to interact with a GTP exchange factor (GEF) for Rho and thereby could regulate activity of Rho GTPases similar to the closely related p120-catenin, which is known to inhibit Rho A and to activate Rac 1 and Cdc42 (Anastasiadis and Reynolds 2001; Hatzfeld 2007). In addition to their localization within the desmosomal plaque, plakophilin 1 and 2 are also present inside nucleus and plakophilin 2 has been found to be part of the polymerase III complex which is responsible for generation of tRNA and rRNA (Mertens et al. 2001). By these two mechanisms, it is possible that plakophilins may regulate cell adhesion and cell growth (Hatzfeld 2007). Plakin family proteins Plakin family proteins are linkers between the cytoskeleton and cell–cell or cell–matrix contacts (Jefferson et al. 2007). Desmoplakin, which exists in two spice variants of a protein encoded by a single gene on chromosome 6 (desmoplakin I: 322 kDa; desmoplakin II: 259 kDa), is an essential component of the desmosomal plaque and therefore is regarded as the prototype of this family (Armstrong et al. 1999; Hatsell and Cowin 2001; Mueller and Franke 1983; Sonnenberg and Liem 2007). Other members such as plectin, envoplakin and periplakin were also found in desmosomes, but their significance for the structure and function of desmosomes is less clear. Especially, plectin is primarily important in hemidesmosmes, which anchor epithelia to the extracellular matrix. Desmoplakin consists of an aminoterminal plakin domain, which can interact with all other desmosomal plaque proteins such as plakoglobin and plakophilins but also with Dsc 1a (Kowalczyk et al. 1997; Smith and Fuchs 1998; Troyanovsky et al. 1994b). The central coiled-coil rod domain, which is important for dimerization, is followed by the carboxyterminal tail consisting of three globular subdomains with several plakin-repeats, which serve as linkers for different intermediate filament types (Choi et al. 2002; Green et al. 1990; Stappenbeck and Green 1992). It is well established that desmoplakin is the main linker protein between the desmosomal cadherin–plakoglobin complex and the intermediate filament cytoskeleton. This has been shown in vitro and was ultimately demonstrated in desmoplakin-deficient mice, which had a reduced number of desmosomes and died at embryonic stage just after implantation (Table 1) (Bornslaeger et al. 1996; Gallicano et al. 1998). Similar to the findings in epidermal-specific desmoplakin-deficient mice, which suffered from skin blistering, desmosomes were not anchored to intermediate filaments (Vasioukhin et al. 2000). Moreover, when desmoplakin was rescued in extraembryonic tissues so that embryos further developed, defects were present not only in the myocardium and epidermis but also in the vasculature and in the neuroepithelium, underlining the importance of desmoplakin for tissue differentiation (Gallicano et al. 2001). Diversity of desmosomes in different tissues and specific epithelial layers Although it was discovered about 25 years ago that the structure of desmosomes is not identical in all types of cells and tissues (Giudice et al. 1984; Jones et al. 1986b), the knowledge on the diversity of desmosomes is still uncomplete and matter of discussion (Garrod et al. 2002; Getsios et al. 2004b; Green and Simpson 2007; Hatzfeld 2007; Holthofer et al. 2007; Kottke et al. 2006; Yin and Green 2004). The diversity of desmosomes has implications for tissue differentiation and also is of high-medical relevance because diseases caused by genetic alteration of or by an autoimmune reponse against a specific desmosomal component may affect only certain but not all desmosome-containing tissues. Some desmosomal components such as Dsg 2, Dsc 2 and the plaque proteins desmoplakin, plakoglobin and plakophilin 2 are ubiquitously expressed in all cells and tissues in which desmosomes are found. Plakophilin 3 is present in most simple epithelia except hepatocytes as well as in stratified epithelia, whereas plakophilin 1 is restricted to stratified and complex epithelia. In epithelia, the desmosomal cadherins show typical expression patterns. Simple epithelia and urothelium usually express Dsg 2 and Dsc 2 only. Apparently, exceptions are the additional presence of Dsg 1 in the mucosa of uterus, stomach, intestine and in epithelia of liver and pancreas as well as the expression of Dsc 1 in intestine and liver or Dsc 3 in stomach, prostate, salivary gland and urothelium. Dsg 4 has a unique tissue distribution in skin and several simple epithelia such as those present in pancreas, salivary glands, testis, prostate and hepatic epithelium. The Dsg 1/Dsc 1 and Dsg 3/Dsc 3 pairs are largely confined to stratified epithelia where the expression patterns of the Dsg and Dsc isoforms usually conform. Interestingly, in the stratified corneal epithelium, only Dsg 1 and Dsc1 are present indicating that these desmosomal cadherins in the absence of Dsg 1 and Dsg 3 are sufficient to maintain cohesion in stratified epithelia also. In the epidermis, the plaque proteins plakoglobin, desmoplakin and plakophilin 3 are expressed in all layers (Fig. 4). In contrast, plakophilins 1 and 2 display inverse distribution with plakophilin 1 being more abundant in the superficial epidermis. These inverse expression patterns are also typical for the Dsg 1/Dsc 1 and Dsg 3/Dsc 3 pairs. Dsg 1/Dsc 1 are the predominant desmosomal cadherins in the superficial epidermis, whereas Dsg 3/Dsc 3 are primarily expressed in the lower epidermis. In contrast to Dsg 1, which can be detected in some desmosomes in keratinocytes of the basal layer also, Dsc 1 and Dsg 4 are absent in the basal layer (Dusek et al. 2007b; Mahoney et al. 2006; Spindler et al. 2007). Because Dsg 3 is expressed throughout the spinous layers (Fig. 5), the expression patterns of Dsg 1 and Dsg 3 largely overlap in human adult epidermis (Mahoney et al. 2006; Spindler et al. 2007). Dsc 2 is enriched in the deep epidermis with lower levels in the superficial epidermis, whereas Dsg 2 is restricted to basal and suprabasal cells but is present in very faint amounts only (Mahoney et al. 2006) indicating that this pair of proteins may be primarily important for cell cohesion in simple epithelia and myocardium rather than in complex epithelia. However, it is unclear at present which desmosomal cadherin isoforms are capable to heretophilically bind to each other and thus interpretation of these distribution patterns with respect to their relevance for mechanical adhesion is preliminary. In multilayered squamous epithelium of mucous membranes, for instance of the oral cavity, Dsg 1 and Dsg 3 are strongly expressed throughout all layers, whereas Dsg 4 shows strong expression in superficial layers but is missing in the basal layer (Mahoney et al. 2006). Fig. 4Expression patterns of desmosomal components in the epidermis. The schematic drawing of the epidermis (left) indicates the basal (BL), spinous (SL), granular (GL) and corneal (CL) layer of the epidermis. On the right, the expression patterns of desmosomal components in the specific epidermal layers are illustrated. For instance, Dsg 1 and Pkp 1 are most prominent in the superficial layers, whereas expression of Dsg 3 and Dsc 3 is strongest in the deep epidermis. Dsg desmoglein, Dsc desmocollin, Pkp plakophilin, PG plakoglobin, DP desmoplakinFig. 5Immunostaining of Dsg 1 and Dsg 3 in human epidermis. Intact human epidermis was immunostained using monoclonal antibodies against Dsg 1 (a) and Dsg 3 (b). A merge of both panels is shown in c. Dsg 1 is most abundant in the superficial epidermis but is also present in the basal layer. Dsg 3 is expressed in the basal layer as well as throughout the spinous layer indicating that in human epidermis the expression patterns of these two proteins broadly overlap. Scale bar is 20 μm It is important to note that the specific distribution patterns of desmosomal components in stratified epithelia are important for epithelial differentiation and function (Green and Simpson 2007). It was shown that forced overexpression of Dsg 3 in the suprabasal epidermis led to abnormal differentiation and hyperproliferation as well as perinatal lethality due to transepidermal water loss (Elias et al. 2001; Merritt et al. 2002). Similarly, forced suprabasal Dsg 2 and Dsc 3 overexpression resulted in hyperproliferation and formation of papillomas, possibly via altered β-catenin/wnt signalling (Brennan et al. 2007; Hardman et al. 2005). Desmosome assembly and disassembly The mechanisms participating in desmosome assembly and disassembly have been reviewed in detail elsewhere (Getsios et al. 2004b; Green and Simpson 2007; Kitajima 2002; Yin and Green 2004). For instance, extracellular Ca2+ and protein kinase C (PKC) signalling are well known to be involved in desmosome assembly. Ca2+ concentrations >0.1 mM allow formation of adherens junctions and desmosomes (Hennings and Holbrook 1983). Desmosomal plaques with inserted cytokeratin filaments became visible as early as after 5 min after the Ca2+ switch followed by appearance of assymetrical desmosomes after 10 min and of symmetric desmosomes after 1 h. Increased extracellular Ca2+ induced incorporation of desmosomal components such as Dsgs, plakoglobin and desmoplakin into the desmosomal plaque (Hennings and Holbrook 1983; Pasdar et al. 1995; Pasdar and Nelson 1988, 1989). Activation of PKC is required for translocation of desmosomal components to the cell membrane and for desmosome assembly (Sheu et al. 1989), but also was found to reduce desmosomal adhesion and to increase Ca2+-dependence of desmosomes (Kimura et al. 2007) indicating that regulation of desmosomal adhesion by PKC is complex. Before desmosome assembly, adhesion zippers of E-cadherin-containing puncta form on filopodial processes of neighbouring cells, an event that requires both α-catenin and VASP-driven actin reorganization (Vasioukhin et al. 2000). Afterwards, these intermediate junctions mature to adherens junctions and desmosomes are assembled at regions where membranes are brought together. It appears that Dscs initiate the formation of desmosomes. This is based on the observations that Dsc 2 is the first desmosomal component at the cell surface followed by Dsg 2 in MDCK cells (Burdett and Sullivan 2002) and that, in keratinocytes, N-terminally deleted Dsc 3, which compromised desmosome formation was still able to bind to β-catenin. Therefore, it can be speculated that Dsc 3 could localize to pre-existing adherens junctions to induce desmosome formation (Hanakawa et al. 2000). Desmosomal cadherins seem to be transported in vesicles from the Golgi along microtubules whereas non-membranous cytoplasmic particles containing desmoplakin and plakophilin are associated with intermediate filaments and move towards cell-junctions by actin-based motility (Godsel et al. 2005; Green and Simpson 2007). Desmoplakin trafficking seems to be dependent on intracellular Ca2+ levels because patients with Darier’s disease, which results from mutations in a sarcoplasmic reticulum Ca2+ pump show desmoplakin retention in the cytoplasm and altered desmosome structure (Dhitavat et al. 2003; Dhitavat et al. 2004; Sakuntabhai et al. 1999). On the cell surface, the desmosomal cadherins together with plakoglobin and desmoplakin are sufficient to nucleate a desmosomal plaque (Kowalczyk et al. 1997). Further plaque enlargement and desmoglein clustering are dependent on plakoglobin together with plakophilin (Bornslaeger et al. 2001; Koeser et al. 2003). Therefore, keratinocytes deficient for either plakoglobin or desmoplakin display reduced numbers of desmosomes, disturbed plaque formation and reduced anchorage of cytokeratin filaments (Bierkamp et al. 1999; Vasioukhin et al. 2001). It appears that during desmosome assembly, Dsg3-containing clusters are formed in the beginning, which, upon attachment to cytokeratin filaments, become integrated in desmosomes (Sato et al. 2000). Once they are formed, desmosomes are stable throughout the cell cycle and are not disrupted during mitosis, although the desmosomal components are subjected to a significant turnover with a half-life of about 30 min like it was shown for Dsc 2a (Windoffer et al. 2002). Finally, it has to be emphasized that a reciprocal dependence of desmosomes and adherens junctions seems to exist. This can be concluded from experiments in which expression of N-terminally deleted Dsc 3 or desmoplakin deficiency resulted in impaired formation of both desmosomes and adherens junctions (Hanakawa et al. 2000; Vasioukhin et al. 2001). Regulation of keratinocyte proliferation by desmosomal cadherins Evidence is accumulating that desmosomal cadherins such as Dsg 3 regulate keratinocyte proliferation (Muller et al. 2008a). It has been shown that autoantibodies from pemphigus vulgaris patients induce continuing keratinocyte proliferation by impaired Dsg 3/plakoglobin signalling, which finally leads to c-Myc overexpression (Muller et al. 2008a; Williamson et al. 2007, 2006). According to this concept, in healthy epidermis Dsg 3 binding results in inhibition of glycogen synthase kinase 3 (GSK3) via activation of phosphatidylinositol trisphosphate kinase (PI3K) and Akt. In consequence, GSK3 phosphorylation-dependent degradation of plakoglobin is abolished which allows plakoglobin to translocate into the nucleus and to induce growth arrest via suppression of c-Myc (Muller et al. 2008a; Williamson et al. 2006). Desmosome-associated diseases Several diseases have been found in which impaired desmosomal adhesion contributes to pathogenesis. Inactivation of desmosomal function may be reduced by completely different mechanisms including genetic defects of desmosomal components, cleavage of desmosomal cadherins by bacterial toxins and binding of autoantibodies to desmogleins 1, the latter of which is the cause of the autoimmune disease pemphigus. Although altered expression of desmosomal cadherins such as Dsg 2/Dsc 2 and Dsg 3/Dsc 3 have been observed in human carcinomas such as squamous cell carcinoma as well as gastric, colorectal and breast carcinomas, mutations are usually absent (Bazzi and Christiano 2007). Therefore, the role of desmosomal cadherins in cancer is unclear at present. Genetic diseases Mutations in desmosomal plaque components in humans affect the myocardium as well as the epidermis with its appendages (Table 1). Mutations in genes for the essential desmosomal plaque components desmoplakin and plakoglobin result in heart, skin and hair defects (Bazzi and Christiano 2007; Green and Simpson 2007; McGrath 2005). In contrast, genetic alterations of Dsg 2, Dsc 2 and plakophilin 2 selectively lead to heart defects because these are the only isoforms of their protein families in the myocardium. On the other hand, mutations in Dsg 1 and plakophilin 1, which are primarily expressed in the epidermis, cause skin defects whereas loss of Dsg 4 in hair follicles results in hair loss. Genetic heart defects Interestingly, all defects of desmosomal components causing heart defects such as desmoplakin, plakoglobin, plakophilin 2, Dsg 2 and Dsc 2 lead to the phenotype of arrhythmogenic right ventricular cardiomyopathy (ARVC) which is clinically characterized by right bundle block and arrhythmia and histologically by fibrofatty replacement of cardiomyocytes, possibly due to impaired cell adhesion caused by loss and alterations of desmosomes (Asimaki et al. 2007; Gerull et al. 2004; McKoy et al. 2000; Pilichou et al. 2006; Rampazzo et al. 2002; Syrris et al. 2006; Heuser et al. 2006). This is in line with embryonic lethality due to myocardial rupture in mice models deficient in these proteins. Therefore, the thinnest parts of the right ventricle are the most severely affected, but left ventricle involvement also occurs (van Tintelen et al. 2007). However, fibrofatty transdifferentiation of cardiomyocytes cannot be simply explained by impaired desmosomal adhesion, but rather seems to be caused by altered wnt/ β-catenin signalling in response to nuclear translocation of plakoglobin (Garcia-Gras et al. 2006). Genetic defects of skin and its appendages Haploinsufficiency of the gene encoding Dsg 1 results in the autosomal dominant skin disease striate palmoplantar keratoderma (SPPK), which is characterized by linearly arranged thickening of the stratum corneum on the palms, soles, knees, ankles and finger knuckles (Milingou et al. 2006; Rickman et al. 1999). However, blisters are absent indicating that disturbed differentiation is the primary mechanism underyling this entity rather than a loss of keratinocyte adhesive function. Similarly, mutated Dsg 4 leads to autosomal recessive inherited hypotrichosis due to defective hair follicle differentiation, a phenotype related to the lanceolate hair mouse (Kljuic et al. 2003). In contrast, ablation of plakophilin 1 results in the recessive skin-fragility ectodermal dysplasia syndrome, which in 1997 was the first genetic desmosome-associated disease to be described (McGrath 2005). Here, both loss and alterations of desmosomes and lacking insertion of cytokeratin filaments due to inability to recruit desmoplakin cause skin blistering around the mouth as well as on palms and soles accompanied by dystrophic hair and nails (McGrath et al. 1997; McMillan and Shimizu 2001). Mutations in plaque proteins with involvement of various tissues Mutations in plakoglobin are the cause of Naxos disease in which ARVC and palmoplantar keratoderma are associated with woolly hair (McKoy et al. 2000). Interestingly, in contrast to plakoglobin-deficient mice (Bierkamp et al. 1996), acantholysis is absent indicating that some mechanical functions of plakoglobin are maintained in these patients. The most variable phenotypes are the consequence of desmoplakin alterations. An autosomal recessive disorder with dilated cardiomyopathy, keratoderma and woolly hair called Carvajal syndrome is comparable to Naxos disease (Norgett et al. 2000). Haploinsufficiency leads to SPPK, whereas non-sense mutations are accompanied by skin fragility leading to blisters in the face as well as on extremities and trunk and also with wolly hair (Armstrong et al. 1999; Whittock et al. 1999, 2002). However, the most severe disorder is lethal acantholytic epidermolyis bullosa, which is caused by C-terminally truncated desmoplakin and was fatal in a 10-day-old hair and nailless newborn due to extensive blistering leading to transcutaneous fluid loss (Jonkman et al. 2005). On the ultra-structural level, desmosomes were reduced in desmoplakin-related SPPK similar to SPPK caused by mutations in Dsg 1 (Wan et al. 2004), but not in lethal acantholytic epidermolysis bullosa. However, desmosome shedding, alterations of desmosomal plaques and impaired cytokeratin insertion were typically associated with desmoplakin mutations (Jonkman et al. 2005; Norgett et al. 2000; Wan et al. 2004). At present, it is unclear why different mutations affect different tissues. Infectious diseases Staphylococcal scalded skin syndrome (SSSS), which was first described by Ritter von Rittershain in 1878, is the systemic variant of epidemic pemphigus neonatorum or sporadic bullous impetigo and is characterized by superficial epidermal splitting accompanied by fever, erythema and skin tenderness (Farrell 1999; Lyell 1983; Stanley and Amagai 2006). Most cases are caused by staphylococcal exfoliative toxin (ET), a serine protease, which has been shown to selectively cleave Dsg 1 between EC 3 and 4 in conformation-dependent manner, but not Dsg 3 or E-cadherin (Table 1) (Amagai et al. 2000a, 2002; Hanakawa et al. 2002; Melish and Glasgow 1970). Assuming that ET does not cleave other superficially expressed desmosomal cadherins such as Dsg 4 or Dsc 1, SSSS is a good example that extensive epidermal blistering can be induced by proteolytic cleavage of a single adhesion molecule. According to its bacterial pathogenesis, SSSS can be effectively treated with antibiotics (Stanley and Amagai 2006). Pemphigus Pemphigus is an autoimmune blistering skin disease, which is characterized by intraepidermal blistering (Lever 1953). The two major types of pemphigus are the more severe pemphigus vulgaris (PV), which accounts for 80–90% of cases, and pemphigus foliaceus (PF) (Bystryn and Rudolph 2005; Schmidt et al. 2000; Stanley and Amagai 2006). Pemphigus is a rare disease with a yearly incidence of 0.75–5 cases per million and apart from being present in humans, is also found in horses, dogs and cats. In contrast to other autoimmune diseases, which primarily affect women, pemphigus is equally distributed between both genders, and is diagnosed mostly between the fourth and sixth decades. In PV, there are two main forms, the mucosal-dominant and the mucocutaneous type. In both cases, the disease most commonly begins with painful non-healing ulcerations not only in the mucous membranes of the mouth, but also in the larynx, nose and vagina. Later on, flaccid blisters may occur on the scalp, trunk, groin and axillae, which easily rupture, leaving sharply outlined erosions and heal without scarring (Fig. 6). In contrast to PV, PF only affects the epidermis and because epidermal splitting is restricted to the superficial epidermis, lesions appear as crusted erosions on the upper torso, face and scalp. It has to be mentioned that pemphigus can be induced by drugs such as penicillamine, penicillin, captopril and β-blockers and also can occur as a paraneoplastic entity accompanying or preceding lymphoma and lung carcinoma (Yeh et al. 2003). Moreover, in South America an endemic form of PF, called Fogo selvagem, exists, which is thought to be transmitted by insect vectors (Aoki et al. 2004; Diaz et al. 1989). Currently, the therapy of pemphigus is based on immunosupression and reduction of autoantibody load. Conventional therapy includes high-dose corticosteroids, intravenous immune globulin and cytotoxic agents. Before systemic corticosteroids were available, 75% of PV patients died within a year. Second-line therapies for refractory PV include rituximab, an antibody directed against B cell CD20, which reduces autoantibody-producing B cells as well as plasmapheresis to physically remove autoantibodies (Ahmed et al. 2006; Shimanovich et al. 2008). Fig. 6Clinical phenotype of Pemphigus vulgaris and Pemphigus foliaceus. Patients suffering from the mucocutaneous form of pemphigus vulgaris (PV) usually have flaccid blisters and erosions on the trunk (a) accompanied by mucosal ulcerations in the mouth (b). In contrast, Pemphigus foliaceus patients are characterized by crusted epidermal erosions (c) whereas involvement of mucous membranes is absent Histology and autoantibody profile in pemphigus Besides the clinical phenotype, diagnosis of pemphigus is based on histology and the patients’ autoantibody profile (Bystryn and Rudolph 2005). The histologic hallmark of pemphigus is acantholysis, i.e. loss of cell–cell adhesion between keratinocytes. In PV, the epidermal cleavage plane is located in the deep epidermis, usually right above the basal layer (Fig. 7). In contrast, in PF, epidermal splitting occurs between granular layers. Skin blisters can be induced by rubbing on healthy-appearing epidermis, a phenomenon referred to as Nikolsky sign. Fig. 7Typical histology of epidermal lesions from pemphigus patients. Hematoxylin eosin-stained paraffin sections from PV (a) and PF (b) patients showed suprabasal epidermal cleavage in the PV and superficial granular blistering in PF. Scale bar is 50 μm Autoantibodies in pemphigus are sufficient to cause blistering in human skin in vivo and in vitro (Anhalt et al. 1982; Schiltz and Michel 1976). In contrast to other autoimmune blistering skin diseases such as bullous pemphigoid or epidermolysis bullosa acquisita (Sitaru and Zillikens 2005; Yancey 2005), pemphigus antibodies do not require the complement system or leukocytes to induce blisters in vivo (Anhalt et al. 1986). It is generally accepted that PV and PF are characterized by different autoantibody profiles, which generally correlate with disease activity (Bystryn and Rudolph 2005; Harman et al. 2001; Ishii et al. 1997; Stanley and Amagai 2006; Stanley et al. 1984; Yeh et al. 2003). Patients suffering from mucosal-dominant PV usually have antibodies directed against Dsg 3 but not Dsg 1, whereas mucocutaneous PV is characterized by both Dsg 3 and Dsg 1 autoantibodies (Amagai et al. 1999; Ding et al. 1997; Jamora et al. 2003; Miyagawa et al. 1999). In contrast, in PF patients usually antibodies against Dsg 1 but not Dsg 3 are detected (Amagai et al. 1999). However, it is also known that in some cases this correlation between the clinical phenotype and the autoantibody profile was not found (Baykal et al. 2002; Jamora et al. 2003; Yoshida et al. 2005; Zagorodniuk et al. 2005). Over the last decade, there is a debate whether these autoantibodies against desmosomal cadherins are pathogenic (Amagai et al. 2006). It has been shown by passive transfer of affinity-purified Dsg antibody fractions as well as by depletion of pathogenic activity by absorption against desmoglein extracellular domains that Dsg 1 antibodies in PF and the combination of Dsg 1 and Dsg 3 autoantibodies in PV as well as in paraneoplastic pemphigus are sufficient to induce skin blistering (Amagai et al. 1995, 1994a, 1992, 1991, 1998; Koulu et al. 1984; Mahoney et al. 1999). An active PV mouse model in which Dsg 3-deficient mice were immunized with Dsg 3 before splenocytes from these animals were transferred to lymphopenic Rag-2-deficient mice supported the notion that Dsg 3 antibodies alone can cause mucosal erosions (Amagai et al. 2000b). Similar in keratinocyte cultures, depletion of Dsg 1-specific antibodies from PF-IgG by preincubation with recombinant Dsg 1 but not after preincubation with VE-cadherin completely abolished keratinocyte dissociation (Waschke et al. 2005). Pemphigus IgG were found to include a plethora of more than 20 different autoantibodies against keratinocyte antigens such as antibodies against Dsg 1, Dsg 4, Dsc 1-3, desmoplakin, plakoglobin and E-cadherin and several other proteins not associated with cell junctions (Amagai et al. 2006; Evangelista et al. 2008; Kljuic et al. 2003; Korman et al. 1989; Nguyen et al. 2000c). For instance, in all PF sera as well as in 79% of mucocutaneous PV sera, autoantibody activities against E-cadherin were detected, most of which were due to Dsg 1 autoantibodies cross-reacting with E-cadherin (Evangelista et al. 2008). Some of the different autoantibodies have clearly been shown not to be pathogenic such as the Dsg 4-cross-reacting Dsg 1 antibodies in PF (Nagasaka et al. 2004). Therefore, similar to other autoimmune diseases, the pathogenetic relevance of autoantibodies against a specific protein in pemphigus has to be challenged until it has been convincingly demonstrated (Amagai et al. 2006). However, it has been reported that antibodies others than those directed to desmogleins also contribute to epidermal blistering because PV-IgG not containing Dsg 1 antibodies were effective to cause blistering in Dsg 3-deficient mice (Nguyen et al. 2000c). It is possible that these antibodies include antibodies to cholinergic receptors and to pemphaxin, which have both been detected in 85% of PV and PF sera (Grando 2006a; Nguyen et al. 2000b). The pathogenic relevance of antibodies against cholinergic receptors was concluded from experiments where preincubation of monkey oesophagus with PV-IgG blocked staining by a rabbit acetylcholine receptor antibody and the fact that this antibody induced keratinocyte dissociation in culture (Nguyen et al. 2000a). However, antibodies against pemphaxin alone were not sufficient to induce skin blistering (Nguyen et al. 2000b). Moreover, it has not been demonstrated so far that autoantibodies from pemphigus patients, which target cholinergic receptors are capable to induce acantholysis. Therefore, at present it is safe to believe that epidermal blistering in pemphigus is primarily caused by antibodies against Dsg 1 and Dsg 3. These pathogenic antibodies in PV and PF mainly belong to the IgG 4 and IgG 1 subclasses (Bhol et al. 1995; Rock et al. 1989; Spaeth et al. 2001). The knowledge that autoantibodies against desmosomal cadherins are sufficient to induce acantholysis in complement- and leukocyte-independent manner makes pemphigus one of the best-characterized models to study the direct mechanisms underlying autoimmune diseases. Besides the role of autoantibodies, the contribution of Dsg 3 autoreactive T helper (Th) cells has also been characterized for PV and endemic PF (Hertl et al. 2006). Th1 and Th2 cells in PV recognize the extracellular domain of Dsg 3 when presented on the HLA class II alleles HLA-DRβ1*0402 and HLA-DQβ1*0503, whereas in Fogo selvagem patients HLA-DRβ1*0402 and HLA-DRβ1*0101 were most common. In PV patients as well as in healthy carriers of the PV-associated HLA II alleles, Dsg 3 and Dsg 1-autoreactive T cells were found. In healthy individuals, Th1 cells with characteristics of regulatory T (Tr1) cells which inhibit T cell activation were most prevalent. In contrast, in PV patients the levels of Tr1 cells were reduced while Th2 cells were increased (Veldman et al. 2004). Therefore, it is possible that an imbalance of autoreactive Tr1 and Th 2 cells plays a role in the induction of PV by promoting the proliferation of anti-Dsg 3 producing B cells. The mechanisms underlying pemphigus skin blistering As a first concept it was proposed that proteolytic cleavage of molecules responsible for intercellular adhesion was the mechanism underlying pemphigus skin blistering. Later on, with the identification of desmosomal cadherins as the target antigens of pemphigus autoantibodies and with more sophisticated cell biologic tools at hand, the ideas of direct antibody-mediated inhibition and of indirect signalling-mediated reduction of desmoglein binding were developed (Fig. 8). Fig. 8The two principal mechanisms underlying pemphigus skin blistering. Two principal mechanisms have been proposed by which autoantibodies specific for Dsg 1 and Dsg 3 could impair desmosomal adhesion. First, antibodies could directly interfere with desmoglein transinteraction (a). Second, antibody binding has been shown to trigger intracellular signalling pathways, which indirectly results in loss of desmoglein-mediated binding (b) Proteolytic cleavage of desmosomal cadherins Proteolytic cleavage of cell adhesion molecules has first been suggested to be involved in pemphigus acantholysis because protease inhibitors blocked pemphigus IgG-induced cell detachment in culture (Farb et al. 1978). Moreover, plasminogen activator activity and expression of the urokinase-type plasminogen activator receptor (uPAR) system were found to be increased following treatment with PV- and PF-IgG/serum in keratinocytes in vitro as well as in PV patients’ skin (Feliciani et al. 2003; Hashimoto et al. 1983; Lo Muzio et al. 2002; Schaefer et al. 1996; Seishima et al. 1997; Yamamoto et al. 2007b), possibly via phospholipase C (PLC)-mediated signalling (Esaki et al. 1995). Anti-uPA antibodies and a PA inhibitor were sufficient to block acantholysis induced by PV- or PF-IgG in several studies (Feliciani et al. 2003; Hashimoto et al. 1983; Morioka et al. 1987) but not in that by Schuh et al. (2003). However, a definitive evaluation of the uPAR system for pemphigus acantholysis became possible since a study using uPA- and tissue PA-deficient mice showed extensive skin blistering in response to PV- and PF-IgG (Mahoney et al. 1999). Thus, the plasminogen activator system does not appear to be essential for pemphigus skin blistering but may aggravate the phenotype, especially when secondary inflammatory mediators such as IL-1α and TNF-α are released (Feliciani et al. 2000, 2003). The same may hold true for other proteases such as matrix metalloproteinases (MMP) or proteases of the ADAM (a disintegrin and metalloproteinase) family. MMP-9, which was overexpressed but not activated following treatment with PV serum was reported to specifically cleave Dsg 3 during apoptosis (Cirillo et al. 2007a, d). ADAM17 on the other hand, was upregulated by activation of the epidermal growth factor receptor (EGFR) and caused shedding of Dsg 2 (Bech-Serra et al. 2006; Santiago-Josefat et al. 2007). These results may be important for PV because EGFR activation was observed following treatment with PV-IgG (Chernyavsky et al. 2007a; Frusic-Zlotkin et al. 2006). The presence of proteolytic enzymes in PV sera may also explain why IgG-depleted PV sera were found to be pathogenic in culture (Cirillo et al. 2007c). However, because no direct evidence was provided that MMP-9 or ADAM17 or any other proteinase cleaves members of the Dsg family in pemphigus, the significance of these findings for acantholysis in pemphigus is unclear and the specific proteolysis hypothesis proposed for pemphigus requires further experimental substantiation (Cirillo et al. 2008). Nevertheless, the fact that specific cleavage of Dsg 1 by staphylococcal exfoliative toxin in bullous impetigo is sufficient to cause a histologic phenotype comparable to PF (Amagai et al. 2000a; Hanakawa et al. 2002) indicates that, in principle, specific proteolysis could be an effective mechanism in pemphigus. Direct inhibition of desmoglein binding Since it was discovered that autoantibodies in pemphigus are directed to desmosomal adhesion molecules, it was believed that these autoantibodies might directly interfere with desmoglein binding (Fig. 8) (Amagai et al. 1991; Jones et al. 1986a; Koulu et al. 1984), a mechanism also refered to as “steric hindrance”(Sharma et al. 2007). This model is attractive because it has been shown that autoantibodies against Dsg 3 and Dsg 1 in PV and PF patients primarily target the aminoterminal part of the EC 1 domain (Futei et al. 2000; Hacker-Foegen et al. 2003; Ishii et al. 2008; Muller et al. 2008b; Sekiguchi et al. 2001). The EC 1 domain, according to morphologic studies on desmoglein transinteraction in desmosomes, is increasingly recognized as the part of the desmosomal cadherin ectodomain, responsible for trans-interaction (Al-Amoudi et al. 2007; He et al. 2003) and may harbour the putative transadhesive interface, based on data from the crystal structure of classical cadherins (Boggon et al. 2002; Overduin et al. 1995; Shapiro et al. 1995). Moreover, it seems that autoantibody reactivity to the aminoterminal parts (EC 1) of the Dsg 3 ectodomain correlates with high disease activity as well as epidermal or mucosal involvement in PV although the titers of these antibodies do not show this correlation (Amagai et al. 1992; Muller et al. 2006, 2008b; Salato et al. 2005). First functional data that anti-Dsg 3 antibodies in PV may directly interfere with Dsg 3 binding were provided using monoclonal antibodies derived from the active PV mouse model (Amagai et al. 2000b). AK 23, which was directed against the aminoterminal part of EC 1 was found to be pathogenic and capable to induce epidermal blistering in vivo, at least when PF-IgG or exfoliative toxin A was added to inactivate Dsg 1 (Shimizu et al. 2005; Tsunoda et al. 2003). Antibodies to other parts of the Dsg 3 extracellular domain such as AK 9 and AK 18 were ineffective to induce blisters. Recently, by using single-molecule atomic force microscopy (AFM), it was shown that PV-IgG as well as AK 23 directly interfered with homophilic Dsg 3 binding under cell free conditions (Heupel et al. 2007) which supports the hypothesis of direct inhibition of Dsg 3 binding in PV (Stanley and Amagai 2006). However, no direct inhibition of Dsg 1 binding by PV-IgG and PF-IgG was detected by AFM. These autoantibodies induced keratinocyte dissociation and reduced binding of both Dsg 3- and Dsg 1-coated microbeads to the surface of cultured keratinocytes, as revealed by laser trapping (Heupel et al. 2007; Waschke et al. 2005). These data suggest that autoantobodies interfere with Dsg 1 binding rather by indirect, cell-dependent mechanisms. Finally, it has to be noted that, if direct inhibition occurs, it is not possible to discriminate at present whether interference with Dsg 3 binding in PV was mediated by steric hindrance, i.e. by blocking trans-interaction of desmoglein molecules by the bound autoantibody, or rather by allosteric effects, i.e. autoantibody-induced conformational changes of the Dsg 3 ectodomain, which in turn interfere with Dsg 3 transinteraction. An antibody directed against the putative transadhesive interface may directly induce steric hindrance, whereas antibodies directed against other parts of the desmoglein ectodomain could indirectly inhibit desmoglein binding by allosteric mechanisms. The fact that AK 18 and AK 9, which were directed to the middle and the carboxyterminal parts of the Dsg 3 ectodomain, were not pathogenic and did not directly interfere with Dsg 3 binding suggests that these specific antibodies were not capable of causing allosteric hindrance (Heupel et al. 2007; Tsunoda et al. 2003). On the other hand, an antibody directed against the EC2 domain, although this part of the molecule may not be involved in transinteraction, might be large enough to cause steric hindrance of Dsg 3 transinteraction. Therefore, “direct inhibition”, instead of “steric hindrance” of desmoglein binding should be used until discrimination between steric and allosteric effects is possible. Desmoglein compensation in pemphigus The desmoglein compensation hypothesis was proposed to explain the different clinical phenotypes of PV and PF on the basis of their different autoantibody profiles (Amagai 2003; Payne et al. 2004; Sharma et al. 2007; Shirakata et al. 1998; Stanley and Amagai 2006; Udey and Stanley 1999). According to this concept, in the deep epidermis which contains both Dsg 1 and Dsg 3, Dsg 3 compensates for the functional loss of Dsg 1 induced by Dsg 1-specific autoantibodies, resulting in more superficial blistering in PF (Fig. 9). In PV, when only Dsg 3 antibodies are present, no epidermal blistering would occur because Dsg 1 is considered to compensate for autoantibody-induced loss of Dsg 3 binding. However, acantholysis occurs in mucous membranes where Dsg 3 is assumed to be the predominantly expressed Dsg isoform, leading to the phenotype of mucosal-dominant PV. When autoantibodies to Dsg 1 are also produced in PV, epidermal blistering occurs. However, it is unclear why the cleavage plane is restricted to the deep epidermis in PV since in PF Dsg 1 autoantibodies cause superficial blistering (Fig. 10). For this reason and other reasons such as the cases of pemphigus in which the autoantibody profiles do not correlate with the clinical phenotype or the presence of other desmosomal cadherin isoforms in the epidermis, this concept has been challenged (Amagai et al. 2006; Bystryn and Rudolph 2005; Muller et al. 2002; Spindler et al. 2007). Fig. 9The desmoglein compensation hypothesis. Based on the different autoantibody profiles in PV and PF together with the findings that Dsg 3 is present in the deep epidermis only whereas Dsg 1 is primarily expressed in the superficial epidermis, the desmoglein compensation hypothesis has been proposed to explain the epidermal cleavage planes typical for PV and PF. According to this model, blistering in PF affects the superficial epidermis because Dsg 3 is present in the deep epidermis to compensate for the autoantibody-induced loss of Dsg 1 binding. In PV, epidermal involvement would occur only when autoantibodies against both Dsg 1 and Dsg 3 are present because Dsg 1 is found in all epidermal layers and could compensate for loss of Dsg 3 binding when antibodies to Dsg 3 are solely presentFig. 10Immunostaining of Dsg 1 and Dsg 3 in PV lesional epidermis. Epidermis from a patient with mucocutaneous PV was stained for Dsg 1 (a) and Dsg 3 (b). A merge of both panels is shown in c. Both Dsg 1 and Dsg 3 are expressed in the basal layer underneath the blister as well as in keratinocytes in the blister roof. However, Dsg 3 staining appears to be fragmented throughout the epidermis whereas Dsg 1 staining is more continuous. Note that in the level of the cleavage plane the apical membrane of basal cells shows strong immunostaining for Dsg 1 and Dsg 3 (arrows). Therefore, based on the desmoglein compensation hypothesis the expression patterns of Dsg 1 and Dsg 3 cannot explain why the cleavage plane is located suprabasally in PV but not in other epidermal layers. Scale bar is 20 μm Experimental support for the desmoglein compensation hypothesis in vivo was obtained in mice. It was shown that PF-IgG were sufficient to cause skin blistering in Dsg 3-deficient mice but not in normal mice (Mahoney et al. 1999). In skin layers where Dsg 1 and Dsg 3 were found, autoantibodies against both desmogleins were required for blistering. In line with these findings, forced expression of Dsg 3 in the superficial epidermis abolished the ability of PF-IgG to induce acantholysis in mice (Wu et al. 2000). In contrast, in human skin and in cultured human keratinocytes in vitro, PF-IgG were effective to induce acantholysis despite of the presence of both Dsg 1 and Dsg 3 (Spindler et al. 2007). The discrepancy between these conflicting findings may be explained in part by the notion that the desmoglein compensation hypothesis is based on the following two assumptions: (1) the expression pattern of Dsg 3 and Dsg 1 do not substantially overlap in epidermal and mucosal layers where the cleavage plane in PV and PF is located. (2) Dsg 1- and Dsg 3-specific autoantibodies only lead to inactivation of either Dsg 1 or Dsg 3, respectively. Because of the latter, the desmoglein compensation hypothesis has been used to promote the idea that autoantibodies reduce Dsg binding by direct inhibition rather than by unspecific proteolysis (Mahoney et al. 1999). Regarding the distribution of Dsg 1 and Dsg 3, it is important to note that Dsg 3 expression patterns in specific epidermal layers are different in mice and men. In mice, expression of Dsg 3 is restricted to the basal and immediately suprabasal epidermal layer (Mahoney et al. 2006, 1999). In human skin, when PV and PF were used for staining, a similar staining pattern was revealed (Amagai et al. 1996; Shimizu et al. 1995). In contrast, when specific antibodies or in situ hybridisation were used for Dsg 3 mapping in human epidermis, it was demonstrated that Dsg 3 is present throughout the spinous layers and thus Dsg 3 distribution showed substantial overlap with expression of Dsg 1 (Arnemann et al. 1993; Mahoney et al. 2006; Spindler et al. 2007). However, immunostaining of human epidermis using another monoclonal antibody detected expression of Dsg 3 in the lower epidermis only (Wu et al. 2000). In oral mucosa, equally strong expression of Dsg 1 and Dsg 3 was found throughout the epithelium when specific antibodies were used (Mahoney et al. 2006), whereas Dsg 1 staining intensity was found to be much lower when PV-IgG were used for immunstaining (Shirakata et al. 1998). Taken together, the expression patterns of Dsg 1 and Dsg 3 broadly overlap in human epidermis and appear to be identical in oral mucosa. With respect to the second assumption the desmoglein compensation is based on, i.e. selective inactivation of Dsg 1 but not of Dsg 3 by Dsg 1-specific antibodies, it was shown recently that both PF-IgG (only containing Dsg 1-specific antibodies) and PV-IgG from patients with only Dsg 3-specific antibodies were equally effective to reduce binding of Dsg 1- and Dsg 3-coated beads to the surface of cultured keratinocytes (Heupel et al. 2007; Spindler et al. 2007). These data indicate that PV-IgG and PF-IgG can reduce binding of Dsg 1 and Dsg 3, at least on the keratinocyte cell surface. Taken together, the relevance of desmoglein compensation for pemphigus pathogenesis in humans cannot be concluded from experiments in mice, especially because distribution patterns of Dsg 1 and Dsg 3 substantially differ in the two species. Therefore, alternative models have to be worked out to explain the different epidermal cleavage planes in PV and PF. These may involve different signalling pathways required for maintenance of desmosomal adhesion in the specific epidermal layers as outlined below. Signalling pathways in pemphigus and in desmosome disassembly Since it has been shown that PV-IgG upon binding to keratinocytes induced a rapid transient increase of intracellular Ca2+ (Seishima et al. 1995), several signalling pathways have been shown to be involved in pemphigus pathogenesis (Fig. 8) (Kitajima 2002; Lanza et al. 2006; Sharma et al. 2007; Sitaru and Zillikens 2005). Interestingly, evidence has been provided that transadhering non-desmosomal cadherins, for instance Dsg 3, are involved in “outside-in” signalling and that binding of pemphigus IgG interferes with this function (Muller et al. 2008a). This can be concluded from experiments which showed that autoantibody binding as well as keratinocyte separation started between desmosomes (Sato et al. 2000; Takahashi et al. 1985) and that non-junctional Dsg 3 and plakoglobin were depleted first before changes in the desmosomal fractions were present (Aoyama and Kitajima 1999; Williamson et al. 2006; Yamamoto et al. 2007a). Interestingly, to trigger Dsg-induced signalling, autoantibody-mediated cross-linking of Dsg 3 and Dsg 1 seems not to be required because monovalent Fab fragments and single-chain variants of PV- and PF-IgG were effective to cause skin blistering in vivo and to disrupt the desmosomal plaque in vitro (de Bruin et al. 2007; Ishii et al. 2008; Payne et al. 2005; Rock et al. 1990). Ca2+, PLC and PKC It was shown that PV-IgG caused a rapid, transient phospholipase C (PLC)-dependent increase of inositol 1,4,5 trisphosphate and of intracellular Ca2+ leading to activation of both PKC and plasminogen activator (PA) (Esaki et al. 1995; Kitajima et al. 1999; Memar et al. 1996; Osada et al. 1997; Seishima et al. 1995, 1999). Because a chelator of intracellular free Ca2+ blocked keratinocyte dissociation in vitro and inhibitors of calmodulin, PLC and PKC were effective to block PV-IgG-induced acantholysis in vivo (Arredondo et al. 2005; Sanchez-Carpintero et al. 2004), it is possible that this signalling pathway may be involved in PV acantholysis. However, because the PA system is not believed to be crucial in this process, PKC signalling may contribute to PV acantholysis by other pathways such as phosphorylation of β-catenin in adherens junctions (Chernyavsky et al. 2007b). This hypothesis is supported by experiments, which showed that keratinocyte adhesion was negatively regulated by pharmacologic PKC activation (Kimura et al. 2007). p38MAPK Activation of p38MAPK at present is the most promising signalling mechanism to be responsible for acantholysis in pemphigus. It has been demonstrated in vivo that p38MAPK and one of its downstream targets, heat shock protein (HSP) 25, were phosphorylated in response to PV-IgG and PF-IgG and that pharmacologic inhibition of p38MAPK abolished blister formation (Berkowitz et al. 2006, 2007b). Similarly, p38MAPK and the human homolog HSP 27 were found to be phosphorylated in skin lesions of PV and PF patients (Berkowitz et al. 2007a). In cultured human keratinocytes, phosphorylation of p38MAPK and HSP27 occured after 30 min of exposure to PV-IgG (Berkowitz et al. 2005; Kawasaki et al. 2006). However, another study found that activity of p38MAPK was not increased before 120 min and that activity peaked after 240 min of PV-IgG treatment (Chernyavsky et al. 2007a). In the latter study it was shown that activation was mediated, at least in part, by Dsg 1- and/or Dsg 3-specific antibodies because Dsg depletion by siRNA reduced p38MAPK activation by 50%. Inhibition of p38MAPK blocked autoantibody-induced keratinocyte dissociation, Rho A inactivation, cytokeratin retraction and reorganization of the actin cytoskeleton (Berkowitz et al. 2005; Chernyavsky et al. 2007a; Waschke et al. 2006). These results demonstrate that p38MAPK is involved in the mechanisms leading to acantholysis and that inhibition of p38MAPK could be a beneficial approach to treat pemphigus patients. Nevertheless, based on the finding that activation of p38MAPK can also be a consequence of cell detachment in rat intestinal epithelium (Rosen et al. 2002), it was proposed that p38MAPK activation is a consequence of cell dissociation (Sharma et al. 2007). However, this is unlikely because significant pemphigus-IgG-induced acantholysis usually takes 12–24 h to occur (Berkowitz et al. 2006; Caldelari et al. 2001; Lanza et al. 2008; Mahoney et al. 1999; Nagasaka et al. 2004; Nguyen et al. 2000c; Shu et al. 2007; Spindler et al. 2007; Takahashi et al. 1985; Tsunoda et al. 2003; Waschke et al. 2005; Waschke et al. 2006; Williamson et al. 2006). The mechanisms by which p38MAPK and HSP27 lead to keratinocyte dissociation are largely unclear but may involve reorganization of the keratinocyte cytoskeleton. It has been shown recently that serine phosphorylation of cytokeratin 8 by p38MAPK induced cytokeratin network disassembly (Woll et al. 2007). Similarly, HSP27 was found to associate with cytokeratin filaments and to inhibit assembly of glial fibrillary acidic protein (GFAP) (Perng et al. 1999). Similarly, HSP25 was found to interact with actin filaments and to prevent formation of aggregates of thermally denatured actin (Panasenko et al. 2003). However, because p38MAPK-mediated phosphorylation of HSP27 or HSP 25 promoted actin assembly and stabilized F-actin against depolymerization in response to cytochalasin D or heat shock (Benndorf et al. 1994; Geum et al. 2002; Guay et al. 1997), one would expect that PV-IgG-induced HSP27 phosphorylation also would stabilize actin filaments. Therefore, it is more likely that HSP27 is part of a salvage pathway in response to PV-IgG and does not directly contribute to PV-IgG-induced actin reorganization. Rho A GTPase Rho GTPases are important regulators of the cytoskeleton and of cell adhesion (Braga and Yap 2005; Bustelo et al. 2007; Fukata et al. 1999; Jaffe and Hall 2005). In the epidermis, Rac 1 was found not to be crucial for maintaining epithelial integrity but for stem cell differentiation, presumably via negative regulation of c-Myc expression (Benitah et al. 2005; Chrostek et al. 2006). Recently, it has been shown that PV- and PF-IgG-induced epidermal splitting, keratinocyte dissociation, as well as loss of Dsg 1 and Dsg 3 binding in vitro were accompanied by p38MAPK-dependent inactivation of Rho A and that specific activation of Rho A by the bacterial toxin cytotoxic necrotizing factor y (CNFy) abolished these effects (Spindler et al. 2007; Waschke et al. 2006). Moreover, toxin-mediated inactivation of Rho A and Rac 1 resulted in epidermal splitting, keratinocyte dissociation and actin reorganization similar to treatment with pemphigus IgG. Especially, inactivation of Rho GTPases by toxin B caused deep epidermal acantholysis comparable to the effects of PV-IgG. Based on this observation, it is tempting to speculate that the cleavage plane in pemphigus may in part be explained by different, layer-specific signalling pathways important for keratinocyte cohesion. Because cytokeratin retraction and actin reorganization in response to pemphigus IgG were also abrogated by activation of Rho A, these results suggested that Rho A could be involved in the regulation of desmoglein cytoskeletal anchorage (Waschke et al. 2006). This idea was supported by Triton extraction experiments which showed that Rho A activation reduced Dsg 3 in the non-cytoskeleton bound fraction. Alternatively, Rho GTPases may regulate endocytosis of desmosomal cadherins similar to their role in E-cadherin turnover (Akhtar and Hotchin 2001; Izumi et al. 2004; Kamei et al. 1999). Interestingly, endocytosis and depletion of Dsg3 are increasingly recognized to contribute to pemphigus acantholysis (Calkins et al. 2006; Yamamoto et al. 2007a). Taken together, Rho GTPases seem to be important for maintenance of desmosomes and activation of Rho A could be used to develop new strategies in pemphigus treatment. These data are in contrast to previous findings, which suggested that Rho GTPases are not involved in the regulation of desmosomal adhesion. This was concluded from experiments in keratinocytes which showed that Rac 1 inactivation by transfection with a dominant-inactive mutant as well as inactivation of Rho A by C3 toxin for 25 min was sufficient to displace E-cadherin but not desmoplakin from cell junctions (Braga et al. 1997). These studies led to the hypothesis that Rho GTPases are important regulators of keratinocyte adherens junctions but not desmosomes (Braga and Yap 2005). However, because inactivation of Rho A by a cell-permeable C3 fusion toxin was sufficient to displace Dsg 3 from cell borders after 180 min, the negative results from the early study might be caused by the shorter incubation period (Waschke et al. 2006). Plakoglobin Besides its involvement in PV pathogenesis, plakoglobin is important for several aspects of desmosomal adhesion. Plakoglobin deficiency resulted in subcorneal acantholysis, loss of desmosomes, and impaired cytoskeletal anchorage of desmoplakin and desmogleins in vivo and in vitro, supporting the notion that plakoglobin is an important cytoskeletal linker and is crucial for desmosome assembly (Bierkamp et al. 1999; Yin et al. 2005). However, at least in part, plakoglobin functions seem to be compensated by other desmosomal plaque proteins such as plakophilin 1 because a plakoglobin-deficient keratinocyte cell line displayed Dsg 3 cytoskeletal anchorage and keratinocyte aggregation similar to wild-type cells (Caldelari et al. 2001). The linker function of plakoglobin seems to be regulated by tyrosine phosphorylation of plakoglobin because following EGFR activation, phosphorylated plakoglobin remained associated with Dsg 2 but not with desmoplakin (Gaudry et al. 2001). Moreover, plakoglobin phosphorylation was shown to be required for EGFR-induced loss of cell adhesion and to regulate binding of plakoglobin to desmoplakin (Miravet et al. 2003; Yin et al. 2005). These data indicate that plakoglobin is required to integrate the effects from extracellular cues and from different signalling pathways in order to allow coordinated modulation of desmosomal adhesion. In addition, plakoglobin regulates the turn over of desmosomal components because the protein levels of desmogleins as well as of plakoglobin and desmoplakin were decreased in plakoglobin-deficient cells (Yin et al. 2005). With respect to pemphigus, it has been shown that plakoglobin is critical for PV pathogenesis because keratinocytes from plakoglobin-deficient mice were resistant to PV-IgG-induced keratinocyte dissociation, cytokeratin retraction and disruption of the desmosomal plaque (Caldelari et al. 2001; de Bruin et al. 2007). These studies provided first evidence that direct inhibition of Dsg binding alone cannot account for acantholysis in PV and raised the hypothesis that plakoglobin could be part of a receptor complex required to transfer the signal from autoantibody-bound Dsg 3 into the cell, a phenomenon referred to as “outside-in” signalling (Muller et al. 2008a). As outlined above, it has been shown that plakoglobin is involved in c-Myc repression, which is required for keratinocyte differentiation and that PV-IgG by depleting noncytokeratin-anchored plakoglobin on the cell surface also led to c-Myc overexpression (Williamson et al. 2006). However, as discussed below, the role of c-Myc signalling in acantholysis is unclear at present. EGFR The involvement of the EGFR in PV pathogenesis has been reported by showing that EGFR activity increased after 30 min and peaked after 60 min of PV-IgG incubation. This was followed by activation of ERK1/2, c-Jun phosphorylation and finally by keratinocyte apoptosis (Chernyavsky et al. 2007a; Frusic-Zlotkin et al. 2006). Inhibition of EGFR reduced EGFR signalling as well as apoptosis and acantholysis (Frusic-Zlotkin et al. 2006). This finding was surprising because EGFR signalling usually is considered to promote cell survival and proliferation (Bazley and Gullick 2005; Muller et al. 2008a). However, because 60 h were required to induce acantholysis, very high concentrations of IgG (5 mg/ml) were used and activation of p38MAPK was not detected in this study, the significance of these results is unclear. EGFR could contribute to PV pathogenesis by stimulating desmosome disassembly. It has been shown that EGF-induced loss of keratinocyte aggregation was mediated by plakoglobin phosphorylation, which led to uncoupling of the desmoglein 2-plakoglobin complex from desmoplakin and thus from the cytoskeleton (Gaudry et al. 2001; Yin et al. 2005). Moreover, EGFR seems to negatively regulate the protein levels of desmosomal cadherins, in part by promoting their metalloproteinase-mediated degradation, as well as the incorporation of desmosomal cadherins into the desmosomal plaque (Bech-Serra et al. 2006; Lorch et al. 2004; Santiago-Josefat et al. 2007). Src The tyrosine kinase Src is activated by PV-IgG within 30 min and seems to contribute to EGFR and p38MAPK activation because inhibition of Src reduced EGFR and p38MAPK phosphorylation by 45 and 30%, respectively (Chernyavsky et al. 2007a). Because Src inhibition reduced PV-IgG-induced loss of keratinocyte cohesion, cytokeratin retraction and apoptosis in vitro and inhition of tyrosin kinases blocked PV-IgG-induced acantholysis in vivo (Chernyavsky et al. 2007a; Sanchez-Carpintero et al. 2004), these results indicate that Src is significantly involved in PV acantholysis. Besides activation of p38MAPK and EGFR, a recent study provided another possible underlying mechanism: Src was shown to directly phosphorylate p120catenin in response to PV-IgG, which correlated with the degree of acantholysis (Chernyavsky et al. 2007b). Because p120catenin is involved in stabilization of classical cadherins such as E-cadherin on the cell surface as well as in cadherin-dependent regulation of Rho A and Rac 1 activity (Alema and Salvatore 2007; Kowalczyk and Reynolds 2004), Src-mediated phosphorylation of p120catenin may be important for PV pathogenesis. Cholinergic receptors The cholinergic system of the human epidermis involves two classes of cholinergig receptors, the nicotinic and the muscarinic acetylcholine receptors (nAChR and mAChR), which in keratinocytes are involved in the regulation of cell–cell and cell–matrix adhesion as well as in cell migration (Grando 2006a). As outlined above, it is unclear whether autoantibodies against cholinergic receptors in pemphigus directly contribute to acantholysis. However, it was shown that cholinergic agonists can ameliorate PV acantholysis in vivo and in vitro (Nguyen et al. 2004b) and that local application of cholinergic agonists such as pilocarpine had therapeutic effect on oral and skin lesions in PV patients (Iraji and Yoosefi 2006; Namazi 2004). Cholinergic agonists increased protein levels of Dsg 1, Dsg 3 and E-cadherin and antagonists to cholinergic receptors resulted in keratinocyte dissociation, which was paralleled by phosphorylation of these molecules (Grando 2006a). Recently, a mechanism by which signalling from cholinergic receptors could directly interfere with PV-IgG-induced effects was demonstrated. The M1 mAChR agonist pilocarpine inhibited PV-IgG-induced acantholysis by reducing serine phosphorylation of β-catenin and Src-mediated tyrosine phosphorylation of p120catenin by activation of the specific protein phosphatases (Chernyavsky et al. 2007b). Similarly, an agonist of the α 7 nAChR reduced p120catenin phosphorylation both on the level of Src activation as well as by activation of the tyrosine phosphatase. Moreover, it was shown that M3 and M4 receptors can lead to activation of Rho A and that carbachol, which activates α 3 and α 7 nAChR, activates Rho A, Rac 1 and Cdc42 (Chernyavsky et al. 2004a, b; Ruiz-Velasco et al. 2002), which also might be involved in the inhibition of PV-IgG-induced acantholysis. Taken together, these data indicate that activation of cholinergic receptors could provide a promising strategy to treat PV patients. Regulation of cell cycle and gene expression Recent data support the hypothesis that PV-IgG change expression patterns of molecules involved in cell cycle regulation and that these molecules might be involved in acantholysis. As outlined above, PV-IgG-induced plakoglobin depletion resulted in c-Myc overexpression in cultured keratinocytes (Muller et al. 2008a; Williamson et al. 2006). Accordingly, c-Myc overexpression was found in the epidermis of PV patients, but interestingly not in PF skin, indicating that this mechanism may only be important for PV (Williamson et al. 2007, 2006). Because pharmacologic inhibition of c-Myc as well as of plakoglobin degradation by blocking GSK 3 abrogated pemphigus-IgG-induced skin blistering, these results indicate that this mechanism could also be important to induce acantholysis. However, the findings that c-Myc overexpression was found after 24 h, whereas PV-IgG-induced loss of cell aggregation was observed as early as after 12 h and that c-Myc overexpression was absent in PF skin indicate that c-Myc overexpression likely contributes to ongoing proliferation of keratinocytes in PV, whereas it seems not to be essential for acantholysis. Therefore, other mechanisms must exist in addition to induce keratinocyte dissociation. These mechanisms may, amongst others, involve expression of genes different from c-Myc, which is suggested by the fact that plakoglobin deficiency resulted in protection of keratinocytes from apoptosis, possibly via overexpression of the anti-apoptotic molecule Bcl-XL (Dusek et al. 2007a). Similar to c-Myc, cyclin-dependent kinase 2 (cdk2), another kinase involved in the regulation of cell cycle progression, seems to be involved in keratinocyte proliferation and acantholysis in pemphigus (Lanza et al. 2008). PV serum increased protein levels of cdk2 and PV serum-induced acantholysis in vivo was abolished by pharmacologic inhibition of cdk2. Moreover, siRNA-mediated down-regulation of cdk2 blocked cell dissociation in cultured keratinocytes indicating that cdk2 may be important for PV acantholysis in vivo and in vitro. However, at present, it is unclear how continuing keratinocyte proliferation in consequence to increased c-Myc and cdk2 signalling contributes to acantholysis in PV. One explanation is that reduced expression of desmosomal components in hyperproliferating cells may foster the loss of keratinocyte cohesion in pemphigus. PV-IgG do not only induce direct signalling effects on cell junctions and the cytoskeleton but also change gene expression patterns, which may also contribute to pemphigus pathogenesis. It has been shown by DNA microarray that PV-IgG within 8 h down-regulated transcription of 198 genes including genes encoding for adhesion molecules including Dsg 3 and desmoplakin, cytoskeletal proteins such as different cytokeratins as well as molecules involved in cell cycle regulation such as p53 and cyclin D2 in cultured keratinocytes and that expression of some of these genes was antagonistically regulated by methylprednisolone (Nguyen et al. 2004a). When PV serum was used for 24 h, the expression of even more genes was altered. In cultured human keratinocytes, transcription of 231 genes was decreased including Dsc 2 and plakophilin 3 whereas transcription of 329 genes was increased. In vivo, expression of 1114 genes was reduced whereas transcription of 349 other genes was inceased (Lanza et al. 2008). Pharmacologic inhibition of cdk2 blunted the effect of PV-IgG on the expression of most of the genes including desmoplakin and also blocked PV-IgG-induced acantholysis. Thus, it is conceivable that altered desmosome formation may be involved in the mechanisms underlying keratinocyte dissociation in pemphigus. Apoptosis Apoptosis has been detected in skin lesions and in perilesional skin of PV and PF patients (Gniadecki et al. 1998; Puviani et al. 2003; Wang et al. 2004a) and hallmarks of apoptosis such as DNA fragmentation, increased expression of pro-apoptotic molecules Fas, FasL, Bax, p53, depetion of anti-apoptotic Bcl-2 and FLIPL as well as activation of caspases 1, 3 and 8 have been observed following treatment of cultured keratinocytes with PV-IgG or PV serum (Arredondo et al. 2005; Baroni et al. 2004; Chernyavsky et al. 2007a; Frusic-Zlotkin et al. 2005, 2006; Pelacho et al. 2004; Puviani et al. 2003; Wang et al. 2004a, b). Hence, compelling evidence has been provided for the presence of programmed cell death in PV although the phenotype of acantholytic cells is different from cells undergoing apoptosis (Arredondo et al. 2005). However, for several experiments prolonged incubation times of 48–72 h (Arredondo et al. 2005; Baroni et al. 2004; Frusic-Zlotkin et al. 2005, 2006; Wang et al. 2004a, b) were required whereas in most studies from the literature, acantholysis was clearly present after 18–24 h (Berkowitz et al. 2006; Caldelari et al. 2001; Lanza et al. 2008; Mahoney et al. 1999; Nagasaka et al. 2004; Nguyen et al. 2000c; Shu et al. 2007; Spindler et al. 2007; Takahashi et al. 1985; Tsunoda et al. 2003; Waschke et al. 2005; Waschke et al. 2006; Williamson et al. 2006). In another study, apoptosis was detected by TUNEL reactivity starting after 6 h of PV-IgG treatment (Chernyavsky et al. 2007a). However, in this study, loss of intercellular adhesion was present after 120 min and thus before onset of apoptosis. Therefore, it is conceivable that apoptosis may be an event parallel to or in consequence of acantholysis. On the other hand, it was reported that activated caspase 3 cleaves Dsg 3 (Weiske et al. 2001) and that caspase and calpain inhibitors can block PV-IgG-induced acantholysis in keratinocyte monolayers and in skin organ culture (Arredondo et al. 2005; Wang et al. 2004a, b; Weiske et al. 2001). Taken together, it is unclear at present to which extent PV-IgG-induced acantholysis is caused by apoptosis. Targets of signalling pathways in pemphigus Pemphigus is a desmosomal disease because pathogenic autoantibodies are directed against Dsg 1 and Dsg 3. Moreover, these antibodies result in the depletion of desmosomal components from the cell surface, alterations of desmosomal plaques and a loss of desmosomes (Aoyama and Kitajima 1999; Aoyama et al. 1999; Calkins et al. 2006; de Bruin et al. 2007; Sato et al. 2000; Shu et al. 2007; Waschke et al. 2006; Williamson et al. 2006; Yamamoto et al. 2007a). However, it is unclear at present whether desmosomes are the primary targets of autoantibodies or whether components outside of desmosomes cause disassembly of desmosomes via mechanisms involving adherens junctions or the cytoskeleton, which finally results in acantholysis. The latter hypothesis is supported by the observation that intercellular spaces between desmosomes widen before desmosomes separate (Takahashi et al. 1985). Effects on desmosomes and adherens junctions It is well established that pemphigus autoantibodies lead to depletion of desmosomal components from desmosomes as well as from the cell surface. Within 30–60 min, PV-IgG induced internalization and lysosomal degradation of cytoskeleton-unbound Dsg 3 together with plakoglobin (Aoyama and Kitajima 1999; Aoyama et al. 1999; Calkins et al. 2006; Sato et al. 2000; Williamson et al. 2006; Yamamoto et al. 2007a). After 24 h, Dsg 3 but not Dsg 2, plakoglobin or desmoplakin was also depleted from the cytoskeletal fractions leading to reduced total cellular Dsg 3 levels (Calkins et al. 2006; Yamamoto et al. 2007a). This effect was mediated by Dsg 3-specific antibodies because monoclonal Dsg 3 antibodies such as AK23 were effective to induce Dsg 3 depletion and this effect correlated with the pathogenic activity of these autoantibodies in vivo and in vitro (Shu et al. 2007; Yamamoto et al. 2007a). However, no strong correlation between Dsg 3 depletion and keratinocyte dissociation was observed. Dsg 3 depletion was observed in DJM-1 cells at 40% confluence as well as in normal human keratinocytes. In contrast, in confluent HaCaT monolayers, a reduction of Dsg 3 half-life but no depletion of total cellular Dsg 3 was detected (Cirillo et al. 2006; Waschke et al. 2006). Because depletion of Dsg 3 was first found in parallel with serine phosphorylation of Dsg 3, it was suggested that phosphorylation might be required for Dsg 3 degradation (Aoyama and Kitajima 1999; Aoyama et al. 1999; Bystryn and Rodriguez 1978; Rodriguez and Bystryn 1977). However, phosphorylation of Dsg 3 was observed in some studies but not in others, which might be due to different cell culture models indicating that Dsg phosphorylation is not necessary for acantholysis (Chernyavsky et al. 2007b; Nguyen et al. 2004a). Some evidence exists that depletion of Dsg 1 may also occur in pemphigus. However, studies to systematically investigate Dsg 1 turnover in pemphigus are lacking. Dsg 1 internalization and decreased immunostaining of Dsg 1 have been observed following treatment with PF-IgG/-serum and PV serum (Cirillo et al. 2007b; Waschke et al. 2005). This is in line with previous findings from the literature, which demonstrated that PF-IgG did not bind to keratinocytes in the lower epidermis of PF patients but strongly labelled basal keratinocytes in the epidermis of healthy subjects suggesting that in PF the lower epidermis was depleted from Dsg 1 (Bystryn and Rodriguez 1978; Rodriguez and Bystryn 1977). Taken together, it is unclear at present to which extent Dsg depletion from desmosomes causes antibody-mediated acantholysis. However, even if depletion of desmosomal components occured secondary to keratinocyte dissociation, it likely would further aggravate the loss of keratinocyte cohesion. The involvement of adherens junctions in pemphigus pathogenesis is unclear at present. It has been shown that formation of adherens junctions is a prerequisite for desmosome assembly and α-catenin-deficient keratinocytes had reduced numbers of desmosmes (Vasioukhin et al. 2000). Because Rho GTPases are known to primarily regulate adhesion mediated by classical cadherins in adherens junctions whereas the requirement of Rho A and Rac 1 activity for maintenance of desmosomal adhesion was reported just recently (Braga and Yap 2005; Fukata et al. 1999; Spindler et al. 2007; Waschke et al. 2006), it was suggested that pemphigus IgG-induced inactivation of Rho A causes disassembly of desmosomes via destabilization of adherens junctions (Sharma et al. 2007). Although alterations of adherens junctions in response to PV-IgG were negligible compared to the effects on desmosomes (Calkins et al. 2006; de Bruin et al. 2007; Muller et al. 2008a; Waschke et al. 2006), this scenario cannot be completely ruled out at present. In fact it was reported recently that PV-IgG lead to phosphorylation of β-catenin and p120catenin by PKC and Src, respectively (Chernyavsky et al. 2007b). However, more studies are required to address the question how catenin phosphorylation in adherens junctions is associated with keratinocyte dissociation in pemphigus. Effects on the cytoskeleton Alterations of the cytoskeleton have been described both in pemphigus skin lesions as well as in keratinocytes exposed to pemphigus autoantibodies,. These effects include dysorganization of cytokeratin filaments, the actin cytoskeleton as well as of microtubules. First, it has been described that cytokeratin filaments become detached from cell junctions and accumulate in the perinuclear region, a phenomenon commonly referred to as “keratin retraction” (Berkowitz et al. 2005; Caldelari et al. 2001; Calkins et al. 2006; Chernyavsky et al. 2007a; Jinbu et al. 1992; Kitajima et al. 1987; Waschke et al. 2006; Wilgram et al. 1961, 1964; Williamson et al. 2006). It was shown that cytokeratin retraction could be detected as early as 4–6 h after addition of PV-IgG, i.e. when first alterations of desmosomes became visible, and was fully established after 24 h (Berkowitz et al. 2005; Calkins et al. 2006; Chernyavsky et al. 2007a). However, because no cytokeratin filament reorganization was observed before cell detachment and the desmosomes of acantholytic cells remained anchored to cytokeratin filament bundles (Kitajima et al. 1986; Shimizu et al. 2004; Spindler et al. 2007), it appears that cytokeratin retraction is not required for acantholysis but may be a secondary event. Recent studies demonstrated that acantholysis in pemphigus was paralleled by alterations of actin filaments including increased stress fiber formation as well as fragmentation of actin filament bundles. Moreover, pharmacological inhibition of pemphigus IgG-induced acantholysis also was effective to reverse the effects on the cytoskeleton (Berkowitz et al. 2005; Spindler et al. 2007; Waschke et al. 2005; Waschke et al. 2006). Similarly, microtubule distribution was found to be altered by pemphigus IgG (Kitajima et al. 1986). Taken together, pemphigus IgG cause profound cytoskeletal reorganization but it is unclear at present whether these effects contribute to acantholysis or are triggered consequent to keratinocyte dissociation. Nevertheless, it is likely that alterations in the cytoskeleton may account for the changes in cell shape during keratinocyte dissociation, which seem to start between existing desmosomes (Bystryn and Grando 2006; Takahashi et al. 1985). The role of the different mechanisms in blister formation As outlined above, both direct and indirect mechanisms finally leading to a loss of desmoglein-mediated adhesion have been found to be involved in pemphigus acantholysis. At present, compelling evidence indicates that acantholysis in PV and in PF is initiated by cellular signalling pathways rather than by direct inhibition of Dsg binding (Fig. 11). This was first shown using plakoglobin-deficient keratinocytes, which were resistant against PV-IgG-induced acantholysis and was further supported by the findings that inhibition of p38MAPK and activation of Rho A is sufficient to abolish PV-IgG and PF-IgG-induced skin blistering (Berkowitz et al. 2005, 2006, 2007b; Caldelari et al. 2001; Waschke et al. 2006). Moreover, low temperature-abrogated keratinocyte dissociation but not antibody binding to the keratinocyte cell surface indicating that direct inhibition of cadherin transinteraction by autoantibody binding is not sufficient to cause acantholysis (Calkins et al. 2006). If extracellular autoantibody-mediated direct inhibition of Dsg binding would be the primary cause of acantholysis, it is hard to imagine how this process should be blocked by modified intracellular signalling or low temperature. It is possible that antibody-mediated direct inhibition of Dsg binding may contribute to trigger cellular signalling events (Muller et al. 2008a; Sharma et al. 2007; Tsunoda et al. 2003). However, because PF-IgG were shown to induce cellular signalling events but not to directly reduce Dsg 1 binding, it seems that direct inhibition of Dsg transinteraction is not essential to alter Dsg-mediated signalling (Heupel et al. 2007; Waschke et al. 2005). Fig. 11The mechanisms involved in pemphigus acantholysis. Accumulating evidence indicates that PV-IgG and PF-IgG initiate keratinocyte dissociation via intracellular signalling pathways including p38 MAPK, Rho A and plakoglobin (PV only). In addition, other mechanisms such as direct inhibition of Dsg 3 binding and Dsg 3 depletion from desmosomes as well as other signalling events seem to contribute to PV pathogenesis, whereas their role for acantholysis in PF is unclear. These mechanisms may account for the more severe clinical phenotype of PV compared to PF. PLC phospholipase C, PKC protein kinase C, cdk 2 cyclin-dependent kinase 2, EGFR epidermal growth factor receptor It is tempting to speculate that the clinical phenotype of PV may be more severe compared to PF because additional mechanisms could be involved in PV (Fig. 11). Indeed, some mechanisms have only been found to contribute to acantholysis in PV but not in PF. For instance, direct inhibition of Dsg 3 transinteraction was found in PV but not in PF under conditions where antibodies caused acantholysis (Heupel et al. 2007). Therefore, it is possible that direct inhibition as a second pathway to reduce Dsg 3 binding in PV may explain the more severe phenotype of PV. Similarly, depletion of Dsg 3 has been convincingly shown in PV whereas depletion of Dsg 1 in both PV and PF is less clear (Yamamoto et al. 2007a). Moreover, altered plakoglobin signalling leading to c-Myc overexpression was found in PV patients’s skin but not in PF (Williamson et al. 2007, 2006). Other pathways have been investigated for PV only whereas data for PF are lacking. As outlined above, inhibitors of calmodulin, PKC, EGFR, cdk2 as well as of tyrosin kinases have been shown to inhibit PV-IgG-induced acantholysis. Finally, it has to be emphasized that desmoglein proteolysis, mechanical stress as well as secondary changes for example in the extracellular Ca2+ concentration might contribute to skin blistering. For instance, acantholysis may foster skin blistering by derangement of the epidermal Ca2+ gradient, which in turn would result in loss of desmosomal cadherin binding, especially when the tight junctions of the granular layer are affected. Concluding remarks The major goal for the future is to elucidate the primary signalling pathways responsible for the diverse effects of pemphigus IgG such as inhibition of desmoglein binding, depletion of desmosomal components, loss of desmosomes, reorganization of the cytoskeleton and finally the induction of acantholysis. Other pathways will be identified to represent salvage mechanisms to rescue keratinocyte cohesion (Grando 2006b). Moreover, future experiments may reveal how the different signalling pathways are involved in the regulation of desmosomal adhesion in the specific epidermal layers. This knowledge may elucidate why the cleavage planes in PV and PF usually are different. The second goal is to characterize whether acantholysis in pemphigus is caused by mechanisms directly targeting desmosomes or whether cellular signalling pathways regulate desmosomal adhesion via reorganisation of the cytoskeleton or via other types of cell contacts such as adherens junctions.

Mol_Immunol-1-5-2075531

The different effector function capabilities of the seven equine IgG subclasses have implications for vaccine strategies

Recombinant versions of the seven equine IgG subclasses were expressed in CHO cells. All assembled into intact immunoglobulins stabilised by disulphide bridges, although, reminiscent of human IgG4, a small proportion of equine IgG4 and IgG7 were held together by non-covalent bonds alone. All seven IgGs were N-glycosylated. In addition IgG3 appeared to be O-glycosylated and could bind the lectin jacalin. Staphylococcal protein A displayed weak binding for the equine IgGs in the order: IgG1 > IgG3 > IgG4 > IgG7 > IgG2 = IgG5 > IgG6. Streptococcal protein G bound strongly to IgG1, IgG4 and IgG7, moderately to IgG3, weakly to IgG2 and IgG6, and not at all to IgG5. Analysis of antibody effector functions revealed that IgG1, IgG3, IgG4, IgG5 and IgG7, but not IgG2 and IgG6, were able to elicit a strong respiratory burst from equine peripheral blood leukocytes, predicting that the former five IgG subclasses are able to interact with Fc receptors on effector cells. IgG1, IgG3, IgG4 and IgG7, but not IgG2, IgG5 and IgG6, were able to bind complement C1q and activate complement via the classical pathway. The differential effector function capabilities of the subclasses suggest that, for maximum efficacy, equine vaccine strategies should seek to elicit antibody responses of the IgG1, IgG3, IgG4, and IgG7 subclasses. 1 Introduction Only recently has the full complement of horse Ig heavy (H) chain constant region genes (one δ, one μ, seven γ, one α and one ɛ) been described (Wagner et al., 2004). The horse has the highest number of IgG constant region (γ or IGHG) genes of any mammalian species examined to date and all the seven IgG subclasses they encode appear to be expressed in vivo (Wagner et al., 2004). Early studies described five equine IgG subclasses named IgGa, IgGb, IgGc, IgG(T), and IgG(B) or ‘aggregating immunoglobulin’. Following identification of the seven horse heavy chain constant region genes, the IgG subclasses have been reassigned as IgG1 to IgG7 (Wagner, 2006). Of the originally described IgG subclasses, IgGa corresponds to IgG1, IgGb to IgG4 and IgG7, IgGc to IgG6, and IgG(T) to both IgG3 and IgG5 (Wagner et al., 2002, 2004). IgG is the predominant antibody class in equine serum and colostrum (Sheoran et al., 2000) and is present at the mucosal surfaces, where it is the most abundant isotype in the equine urinary tract, lower respiratory tract and lung (Butler, 1998). In equine serum, IgGb (IgG4 and IgG7) is the most prevalent isotype followed by IgG(T) (IgG3 and IgG5), IgGa (IgG1) and IgGc (IgG6). In colostrum, IgGb is predominant, followed by IgGa and IgG(T), while IgGc is barely detectable. In nasal wash samples from adult horses only the IgGa and IgGb subclasses have been detected (Sheoran et al., 2000). Systemic and/or mucosal IgG antibody responses are known to play an important role in protection against several equine pathogens including equine influenza virus (Nelson et al., 1998; Breathnach et al., 2006) and Streptococcus equi (Galan and Timoney, 1985; Galan et al., 1986; Sheoran et al., 1997), and may limit the severity and spread of equine herpes virus (EHV)-1 (Kydd et al., 2006). Although a role for equine IgG antibodies in protection against disease has long been recognised, the structures and functions of the individual IgG subclasses are not well characterised. Identification and cloning of the full complement of IgG H chain genes has provided a fresh resource for the study of equine IgG proteins. Here, we describe the first expression of recombinant versions of all seven equine IgG subclasses and present an analysis of their individual physical and biological properties. 2 Materials and methods 2.1 Construction of equine γ H chain expression vectors The mouse VH gene specific for NIP was subcloned as a HindIII-BamHI fragment from a previously described human IgA1 expression vector (Morton et al., 1993) into pcDNA3.1/Hygro (+) (Invitrogen, Paisley, UK) to produce pcDNA3.1Vnip. Isolation of genomic DNA for equine IGHG3, IGHG4 and IGHG7, and cDNA for equine IGHG1, IGHG2, IGHG5 and IGHG6 has been previously described (Wagner et al., 2002, 2004). For construction of the equine IgG3 H chain vector, a 2.5 kb BamHI fragment containing the IGHG3 gene along with 630 bp of 5′ UTR and 577 bp of 3′ UTR was placed downstream of the mouse VH gene in pcDNA3.1Vnip. To construct mammalian expression vectors for the remaining six horse IgGs, each γ constant region was amplified by PCR and placed downstream of the 630 bp horse IGHG3 5′ UTR (also amplified by PCR) within the cloning vector pBluescript II (Stratagene, Amsterdam, The Netherlands). For each subclass, the 5′ UTR and γ constant region cassette was then subcloned into pcDNA3.1VNip downstream of the mouse VH gene. 2.2 Expression of reqIgGs in CHO-K1 cells CHO-K1 cells, which had previously been stably transfected with the mouse λ L chain specific for NIP, were transfected with one of the γ1–γ7 H chain expression vectors, as previously described (Morton et al., 1993). Supernatant from individual resistant CHO clones was screened for Ig production by antigen-capture ELISA as previously described (Morton et al., 1993) except that the detection antibodies used were either rabbit anti-mouse λ L chain–HRP conjugate (0.25 μg/ml in PBS-0.5% Tween [PBS-T]) or goat anti-horse IgG Fc-HRP conjugate (0.4 μg/ml in PBS-T) (both Rockland, Gilbertsville, PA, USA). 2.3 Purification and analysis of reqIgGs ReqIgGs were purified using NIP-affinity chromatography as previously described (Morton et al., 1993), then subjected to gel filtration on a Superose6 column using an ÄKTA FPLC system (GE Healthcare, Little Chalfont, UK). Only fractions corresponding to intact antibody (H2L2) were pooled for further analysis. The antibodies were analysed by SDS-PAGE and Western blotting on 8.4% acrylamide gels under reducing and non-reducing conditions. Gels were stained with Coomassie brilliant blue and Western blots were probed with HRP-conjugated rabbit anti-mouse λ L chain (0.2 μg/ml) or anti-horse IgG antibodies. Horse serum IgG (Sigma–Aldrich, Poole, UK) and recombinant human IgG1 (Pleass et al., 1999) were used as controls. 2.4 Reactivity of commercially available anti-horse IgG antibodies with the reqIgGs The following antibodies were tested for reactivity against the reqIgGs in ELISA: goat HRP-conjugated polyclonal antibodies specific for horse IgGa, IgGb, IgGc or IgG(T), (kindly provided by Serotec, Oxford, UK and Bethyl Laboratories, Montgomery, TX, USA), and mouse monoclonal antibodies (mAb) against horse IgGa (CVS48), IgGb (CVS39), IgGc (CVS53) and IgG(T) (CVS38) (kindly provided by Serotec). Secondary antibody used to detect binding of the mAb was a goat anti-mouse IgG1-HRP conjugate diluted 1/10,000 (kindly provided by Serotec). In addition, HRP-conjugated goat polyclonal anti-horse IgG H + L chain (kindly provided by Bethyl Laboratories) and goat anti-horse IgG Fc (Rockland) were tested in ELISA. Antigen-capture ELISAs were performed as above using 250 ng/ml of each purified reqIgG subclass. Detection antibodies were diluted 1/10,000 (goat polyclonals) or 1/100 (anti-IgGa mAb), 1/200 (anti-IgGb mAb), 1/10 (anti-IgGc mAb) and 1/500 (anti-IgG(T) mAb). The goat polyclonal anti-horse IgGa, IgGb, IgGc and IgG(T) antibodies were also tested in immunoblotting. Purified reqIgGs (1 μg) were electrophoresed under reducing and non-reducing conditions and the detection antibodies were diluted as for ELISA. 2.5 Analysis of reqIgG glycosylation ReqIgGs (1 μg) were electrophoresed on 8.4% acrylamide gels under reducing conditions alone or following reduction and de-glycosylation. Removal of N-linked sugars was carried out using a GlycoPro enzymatic deglycosylation kit (Prozyme, Inc., Leandro, CA, USA) according to the manufacturer's instructions. Following transfer to nitrocellulose, membranes were probed with the appropriate HRP-conjugated goat anti-horse IgG subclass antibody or either Con A-biotin (diluted 1/10,000) or jacalin-biotin (diluted 1/5000) (both Vector Laboratories, Inc., Burlingame, CA, USA) followed by streptavidin-AP (Dako, Ely, UK). Recombinant human IgG1 and IgA1 (Pleass et al., 1999) were used as positive controls. 2.6 Interaction with staphylococcal protein A and streptococcal protein G Interaction of reqIgGs with protein A and protein G was analysed by ELISA. NIP-BSA coated wells were incubated with 100 μl (0.25 μg/ml) of one of the purified reqIgGs or recombinant human IgG1 for comparison. Binding to protein A or protein G was detected by incubation with 100 μl of serial dilutions of either HRP-conjugated protein A (0–50 μg/ml in PBS-T) or HRP-conjugated protein G (0–5 μg/ml in PBS-T) (both Sigma, Poole, UK). Plates were developed with ABTS and absorbance read at 405 nm. 2.7 Functional assays Chemiluminescence assays were carried out as previously described (Pleass et al., 1999) using 10 μg/ml of each reqIgG and equine PBLs, isolated from blood drawn under appropriate licence from the UK Home Office. Complement binding and activation was assessed by ELISA on NIP-BSA coated wells incubated with recombinant human IgG3 (Bruggemann et al., 1987) or each of the reqIgGs (0–10 μg/ml in PBS-T, 100 μl/well) for 1 h at room temperature. Following washing, plates were incubated with human serum diluted 1/100 in veronal buffered saline containing 0.5 mM MgCl2, 2 mM CaCl2, 0.05% Tween-20, 0.1% gelatin and 0.5% BSA for 1 h at room temperature. After washing, plates were incubated with either a 1/800 dilution of sheep anti-C1q-HRP (Serotec) or a 1/500 dilution of biotin-conjugated anti-C5b-9 (Quidel, Santa Clara, CA, USA), followed by streptavidin-HRP (Dako) diluted 1/1000 in PBS-T, 0.5% BSA for 1 h at room temperature. Plates were developed as above. 3 Results 3.1 Analysis of purified antibody by SDS-PAGE and Western blotting Transfection of each of the equine γ1–γ7 H chain expression vectors into CHO cells stably expressing a compatible mouse λ light (L) chain allowed expression of all seven subclasses. Expressed Ig was purified by antigen affinity chromatography followed by FPLC. All seven IgGs eluted from FPLC as a single major peak representing monomer. Minor peaks eluting earlier were attributed to antibody that had aggregated during purification. For each subclass, fractions corresponding to the major peak were pooled and used for all subsequent experiments. SDS-PAGE analysis of FPLC purified reqIgGs confirmed that the reqIgs assembled into intact Igs stabilised by interchain disulphide bridges (Fig. 1A). Under reducing conditions, analysis of the seven IgGs revealed that the IgG2 H chain migrated the most quickly, followed closely by those of IgG4 and IgG7, then IgG6, IgG1, IgG5 and IgG3 (Fig. 1B). Under the more sensitive conditions of Western blotting using an anti-mouse λ L chain antibody to detect the horse IgGs (Fig. 1C), some lower molecular weight species were noted for IgG4 and IgG7 under non-reducing conditions. These may represent species such as half molecules (HL) and L chain dimers (L2), suggesting that a small proportion of the IgG4 and IgG7 molecules lack inter-H and/or H–L disulphide bonds and are stabilized by non-covalent interactions alone. Hence, under the native conditions of FPLC only a single major peak of fully assembled IgG was seen. 3.2 Reactivity of anti-horse IgG antibodies with the reqIgG subclasses Currently available anti-horse IgG antibodies are categorised on the basis of reactivity against horse IgG subclasses defined under the previous nomenclature of IgGa, IgGb, IgGc, and IgG(T). We sought to analyse the reactivity of these antibodies with the seven reqIgG subclasses. The results, obtained by ELISA, are summarised in Table 1. The polyclonal antibody directed against horse IgGa was specific for a single subclass, namely IgG1. Similarly a polyclonal antibody preparation directed against IgGc was found to recognise just one subclass, in this case IgG6. Goat anti-horse IgGb showed strong reactivity with IgG4 and IgG7 but also weakly recognised IgG1. The goat anti-horse IgG(T) polyclonal antibody reacted most strongly with IgG5, but also recognised IgG2 and IgG3. The pattern of recognition of the IgG subclasses by the mAbs was similar to that of the polyclonal antibodies except that anti-horse IgGb mAb recognised IgG4 and IgG7 but not IgG1, and anti-horse IgG(T) mAb strongly recognised IgG5, weakly recognised IgG3 but showed no reactivity with IgG2. Reactivity of both the polyclonal and monoclonal anti-IgGb antibodies with IgG4 and IgG7 is not surprising as these two subclasses show 97% amino acid identity. The polyclonal anti-horse IgG (H + L) and anti-horse IgG Fc antibodies recognised all seven IgG subclasses but gave varying signal strengths in the ELISA (see Table 1). Anti-horse IgG (H + L) gave a strong ELISA signal with IgG1, IgG4, IgG5 and IgG7, an intermediate signal with IgG2 and IgG3, and only a weak signal with IgG6. The observed reactivity is directed towards the heavy chain of the recombinant IgGs only, because no binding of the anti-horse IgG (H + L) was seen with recombinant forms of either equine IgA or human IgA (Morton et al., 1993), both of which carry identical light chains to those of the eqIgGs (data not shown). Anti-horse IgG Fc gave strong ELISA signals with IgG1, IgG4 and IgG7, intermediate signals with IgG3 and IgG5, and only weak signals with IgG2 and IgG6. The reactivities of the anti-horse polyclonal Abs with the horse IgG subclasses when assessed by immunoblotting under non-reducing conditions (Fig. 2A) mirrored those of ELISA. However, under reducing conditions (Fig. 2B) goat anti-horse IgGb recognised IgG4 and 7 but no longer recognised IgG1, suggesting that the IgG1 epitope(s) recognised by this antibody may be discontinuous and depend on the intact disulphide-stabilised structure of the antibody. Several previously described anti-equine IgG monoclonal antibodies were thought to recognise conformational epitopes on equine IgG as they functioned under native conditions but not in immunoblotting (Sheoran et al., 1998). Under reducing conditions, goat anti-horse IgG(T) used at a 1/10,000 dilution recognised IgG3 and 5 but in contrast to non-reducing SDS-PAGE and ELISA no longer reacted with IgG2 but with IgG1 instead. However, recognition of IgG2 under reducing conditions could be achieved with a higher concentration (1/1000 dilution) of detection antibody (data not shown). Under non-reducing conditions, immunoblotting with goat anti-IgGb (Fig. 2A) revealed additional molecular weight species for IgG4 and IgG7 consistent with those observed with anti-L chain blotting. These additional bands may represent species such as free H-chain, H-chain dimers (H2) and half molecules (HL), suggesting, again, an absence of inter-H and/or H–L disulphides in small fractions of these two subclasses. 3.3 Analysis of the glycosylation of reqIgGs All seven reqIgGs contain a conserved potential N-glycosylation site within the CH2 domain. Additional potential sites are found in the CH1 domain of IgG3 and IgG5 and in the CH2 domain of IgG2. To determine whether N-glycans were attached to any of these sites, the reqIgGs were probed under reducing conditions with concanavalin A (Con A), which is specific for N-linked glycans. Con A bound to all reqIgG subclasses confirming that the IgGs are indeed N-glycosylated (Fig. 3A). There was, however, some difference in the intensity of the bands, with IgG2 and IgG3 blotting more weakly than the other subclasses, most likely a reflection of the number, type or accessibility of N-glycans present. By contrast, jacalin, which is specific for O-linked glycans, bound only to reqIgG3 and to the O-glycosylated human IgA1 used as a positive control (Fig. 3B). Treatment of reqIgGs with N-glycanase resulted in a noticeable reduction in the molecular weights of their H chains (Fig. 4), confirming their N-glycosylation status. For the IgG2 subclass, reactivity of the detecting antibody (goat anti-horse IgG(T)) was lost following the removal of N-glycans, suggesting that the epitopes recognized by this anti-horse reagent are in some way glycan-dependent. The IgG5 H chain migrated as two discrete bands following treatment with N-glycanase, possibly reflecting partial removal of the two potential N-linked glycans in a proportion of the molecules treated. 3.4 Interaction with staphylococcal protein A and streptococcal protein G Binding of protein A to reqIgGs was generally much weaker than the binding of protein G (Fig. 5), in keeping with earlier reports (Sheoran and Holmes, 1996; Burton and Woof, 1992). A 10-fold higher concentration range of protein A than protein G was required to see significant binding to equine and human IgG. Protein A showed the highest level of binding to IgG1 and relatively low levels of binding to the other subclasses in the order: IgG3 > IgG4 > IgG7 > IgG5 = IgG2 > IgG6. Sheoran and Holmes (1996) reported moderate adherence of protein A to serum IgGa (i.e., IgG1) but saw no binding of IgGb (IgG4 and 7) or IgG(T) (IgG3 and IgG5). The fact that slight binding of protein A to IgG3, IgG4, IgG5 and IgG7 was observed in our study may be explained by the use of high concentrations of protein A (up to 50 μg/ml) and a sensitive method of assessing the interaction. Sugiura et al. (2000) also reported some binding of equine IgGb and IgG(T) to protein A. In agreement with previous work (Sheoran and Holmes, 1996; Sugiura et al., 2000), binding of IgG6 to protein A was barely detectable above background levels. Protein G showed strong binding to IgG4 and IgG7, closely followed by IgG1. IgG3 displayed intermediate binding, while binding to IgG6 and IgG2 was low, and IgG5 did not bind. Sheoran and Holmes (1996) also showed that the binding affinity of IgGb (IgG4 and IgG7) to protein G was higher than that of IgGa (IgG1). Furthermore, they identified two fractions of IgG(T), one which bound strongly to protein G and one that bound weakly, which most likely correspond to IgG3 and IgG5, respectively. 3.5 Functional assays We studied FcγR-mediated cellular responses by assessing the capacity of the antibodies to trigger a respiratory burst in equine peripheral blood leukocytes (PBL). Recombinant IgG1, IgG4, IgG5 and IgG7, and to a lesser extent IgG3, all triggered strong respiratory bursts, whereas IgG2 and IgG6 elicited only very weak responses (Fig. 6). Binding of C1q and activation of the classical complement pathway by the reqIgGs was assessed using ELISA. Human serum was used as a source of complement due to the ready availability of antibodies to detect human C1q and human C5-9 complex. The C1q–IgG interaction is highly conserved throughout evolution and C1q is known to react with IgG from different species (Burton and Woof, 1992). The ability of the reqIgGs to bind C1q (Fig. 7A) was reflected in their ability to activate the classical complement pathway to its terminal components (Fig. 7B). In control experiments, both C1q binding and C5-9 deposition were reduced to zero for all IgGs following heat inactivation of serum (data not shown). IgG3 was the most potent activator of complement, followed by IgG1, IgG4 and IgG7, which all bound C1q and activated complement to the same extent. Recombinant IgG2, IgG5 and IgG6 did not bind C1q or activate complement. 4 Discussion Our studies show that the CHO expression system previously described for human antibodies (Morton et al., 1993) can be used to generate a stable supply of equine IgG antibodies. These recombinant IgGs retain affinity for their antigen, are recognised by anti-mouse λ L-chain and anti-horse IgG antibodies, are glycosylated, and assemble as covalently stabilised four chain (H2L2) molecules with the expected relative molecular weight of approximately 150 kDa. Hence, in terms of structure, they appear to be representative of natural horse IgGs. Interestingly, horse IgG3 appears to have O-linked as well as N-linked glycans attached to the H chain. O-glycosylated forms of rabbit IgG and mouse IgG2b in which the O-glycans are attached to the hinge region have been described (Fanger and Smyth, 1972a, 1972b; Kim et al., 1994; Kabir et al., 1995). Serine and threonine residues, often in clusters, are known sites for O-glycan attachment. A prominent feature of O-glycosylation sites is an increased frequency of proline residues, especially at positions −1 or +3 relative to the glycosylated residue (Wilson et al., 1991). The horse IgG3 hinge is rich in proline and threonine and contains two Thr residues with a Pro at position −1 and a number of Thr residues with a Pro at position +3. Hence, it is probable that O-glycans are attached to the hinge region of horse IgG3. O-glycosylation of the hinge region of IgG confers an increased resistance to proteolysis by various proteases including pepsin, papain and elastase (Fanger and Smyth, 1972b; Kabir et al., 1995; Parham, 1983) and therefore may play a role in protection of eqIgG3 from degradation. A minor feature of both equine IgG4 and IgG7 was the presence of species such as free H chain, H chain dimers (H2), half molecules (HL) and L chain dimers (L2) under denaturing conditions, suggesting that a small proportion of the molecules lack H–L or H–H disulphide bonds, akin to human IgA2m(1) and human IgG4, respectively. The human IgG2, IgG3 and IgG4 subclasses are disulphide bonded to the L chain through Cys131 in the CH1 domain. The equivalent Cys is missing in equine IgG4 and IgG7, and may account for the tendency for a minor proportion of these subclasses to exist in forms lacking H–L bonds. The majority of equine IgG4 and IgG7 molecules are present at molecular weights corresponding to intact (H2L2) antibodies, suggesting that H–L disulphide bonds can exist. Human IgG1, which lacks Cys131 in the CH1 domain, instead forms a disulphide bond to the L chain through Cys220 in the upper hinge (Burton and Woof, 1992). Hence, it is feasible that hinge cysteines in equine IgG4 and IgG7 may participate in H–L disulphide bond formation. Likewise, this could be the case for equine IgG6, which also lacks the corresponding CH1 Cys residue. The appearance of half molecules of equine IgG4 and IgG7 would suggest that a fraction of both these subclasses lack inter-H chain disulphide bonds, with the molecules being stabilised by non-covalent interactions instead. This arrangement has already been noted in a fraction of human IgG4 molecules which are seen as HL half molecules under denaturing, non-reducing conditions (Aalberse and Schuurman, 2002). The deficiency in inter-H-chain bonds in human IgG4 has been attributed to a Ser within the core hinge sequence (CPSC), which replaces the Pro (CPPC) found in isotypes such as human IgG1 and IgG2 that do not have this deficiency. This Ser seems to increase the likelihood that the two core hinge Cys residues form intra- rather than interchain disulphide bonds possibly due to greater hinge flexibility (Angal et al., 1993; Bloom et al., 1997). In equine IgG1, IgG3, IgG5 and IgG6, which all form disulphide bonded H2L2 molecules, the corresponding core hinge sequence is CPKC. In contrast, the core hinge sequences of eqIgG7 and IgG4 are CPTC and CPAEC respectively, suggesting that deviations from central Lys to either Thr or Ala-Glu may decrease the efficiency of interchain disulphide bridge formation in some way. Our results for the binding of the reqIgGs to protein A and protein G agree well with those for the binding of horse IgG subclasses which had been purified from serum and classified under the previous nomenclature (Sheoran and Holmes, 1996; Sugiura et al., 2000). The binding sites for protein A and G on the human IgG Fc lie at the CH2-CH3 domain interface, with considerable but incomplete overlap (Deisenhofer, 1981; Sauer-Eriksson et al., 1995) (Fig. 8). The amino acid sequences of equine IgG2, IgG3, IgG4, IgG5, IgG6 and IgG7 all contain at least one amino acid difference from human IgG1 within the regions contributing to protein A binding (Fig. 8). These differences may be sufficient to explain the low affinity of these IgGs for protein A because a single amino acid substitution of His to Arg at position 435 in a human IgG3 allotype is known to prevent binding to protein A (Burton and Woof, 1992). Similarly the differential reactivity of the subclasses with protein G can be rationalized on the basis of amino acid substitutions at key positions in the protein G interaction site (Fig. 8). The fact that the highest level of binding is seen with eqIgG1, IgG4 and IgG7, which like human IgG1 all possess His433 and Tyr436, may suggest that these two residues play key discriminatory roles in the binding of horse IgGs to protein G. IgGs mediate effector function through interaction with leukocyte FcγRs and activation of the classical pathway of complement. These effector functions differ with IgG subclass (Burton and Woof, 1992). There is little available information with regard to comparative binding of the horse IgG subclasses to FcγRs. An early study found that eqIgG (subclass unspecified) but not eqIgG(T) (i.e., eqIgG3 and/or IgG5) was able to interact via the Fc region with equine monocytes and neutrophils (Banks and McGuire, 1975). In the present study, we have demonstrated that equine IgG1, IgG3, IgG4, IgG5 and IgG7 are able to elicit a strong respiratory burst from equine PBL, predicting that these five subclasses can interact with FcγR on the surface of these cells. IgG2 and IgG6 stimulated little or no response suggesting that these two subclasses are unable to interact efficiently with FcγR. Several amino acids in the lower hinge region of human IgG1 at the N-terminus of the CH2 domain (Leu234-Leu235-Gly236-Gly237-Pro238-Ser239) (Fig. 8) play a central role in binding to human FcγR and deviations from this motif help to explain human IgG subclass specificity for FcγR (Burton and Woof, 1992; Woof and Burton, 2004). In the horse IgGs, IgG1 and IgG3, which elicit bursts, retain this motif while IgG2 and IgG6, which do not elicit bursts, display significant differences from the motif (Fig. 8). Thus, amino acid differences in this region may account, at least in part, for the observed differences in ability to trigger a respiratory burst. However, despite IgG4, IgG5 and IgG7 lacking the full motif, these subclasses are still capable of eliciting strong respiratory bursts, suggesting that regions other than the lower hinge region may influence interaction with at least some classes of equine FcγRs. Activation of the classical complement pathway by IgG may be initiated through the binding of the first complement component C1q, to two or more adjacent IgG molecules. Human IgG subclasses bind C1q in the rank order IgG3 > IgG1 > IgG2 while IgG4 shows no significant binding (Burton and Woof, 1992). Early studies of equine anti-Lac antibodies (Klinman et al., 1966) using guinea pig complement showed that a preparation comprising IgGa and IgGb (i.e., IgG1, IgG4 and IgG7) was able to fix complement whilst IgGc (i.e., IgG6) was not. The current study confirmed and extended these results. We found that eqIgG3 was the most potent activator of complement, followed by eqIgG1, IgG4 and IgG7, while eqIgG2, IgG5 and IgG6 failed to bind C1q or activate complement. While our results were obtained using human serum as a source of complement, and we cannot rule out the possibility of subtle differences of reactivity with horse complement, we feel that our findings provide a good reflection of the likely relative complement-activating capabilities of the different eqIgG subclasses. The C1q binding site lies within the CH2 domain of IgG, but appears not to be conserved precisely across species. Glu318, Lys320 and Lys322 are necessary for C1q binding by mouse IgG2b, whilst in human IgG1 or IgG3 Asp270, Lys322, Pro329 and Pro331 appear important with minimal contribution from Glu318 and Lys320 (Duncan and Winter, 1988; Idusogie et al., 2000; Thomessen et al., 2000). A comparison of the equine IgG CH2 amino acid sequences does not reveal a common difference between the complement activating and non-activating subclasses at the residues corresponding to Asp270, Glu318, Lys320, Pro329 or Pro331 (Fig. 8). However, all of the complement-activating subclasses have a Lys corresponding to Lys322. The fact that this residue is a serine in eqIgG2 and IgG5 may account, at least in part, for the inability of these subclasses to activate complement. Curiously, IgG6, which possesses a lysine at position 322, does not activate complement, underlining the subtle species differences that are apparent in complement activation. Our data on the effector function capabilities of the IgG subclasses have important new implications for the design of effective horse vaccines. It is clear that to achieve maximal protection via FcγR- and complement-mediated elimination mechanisms, vaccines should seek to elicit IgG antibodies of the IgG1, IgG3, IgG4 and IgG7 subclasses. Vaccines that trigger only IgG2, IgG5 or IgG6 antibodies are predicted to offer less effective protection. Since IgG plays key roles in both serum and mucosal compartments in the horse, these considerations are applicable to both systemic and mucosal vaccination strategies. Further, our findings on the relative effector function capabilities of the different IgG subclasses lend themselves to a rationalization of the protection offered by natural IgG immune responses against various infectious agents. Thus, investigations into the role of individual subclasses in protective immunity have demonstrated that IgGa and IgGb (i.e., the FcγR- and complement-engaging isotypes IgG1, IgG4 and IgG7) contribute to protection against equine influenza virus (Nelson et al., 1998; Breathnach et al., 2006; Soboll et al., 2003), Streptococcus equi (Sheoran et al., 1997) and Rhodococcus equi (Lopez et al., 2002). Indeed, IgGb (IgG4 and IgG7) has been suggested to be most important in equine protective antibody-mediated immune responses to intracellular pathogens (Nelson et al., 1998; Lopez et al., 2002; Goodman et al., 2006), whereas IgGc (IgG6) is suggested to be least important (Sheoran et al., 2000). The current study helps rationalize this observation as both IgG4 and IgG7 are able to stimulate a potent respiratory burst and activate complement via the classical pathway. By contrast, these effector functions are absent in the IgGc (IgG6) subclass. The requirement of equine IgGs to recruit effector molecules for effective immunity is illustrated by several studies. For example, IgG-mediated neutralisation of equine arteritis virus (EAV) (Balasuriya and MacLachlan, 2004) and acute phase EHV-1 (Snyder et al., 1981) is complement dependent, and opsonisation by specific IgG is necessary for efficient phagocytosis of Rhodococcus equi, Escherichia coli and Actinobacillus equuli by equine neutrophils or alveolar macrophages (Hietala and Ardans, 1987; Grondahl et al., 2001; Cauchard et al., 2004). Drawing all this information together, it is clear that the enhanced effector function capabilities of equine IgG1, IgG3, IgG4 and IgG7 equip these subclasses for key protective roles. International meetings have highlighted the need for further research into the functional roles of equine IgG subclasses and for the development of tools to study equine Igs (Lunn et al., 1998). The reqIgGs described here are a reliable source of pure and homogeneous equine IgG subclasses and will serve as useful reference proteins for the production and screening of equine specific reagents. In particular, the generation of mAb able to discriminate between the seven subclasses is an ongoing project. Moreover, they will provide a valuable resource for future research, in particular in delineation of the function of individual IgG subclasses in antibody-mediated immunity of the horse.

Cogn_Process-4-1-2254470

Coherence and recurrency: maintenance, control and integration in working memory

Working memory (WM), including a ‘central executive’, is used to guide behavior by internal goals or intentions. We suggest that WM is best described as a set of three interdependent functions which are implemented in the prefrontal cortex (PFC). These functions are maintenance, control of attention and integration. A model for the maintenance function is presented, and we will argue that this model can be extended to incorporate the other functions as well. Maintenance is the capacity to briefly maintain information in the absence of corresponding input, and even in the face of distracting information. We will argue that maintenance is based on recurrent loops between PFC and posterior parts of the brain, and probably within PFC as well. In these loops information can be held temporarily in an active form. We show that a model based on these structural ideas is capable of maintaining a limited number of neural patterns. Not the size, but the coherence of patterns (i.e., a chunking principle based on synchronous firing of interconnected cell assemblies) determines the maintenance capacity. A mechanism that optimizes coherent pattern segregation, also poses a limit to the number of assemblies (about four) that can concurrently reverberate. Top-down attentional control (in perception, action and memory retrieval) can be modelled by the modulation and re-entry of top-down information to posterior parts of the brain. Hierarchically organized modules in PFC create the possibility for information integration. We argue that large-scale multimodal integration of information creates an ‘episodic buffer’, and may even suffice for implementing a central executive. Introduction When we return home from work we sometimes find ourselves deeply engaged in thinking about an unsolved problem. Yet, at the same time we manage to find our way home and avoid accidents. This indicates that apparently we are able to automatically control our behavior without needing conscious attention. If suddenly an obstacle appears, however, we are also able to switch attention in a split second. Given the right cues, we may also remember that in the morning we planned to do some shopping on the way home. That too would put solving the problem to a halt and bring us to figure out the best route to the shop. This shows that control may be taken over by retrieving a plan, and that in order to carry it out we need access to a large database, including a route planner. If later on we resume thinking about the problem, we may figure out a solution by combining known elements in a novel configuration, and then start thinking how to use it in the future. The description given above is an example of the coordinated and goal-directed thoughts and actions that are generated in our brains all the time. Clearly, such coordinated behavior requires a versatile system of cognitive control for which as yet there does not seem to exist an adequate concept. Some of the contours of such a control system, however, are discernable in the example given above, like the ability to maintain representations, to direct attention and to combine previously unrelated information. An early and influential framework for the study of control and coordination of cognitive behavior is the multi-component model of working memory (Baddeley and Hitch 1974). Working memory (WM) refers to a limited capacity system for temporary holding and manipulating information, that is required for performing a wide range of cognitive tasks such as comprehension, learning and reasoning. Originally the model distinguished three-components. A general control system, or ‘central executive’, and two subsidiary slave systems, a phonological loop and a visuo-spatial sketch pad, each capable of maintaining a limited amount of information. More recently (Baddeley 2003), an episodic buffer was added to connect working memory with long-term memory and to allow binding together information from different sources into integrated episodes. The central executive is the most important but least understood component of WM, and it looks conspicuously like a homunculus. Baddeley is well aware of this criticism, but he defends the central executive concept by pointing out that it defines a problem area for which the processes have to be specified. An important conceptual model was suggested by Norman and Shallice (1986). They proposed a ‘data base system’ for automatic habitual action routines, containing well-learned stimulus-response associations and cognitive and behavioral skills, and a ‘supervisory attentional system’ capable of controlling behavior by selectively biasing and reconfiguring the available skills and schemas. Automatic control develops gradually through practice as learning processes create associative pathways between perception and action. The conscious form of control is applied when we are confronted by novel and unexpected stimuli, and when we have the intention to attain specific goals. In this case, automatic actions to stimulus patterns in the environment have to be suppressed and replaced by novel task- and goal-directed actions. The model of Norman and Shallice aimed to explain deficits of executive control that are often observed in patients with damage in prefrontal cortical areas. It is now generally accepted that the prefrontal cortex (PFC) is of crucial importance when behavior must be guided and controlled by internal states and intentions, when automatic responses have to be suppressed, and when tasks require the establishment of new or rapidly changing mappings between perception and action (e.g., Goldman-Rakic 1996; Miller and Cohen 2001; Smith and Jonides 1999; Wood and Grafman 2003). Anatomically, the PFC is well positioned to coordinate processing in the rest of the brain. It consists of a number of strongly interconnected areas that collectively have reciprocal connections with virtually all other neocortical and subcortical structures. It is also an area that shows late development, both phylogenetically and ontogenetically, reaching maturity only in adolescence (Fuster 2001). Based on these anatomical considerations, we have suggested (Phaf and Wolters 1997) that working memory or, more general, executive functions, may have emerged as a consequence of an evolutionary development of the PFC. This development may have created the possibility for neural processes that could run (partly) independent of present input and output, but that also could modulate, and thus control, perception-action relations in the rest of the cortex. In this view, the role of PFC in controlling behavior is modulatory rather than transmissive, which is similar to the role assigned to PFC in several other models (e.g., Norman and Shallice 1986; Miller and Cohen 2001; O’Reilly et al. 1999). Whereas simple adaptive behavior rests on a cycle of perception, action, and perception-of-action results, the added PFC would allow an internalization of this loop, freeing the organism of the restriction of being aware of, or acting upon, physically present objects or situations only. In fact, it would implement a medium for creating a virtual world, i.e., manipulating internal representations that are independent of the present environment (for a similar idea, describing thinking as simulated behavior and perception, see Hesslow 2002). More specifically, we suggested that the development of PFC created the possibility to maintain physically absent information in an active state by recurrent connections (loops) between PFC and the rest of the cortex. The recurrent activity in these loops may affect subsequent perceptual and motor processing, i.e., it can redirect attention and control actions by activating or inhibiting particular motor programs. By assuming interactions and integration between loops, more complex forms of representation and control may develop. For instance, recurrent connections with memory systems would allow access to all stored information, and mechanisms for combining information in the loops would allow the formation and updating of future goal states, and ways to achieve them. A taxonomy of working memory functions Although there is no generally accepted taxonomy of executive functions, we may speculate which functions would be required. It can be argued, for instance, that to be able to exert top-down control over cognitive processes, patterns of activity representing task relevant information must be actively maintained in the PFC. Moreover, PFC must be able to generate biasing signals to guide the flow of information by selectively inhibiting or activating particular representations and pathways in other parts of the brain. This interaction between PFC and posterior parts of the brain would also be needed to retrieve any relevant information stored in memory. Finally, PFC must be able to integrate information from different sources to implement goal directed behavior over time. We suggest, therefore, that three main executive functions have to be distinguished in PFC. The first function is maintenance, i.e., holding a limited amount of currently needed information, i.e., all task-relevant information supplied by preceding events, in an active form. The second function is attentional control, i.e., top-down controlled selective activation of task-relevant stimulus representations and responses. Only selective activation is required, because in a competitive system this automatically induces selective inhibition of task-irrelevant stimuli and responses. The third function is integration. This function consists of the ability to flexibly combine and reorganize information from different sources in the service of controlling task execution. This also includes control over search in LTM, monitoring and evaluating results of actual or imagined actions, and sequencing operations that are needed for planning, decision making and problem solving. In our view, the central executive operations in PFC are strongly interdependent and continuously interacting with processing in other cortical areas (see also Duncan 2001, for emphasis on interdependence of control, working memory and attention in PFC). A somewhat similar view of PFC functions was suggested by Smith and Jonides (1999). They proposed a distinction between short-term storage and two executive processes, selective attention and task management. In this paper we will first discuss the neuroanatomical correlates of these hypothesized interdependent working memory functions. Then we will present a neural model for binding and segregation in working memory to simulate the maintenance function. This model is based on recent ideas about synchronization of activation patterns in networks and it is able to explain capacity limitations of maintenance in WM. Next, extensions of this model will be discussed to realize the attentional control mechanism. It is suggested that modulatory effects within PFC may selectively enhance or suppress representations in posterior cortex by recurrent connections. Finally, we will present some ideas on how to realize integration and manipulation functions, given that large-scale multi-modal integration is necessary to explain coherence and coordination of behavior. Neuroanatomical correlates of working memory in PFC Generally, the brain shows a remarkable specialization of anatomically distinctive areas performing specific cognitive functions. Also the PFC comprises a number of cyto-architectural distinctive areas or modules that differ in patterns of connectivity with other brain regions (e.g., Miller and Cohen 2001; Wood and Grafman 2003). Much effort has been invested in finding functional specializations within the PFC, i.e., the extent that different areas or modules of PFC are selectively involved in encoding particular types of information (e.g., verbal versus visual), or in implementing various processing functions (e.g., maintenance versus attentional control). Although this debate is still ongoing, there seems to be a growing consensus that broadly speaking, three major areas in PFC seem to be distinguishable in terms of processing specific types of information (Fuster 2001; Wood and Grafman 2003), probably with additional sub-specializations. First, ventromedial and orbitofrontal regions (BA 11/12/47) seem specifically involved in representing and processing reward and affective information, and thus participate in emotionally and motivationally driven behavior (e.g., Fuster 2001; Rolls 2000). These regions have major reciprocal connections with the temporal lobe and the amygdale complex. Second, medial regions of the PFC (in particular the anterior cingulate cortex, BA 24/32) are suggested to be involved in error monitoring, and in detecting conflicts between competing stimuli and responses (e.g., Ridderinkhof et al. 2004). Third, lateral and anterior PFC areas (BA 9/10/44/45/46; see Petrides 2005, for a review of the architectonic organization of this area) are supposedly involved in all executive functions that are needed for organized goal-directed behavior, such as the selection of goal-relevant information, the manipulation and maintenance of information, and monitoring multiple-events (e.g., Miller and Cohen 2001; Petrides 2000). This area has strong reciprocal connections with parietal and temporal association areas, the hippocampal formation and with all other PFC regions. Damage to lateral PFC typically impairs the ability to formulate and carry out plans and sequences of actions (Fuster 2001) and to control long-term memory encoding and retrieval (Blumenfeldt and Ranganath 2006; Tomita et al. 1999). There is also increasing evidence, however, that localization of informational content is not precise, but rather a matter of degree. A review of imaging results showed that quite different cognitive demands (i.e., response conflict, task novelty, WM delay and load, and perceptual difficulty) induced very similar patterns of prefrontal recruitment, mainly involving lateral and medial PFC areas (Duncan and Owen 2000). This finding is corroborated by single cell studies that have shown a substantial adaptability of function at the level of individual neurons. Many lateral PFC neurons show highly specific activity patterns depending upon the current task and task conditions (e.g., Asaad et al. 1998, 2000; White and Wise 1999; Quintana and Fuster 1999; Wallis et al. 2001). These results led Duncan (2001) to postulate the ‘adaptive coding’ model. According to this model, neurons in PFC adjust their function to match the requirements of particular tasks that are carried out. Apparently any given cell can be driven by many different kinds of input, both from posterior parts of the brain and from other PFC areas. The collective responsiveness pattern of all PFC cells thus reflects the particular task relevant conditions. Duncan also noted, however, that the potential of cells to process different types of information, does not exclude regional specializations. This would reconcile results indicating regional specialization, with results from single cell studies showing large scale integration and adaptation. For example, a statistical rather than an absolute specialization seems to apply for left-hemisphere PFC recruitment for verbal material and right-hemisphere PFC recruitment for non-verbal stimuli (see, e.g., Passingham et al. 2002). A slightly different perspective on functional specialization of PFC areas is to assume that some areas are relatively specialized, whereas others serve more general integrative purposes and come into play in almost any task. Gruber and von Cramon (2003), for instance, found both modality specific and amodal areas in PFC in verbal and visuo-spatial working memory tasks. Other findings suggest that many cells in ventrolateral areas of PFC are specifically sensitive to maintaining representations of single stimuli (e.g., objects or locations), whereas cells sensitive to maintaining complex integrated stimuli and elaborative rehearsal are found more often in dorsolateral areas (Prabhakaran et al. 2000; Owen 2000; Wagner et al. 2001). Wallis and Miller (2003) presented evidence suggesting that orbitofrontal cortex, which is specifically involved in reward and affect processing, passes on this information to dorsolateral areas of PFC where it is combined with other information to control behavior. Similarly, Ridderinkhof et al. (2004) concluded from a review that detection of response conflicts and response errors elicits overlapping foci of activation in medial PFC areas, and that this activation serves as a signal that engages regulatory processes in lateral PFC areas. Learning and working memory Our view of working memory is that its informational content consists of the active part of long-term memory representations that is available at any point in time for controlled processing (e.g., Cowan 1999; O’Reilly et al. 1999). These representations presumably are stored in the temporal, parietal and occipital cortex, so the functions of working memory (maintenance, selection and integration) that apply to these representations are realized by an interaction between PFC and posterior cortical areas. In this view, PFC is not itself involved in storing long-term memory information. Its representational role would be limited to temporarily maintaining and integrating information stored elsewhere. This view seems to be endorsed by the fact that the neural circuits and pathways assumed to be involved in the two main learning processes (Eichenbaum and Cohen 2001), and the corresponding procedural and declarative memory systems (Squire and Zola-Morgan 1991), do not include the PFC as a representational medium. The suggestion that PFC implements WM functions, but is not itself involved in associative learning, would fit the requirement that it is maximally flexible in the service of generating novel combinations of representations that are stored somewhere else. There are no indications, however, that the neural architecture and processes of the PFC differ fundamentally from those in posterior cortex which intrinsically generate a learning capacity. Therefore, some learning (and representational) capacity in PFC cannot be ruled out a priori. Some evidence suggests that the information used in working memory tasks does not have to be stored in PFC. First, prefrontal patients generally perform quite well in tasks requiring the retrieval of previously learned procedural and declarative knowledge. However, they are typically impaired in episodic memory tasks requiring for instance elaboration of stimuli, ordered recall and source monitoring. So the problem with prefrontal damage does not seem an inability to store information in memory, but an inability to control encoding and retrieval processes (e.g., Shimamura 1995). Second, several recent studies have linked activity in PFC areas with promoting effective LTM formation (e.g., Blumenfeld and Ranganath 2006; Buckner 2003). They suggest, however, that the information itself is not stored in PFC. Instead, the interaction between PFC and posterior cortex controls what is selectively activated, maintained and elaborated, and what is subsequently stored as a novel representation via the medial temporal lobe and the hippocampus (see, e.g., Ranganath et al. 2005). Third, an increasing number of studies showed that the activation patterns of many single neurons or neuron ensembles in PFC quickly and flexibly adapt to represent completely arbitrary rules (Asaad et al. 2000; Wallis et al. 2001; White and Wise 1999), or to arbitrary action sequences (Averbeck et al. 2006). Although in all of these studies a number of learning trials precedes the development of task specific activation patterns, the speed and flexibility of the adaptation process seems more suggestive of an adaptive coding than of an associative learning process. Other authors have suggested that learning in PFC does occur. O’Reilly et al. (1999), for example, suggested that the information that is maintained in WM consists of activation patterns in PFC that are sustained by strong mutual excitation of the neurons involved. To explain the flexibility in combining these representations without losing their specificity, they assumed that PFC representations had to be relatively isolated from each other. Therefore, learning in PFC was supposed to be slow, taking place over many years, and eventually producing a rich palette of independent PFC representations that enable flexible problem-solving skills. Such representations might have a hierarchical structure with an increasing level of abstraction along a posterior-anterior PFC axis, probably with the most posterior PFC representations being closely connected to detailed memory representations in posterior cortex. A worked out version of this model by Rougier et al. (2005) showed that extensive training of their model with various tasks resulted in the development of abstract rule-like PFC representations that supported flexible generalization in novel tasks. The PFC representations developed slowly, but once learned adaptive behavior was mediated by a search for a task appropriate pattern of activity (cf. the adaptive coding mechanism of Duncan 2001), rather than the need to update connection weights. Although the simple tasks simulated in this model may also apply to more realistic and complex rules, the authors point out that it still leaves unexplained how PFC representations can be dynamically recombined and can interact with other systems (such as episodic memory, language and affect). It remains to be seen whether the supposed PFC learning mechanism is viable. Over brief periods of time, learning probably does not play a major role, but a role of learning over extensive periods cannot be ruled out. Evidence for three interdependent working memory functions in PFC Maintenance Neurophysiological studies have established persistent activity during delays (i.e., after stimulus offset) as the main candidate for a neural substrate of the maintenance function of WM. Single cells showing persistent firing during a delayed-matching task were first discovered in PFC, and later in other neocortical areas like inferotemporal (IT) cortex (Fuster and Alexander 1971; Goldman-Rakic 1996; Nakamura and Kubota 1995). Different areas, especially in posterior parts of PFC, may be involved in maintaining modality specific information. Smith and Jonides (1999), for instance, suggested a relatively specialized role of left and right PFC hemispheres in maintaining verbal and visual information, respectively. Moreover, they proposed that maintenance of spatial information may involve more dorsally located regions than maintenance of object information (see also Baddeley 2003). In principle, there are two ways to maintain activity in neurons during delays. One is to assume that input activation creates an activation pattern in a network of cells that is maintained within the network by auto-association (Amit 1995; Deco and Rolls 2003), and possibly augmented by a dopamine gating mechanism (O’Reilly et al. 1999). The other possibility is that the activation pattern oscillates in a recurrent loop between different networks of cells. There is much evidence that supports an important role of recurrent loops in maintaining information during working memory tasks. Strongest evidence is that during delays continued activity is not only found in PFC, but also in modality specific areas in IT cortex when coding for objects (Fuster et al. 1985; Miller et al. 1993; Tomita et al. 1999; Ungerleider et al. 1998), and in parietal cortex when coding for locations (Curtis and d’Esposito 2003; Rowe et al. 2000; Sakai et al. 2002). Moreover, localized cooling of either PFC or IT cortex interferes with activity in the other area and causes behavioral deficits in a working memory task (see Fuster 2001). From a review of the literature, Ranganath (2006) concluded that information in visual WM tasks is maintained through persistent activity in visual cortical areas (e.g., inferotemporal and parietal areas) that is promoted by top-down input from PFC. So according to Ranganath, maintenance would be an interaction between PFC and modality or object specific posterior areas. An additional loop involving the medial temporal lobe (i.e., the hippocampus and related areas) may be required to maintain complex novel stimuli, and to quickly create new long-term representations (e.g., Ranganath et al. 2005). Ranganath (2006) is somewhat unclear about the nature of the feedback from PFC in maintenance. As a further specification we would suggest a fixed connection scheme (probably shaped by long-term learning) between posterior brain areas, containing long-term memory or newly created representations of stimuli, and adjacent areas in PFC. These pathways convey to PFC what information is being maintained. This may also explain the finding that during maintenance, activation in posterior areas is more vulnerable to distraction than activation in PFC (Miller et al. 1993; O’Connor et al. 2002). Maintenance may resist such distraction because PFC activation can reinstate activation of corresponding lower level representations. Such reinstatement probably also underlies memory retrieval and mental imagery. It has been shown, for example, that in tasks requiring participants to imagine faces or buildings, category specific regions in the IT cortex became activated (e.g., Ishai et al. 2002; O’Craven and Kanwisher 2000). As was argued by O’Reilly et al. (1999), there are two seemingly incompatible core requirements of working memory. The first is to robustly maintain goals and task instructions over time even when faced with strong interference. The second is to rapidly and flexibly update goals and tasks if circumstances change. Their solution to combine robust maintenance with flexible updating has been to assume a dopamine gating mechanism. Under tonic dopamine levels a strong mutual excitation within a cell assembly would lead to sustained activity and resistance to interference by irrelevant input. Phasic shifts of dopamine levels induced by significant events (e.g., success or failure in a task or novel stimuli predicting reward) would enhance the strength of afferent input and cause updating of the activation state, which subsequently may lead to representing novel strategies, rules, goals or task states. Although a dopamine-gating mechanism is an interesting possibility, we believe that re-entry of activation through recurrent neuronal circuits is the main mechanism for the maintenance function of WM. It is likely that there are many of such recurrent loops for different types of information. These loops link the perceptual, memory and motor representational areas in posterior cortex to PFC. They feed information into PFC and, in turn, are activated by the recurrent activation from PFC. Modulating activity in these loops by other PFC sources would modulate the top-down recurrency. We also suggest and that these loops form a hierarchy, at the lowest level maintaining simple stimuli or features and at higher levels maintaining increasingly complex stimulus relations and rules. We will show how this idea can be implemented and explore the feasibility of such an implementation. We agree, however, that especially for the highest and most global or integrative levels of maintaining information, a dopamine gating mechanism for auto-associative maintenance cannot be ruled out. Attentional control Attentional control by executive functions in PFC requires top-down effects on local information processing in posterior brain areas (e.g., Desimone and Duncan 1995; Miller and Cohen 2001; Shimamura 2000). Evidence for such a role comes, for instance, from findings showing that an important characteristic of behavioral deficits following damage to the PFC is extreme distractability (i.e., an inability to suppress interfering information) and disinhibition (i.e., an inability to suppress inappropriate responses). A mechanism for top-down attentional control is suggested by the ‘biased competition model’ (Chelazzi et al. 1993; Desimone and Duncan 1995; Downing 2000; Reynolds et al. 2000; Gazzaley et al. 2005). According to such a model, top-down control activates corresponding representations that are then in a better position to compete with irrelevant information for perceptual awareness and control of motor behavior. Such top-down attentional modulation of neural responses, i.e., relative enhancement of neuronal responses to task-relevant stimuli and relative suppression of neuronal responses to task-irrelevant stimuli, has been shown throughout the visual system as early as the primary visual cortex (e.g., Desimone and Duncan 1995; Reynolds et al. 1999), and the exact areas that are biased depend on the task that is performed (e.g., Kastner et al. 1999). Top-down attentional bias not only pre-activates specific target representations but also spreads activation to related representations. Distractors with visual similarities to targets also attract attention (are positively biased), as well as distractors with semantic or associative links to the target (Moores et al. 2003). Task specific activity in PFC also generates top-down signals involved in the selection of actions, and in long-tem memory retrieval (Hasegawa et al. 1998) and storage (Brewer et al. 1998; Kyd and Bilkey 2003; Blumenfeld and Ranganath 2006; Wagner et al. 1998). This perspective thus suggests a single underlying mechanism of cognitive control by the PFC, namely a top-down biasing effect on processing in specialized sensory, motor and memory systems responsible for task performance (see, e.g., Miller and Cohen 2001). We will explore the feasibility of a mechanism for attentional control based on modulating feedback signals from PFC in the recurrent circuits that are used for maintenance. We suggest that attentional control is mediated by biasing the recurrent loops between PFC and posterior cortex. In this view, the biasing signals stemming from higher integrative areas in PFC would modulate the recurrent feedback to posterior cortical areas (see also Deco and Rolls 2003). Integration The importance of a large-scale integrative system for the coordination of behavior has been widely acknowledged. Such a system would provide the combination and integration of all information necessary for controlling task-relevant behavior, i.e., it would comprise all task-relevant information supplied by the present context and goals, and by knowledge retrieved from memory. Baddeley (2000), for instance, suggested an episodic buffer as an additional subsystem in his WM model which is capable of integrating and maintaining information from different sources. In a similar vein, Miller and Cohen (2001) argued for a cognitive control system in the PFC that represents goals and the means to achieve them. They suggest that this system integrates converging inputs from many sources. It exerts control by divergent feedback signals to sensory, memory and motor areas in posterior cortex which mediate directed attention, response selection, and guide retrieval from LTM. A high level integrative control system was also proposed by Koechlin et al. (2003). They suggested a cascade model, consisting of three nested levels of cognitive control. At the top of this cascade, located in rostral (anterior) parts of the PFC, they assume an episodic control system involved in selecting task-set representations ‘according to events that previously occurred or to ongoing internal goals’ (p. 1181). The view that progressively ‘higher’ neural areas support functions that are increasingly more integrative also has been endorsed also by Fuster (2001). He argued that a cascaded control model, assuming several nested levels of control, with a highly integrative system at the top, may explain neuropsychological results showing that damage to the top system only affects tasks that require a high degree of information integration (e.g., planning and problem solving), but does not interfere with tasks that can be executed at a lower level of control. Potentially the highest level of integration is attained in the anterior PFC (aPFC), in humans corresponding to BA10. In a recent discussion, Ramani and Owen (2004) pointed out this area as a likely candidate for being the apex of a hierarchical system of PFC modules, because it is predominantly (or even exclusively) reciprocally connected to other supramodal PFC areas. Therefore, it is in the position to integrate everything the brain is capable of representing. Moreover, the aPFC has anatomical characteristics (a high dendritic spine density combined with a low density of cell bodies) that make it likely to be involved in integration. The aPFC has been suggested to be specialized for processing internal mental states and introspective evaluation (Christoff and Gabrieli 2000), monitoring successful retrieval (Ranganath and Paller 2000), management and monitoring of goals and sub-goals (Koechlin et al. 2000; Braver and Bongliatti 2002), integration of information over time (Braver et al. 2001; Koechlin et al. 2003; Sakai et al. 2002), and manipulation of relational knowledge (Kroger et al. 2002). A hierarchical model of integration based on the convergence of input from lower level specialized PFC modules on higher integrative levels, is consistent with claims that a distinction can be made between PFC areas involved in simple maintenance, and others involved in maintenance of complex information and executive processes (D’Esposito et al. 1999; Sakai et al. 2002). For example, several findings suggest a distinction between ventrolateral (vlPFC) and dorsolateral (dlPFC) areas in terms of the level of abstractness of information processed (e.g., Koechlin et al. 2003; O’Reilly et al. 2002). Tasks requiring processing of single words or concepts mainly seem to involve vlPFC (e.g., Wagner et al. 2001), whereas in more complex tasks, like sentence processing, enhanced involvement of dlPFC has been found (e.g., Hashimoto and Sakai 2002; Kerns et al. 2004b). A relatively high integrative role for the dlPFC (as compared to vlPFC) is also suggested by findings showing this area to be involved in tasks that go beyond simple maintenance, such as elaborative rehearsal (Blumenfeldt and Ranganath 2006), processing relations between stimuli (Kroger et al. 2002), and solving conflicts that are signalled by medial areas of PFC (Egner and Hirsch 2005; Kerns et al. 2004a). We endorse the idea that the PFC is hierarchically organized, with subordinate modules being (relatively) specialized in processing simple aspects of tasks, and super-ordinate modules, located in dlPFC and frontopolar cortex, being specialized in large scale integration (i.e., binding) of inputs from different sources. At the apex of the supposed hierarchy, the aPFC might integrate all goal related information over space and time (see Fig. 1). Here, emotional and motivational evaluations (from orbitofrontal areas), results of memory retrieval, language and rule-like representations (from dlPFC), and anticipated action effects, response conflicts and bodily states (from medial PFC areas) converge. So the aPFC would be crucial for integrating multiple forms of information in the pursuit of general goals (e.g., Ramnani and Owen 2004), or when information from temporally dispersed events has to be integrated (e.g., Koechlin et al. 2003, 2006). Conversely, the ensuing activation patterns in aPFC set the stage for task and goal directed behavior, with control being exerted by top-down modulation of subordinate modules. In this way a hierarchy of top-down coordinated control is implemented by entraining successively lower order areas, eventually biasing perceptual processing, memory search, and action selection. Fig. 1A schematic view of the hierarchical structure of PFC. Many details regarding sensory input and motor output processing in subcortical centres and cerebellum are not shown. Double arrowheads indicate recurrent connections. Connections and modules shown are not suggested to be anatomically fully complete and accurate. aPFC anterior PFC; dlPFC dorsolateral PFC; vlPFC ventrolateral PFC; oPFC orbital and ventromedial PFC; mPFC medial PFC (anterior cingulate cortex); pm cortex premotor cortex A neurocomputational model of maintenance, control and integration Elsewhere (Raffone and Wolters 2001), we have presented a model for the temporary holding in (visual) working memory of a limited number of neural patterns, simulating either single features or integrated objects. The model implemented a cortical mechanism of maintenance in a network of model neurons with biologically plausible parameters. Although the model implemented a visual working memory system, the principles may be applicable to any form of information or type of working memory. In the model WM was assumed to be based on recurrent connections between IT cortex containing representations of objects or features, and corresponding neurons in PFC. The IT representations were modelled as strongly associated neural assemblies that generate synchronized firing patterns when activated by external input. The simultaneous activation of independent assemblies in IT causes competition via inhibitory interneurons. Due to the neuron characteristics, this leads to desynchronization among the activation patterns of competing assemblies resulting in a sustained phase-locked activation of multiple assemblies over time. Maintenance in cortical circuits of visual working memory was shown to be possible in terms of oscillatory reverberations between PFC and IT modules. Firing rate oscillations induced during stimulus presentation were maintained after stimulus offset by active feedback from prefrontal areas. Neurophysiological plausible model parameters enforced a limitation of about three to four independent assemblies that could be maintained in this way. This number closely coincides with recent estimates of the maintenance capacity of WM (e.g., Cowan 2001). The same mechanism that optimizes coherent pattern segregation, also poses a limit to the number of assemblies (about four) that can concurrently reverberate. The model thus indicated that selective synchronization and desynchronization of feedback-based oscillatory reverberations creates a suitable medium for a visual working memory. Simulations showed that the model was able to explain both the existence of severe limits in the number of assemblies (stimuli) that can be held (e.g., Luck and Vogel 1997; Luck and Beach 1998), and the absence of a limit on the size of assemblies, i.e., representing either simple stimuli or complex chunks (e.g., Ericsson and Delaney 1999). We introduced the concept of ‘chunking fields’ to account for the creation of more complex neural assemblies (e.g., higher order information units or chunks) through previous Hebbian learning (e.g., Hummel and Biederman 1992; Singer 1995). The model can potentially account for different degrees of within-object feature integration (Olson and Jiang 2002) in terms of graded synchrony between neurons coding for features of the same object. Here, we will explore an extension of the model of Raffone and Wolters (2001), simulating not only maintenance, but also a selective attention mechanism and a particular characteristic of an integration mechanism. The network architecture presented here to model these functions, is composed of three modules, which we assume to correspond to an IT module, a ventrolateral prefrontal module (vlPFC), and a dorsolateral prefrontal module (dlPFC), respectively (see Fig. 2). We assume that visual features are coded by individual assemblies of neurons in IT, which are “matched” to one assembly in vlPFC in a recurrent circuit. Moreover, we assume that different subsets of four vlPFC assemblies coding for given visual chunks, are bi-directionally connected to dlPFC assemblies (one for each set of four vlPFC assemblies). Fig. 2Scheme of the cortical network architecture. In the IT module, 20 neural assemblies code for 20 hypothetical visual features or separate representational elements. The figure shows the case with five four-feature chunks, with synchronizing connections between the assemblies coding the features of the same object (depicted as diamond-like configurations). The IT module also comprises an assembly of globally inhibitory neurons, which are implicitly modeled through inhibitory postsynaptic potentials (IPSPs). The vlPFC module is a set of 20 assemblies of neurons, with a coding structure “matching” the IT module structure. For simplicity, each IT assembly is recurrently connected with one vlPFC assembly, with a signaling delay of 15 ms in both directions. The vlPFC assemblies coding for a given chunk are reciprocally connected with a dlPFC assembly, which propagates synchronous firing before Hebbian learning takes place in the IT module In the IT module, strong connections are implemented within and weak connections between assemblies coding different ‘stimuli’. We also implemented a global inhibition (competition) mechanism between IT assemblies. This circuitry, in which a given assembly is inhibited by the firing of the other assemblies in the IT module, mediates an active desynchronization mechanism. Stimulus input to the IT module is given by stochastic spike trains from (not explicitly modeled) lower visual areas, during a limited onset-offset period. During stimulus presentation all stimulus specific IT neurons received a high frequent train of spikes as input, which was added to a continuous stochastic low frequency spike input to all IT neurons. The vlPFC module has a coding structure “matching” the structure of the IT module (see Fig. 2). Each IT assembly is recurrently connected with one vlPFC assembly, with a signaling delay of 15 ms in both directions. Although we implemented a monosynaptic feedback circuit, in real cortical networks this delayed feedback is likely to be mediated by multisynaptic circuitries of the ‘synfire’ type, with a stable transmission along multiple diverging/converging synaptic links (Abeles et al. 1993a, b; see also Villa and Fuster 1992). In vlPFC and dlPFC there are no inhibitory neurons and no inter-assembly connections. Although stimuli are coded by large assemblies of neurons in real cortical networks, in the present model individual stimuli are coded by single neurons. Our simulations will show that the functional processes we have investigated previously at the level of assemblies of neurons (Raffone and Wolters 2001) also hold at the level of single neurons, thus pointing out the robustness of the observed effects. The use of single neurons instead of assemblies of tens of neurons, is also motivated by the sake of running relatively fast simulations with a smaller time-step in numerical integration of neuronal equations (higher computational accuracy) than in our previous investigations. The model, of course, is very simplified with respect to the complexity of the real cortical networks. Feedback from the vlPFC module does no more than maintaining the oscillatory state of IT assemblies after the stimulus offset. More realistic network versions might include “closed” reverberatory circuits within prefrontal areas, making prefrontal assemblies independent from the IT assemblies in maintaining the delay activity. This would allow modality shifts by activating other prefrontal assemblies, which in turn would trigger reverberatory circuitries in lower cortical areas, in the absence of actual physical stimulation. Moreover, we used a simplified one-to-one matching between IT and vlPFC assemblies, whereas it seems more realistic to assume that prefrontal assemblies are relatively non-specific, thus being connected to multiple sets of neurons in posterior areas. Instead of the one-to-one matching of assemblies, dlPFC neurons might simply send back activation to all vlPFC neurons, and these in turn to all IT neurons from which they receive activation. However, the present simple architecture is sufficient for demonstrating the functional principles we have specified earlier. Simulations Simulating maintenance in working memory We first replicated our earlier results (Raffone and Wolters 2001) with stimulus features coded by individual neurons, rather than by sets of neurons, as well as with a higher temporal resolution of the simulations (computational accuracy) and a more realistic global inhibition mechanism. In this simulation the weights of the synchronizing inter-neuron connections in IT were set to zero, so all neurons code for different independent features. As shown in Fig. 3, about four elements are retained in the network after offset of the stimuli, which is in accordance with actual capacity limitations (e.g., Cowan 2001; Luck and Vogel 1997). The frequency of the reverberatory oscillations is approximately 30 Hz. Figure 3a and b show the retention of three out of four retained items, Fig. 3c and d the retention of four out of four, and Fig. 3e and f the retention of four out of eight items. Note the automatic phase-segregation due to the mutual inhibition desynchronizing effects (see Raffone and Wolters 2001, for a systematic investigation of retention capacity with different network parameters). The missed retention of some items (Figs. 3a, b, e, f) is due to the transient inhibition propagated by the firing of other competing neurons in IT. This inhibition “counteracts” the feedback input from the matched vlPFC neuron, thus interrupting the reverberatory cycle after stimulus offset. Such an interruption is more likely to occur with multiple reverberatory activities and stronger inhibition, as in that case the probability of the simultaneous arrival of strong inhibition and feedback signals increases. Fig. 3Limited retention capacity related to between-item segregation. Due to mutual inhibitory activity, the neuronal action potentials (spikes) become spaced in the oscillatory phase, thus allowing a markedly discriminative oscillatory reverberation and retention of the coded items. Often not all reverberations survived the stimulus offset. The panels show the fluctuating membrane potentials of neurons, as well as their spikes by short vertical bars above the membrane potentials. The panels a, c, and e shows the dynamic behavior of IT neurons, and the panels b, d and f the responses of respectively matched vlPFC neurons. In (a, b), three out of the four neurons remained active. In (c, d), all the reverberations remained active, with all the four items being retained. In (e, f), four out of eight neurons maintained their sustained oscillation. Note that the oscillatory reverberations tend to be optimally spaced in the phase-lag, depending on the number of reverberating neurons In this simulation, we did not replicate our previous demonstration (Raffone and Wolters 2001) that the capacity of the reverberatory circuit is independent of the size of the IT representations. As we showed there, larger cell assemblies of connected neurons with fast-signaling positive weights, quickly synchronize and behave as single units. We demonstrated that the maintenance capacity of our model is not a function of the size of a cell assembly, but of the presence and strength of associations between the units of an assembly. This probably mimics the fact that working memory capacity is not a function of the amount of information per se, but of the level of organization or chunking of the material to be maintained (Miller 1956). Simulating attentional control in working memory Given the severe limitation of the capacity of working memory, neural mechanisms are necessary to restrict access to it, depending on the behavioral relevance of the information. As shown by Asaad et al. (1998, 2000), prefrontal neurons of monkeys performing a ‘delayed-matching-to-sample’ task exhibited a higher firing rate when they coded for a rule-defined target-object. Such neurons were suggested to be involved in both the selection and maintenance of behaviorally relevant information. To model a selective attention mechanism by which selection and retention is controlled by working memory cortical circuits, a top-down input to IT neurons has to be supplied by a prefrontal source involved in supervisory control rather than in mere information maintenance. A similar assumption was made in a model for attentional control by Deco and Rolls (2003). Such a mechanism implements the top-down biasing of the competition between neural assemblies in posterior cortex as suggested by Desimone and Duncan (1995). In our model this supervisory control signal could be supplied through the dlPFC and vlPFC modules. We first considered an additive (voltage-independent) input to a subset of four (the bottom four in Fig. 4a) out of eight independent IT neurons all receiving external input. This additional input was modeled in terms of additional spikes with an excitatory postsynaptic potential (EPSP) amplitude equal to the external input signals and a spiking probability equal to 0.04. As can be seen in Fig. 4a, the four items receiving top-down input exhibited a higher (subthreshold) membrane potential before the stimulus onset. This top-down bias was crucial in selecting the items to be retained in terms of reverberatory oscillations. All four biased (and one unbiased) item were maintained after stimulus offset. Fig. 4Attentional modulation of reverberatory maintenance. Four out of the five maintained activities are of neurons (the first four neurons from below in the panels) the activities of which are selectively enhanced either by additive (voltage-independent) modulation (a), or by voltage-dependent (multiplicative) signals (b). Note that with additive modulation the membrane potentials of the four ‘focused’ neurons are higher before the stimulus onset (a), whereas the amplification mostly takes place after stimulus onset in the voltage-dependent modulation case (b) We also implemented a voltage-dependent modulatory input to IT neurons. A voltage-dependent gain effect simulates the possible role of NMDA receptors (i.e., receptors that activate a neuron only if a signal arrives at an already depolarized synapse). NMDA-based gain effects are suggested to play a crucial role in cognitive coordination and control (Phillips and Silverstein 2003, see also Raffone et al. 2003), and such gain effects are often modelled as a multiplication of input signals. This multiplicative effect was modeled in terms of the product of the voltage-independent external input (VI) and voltage-dependent top-down signals (VD) with amplitude equal to 0.5/5, and spiking probability equal to 0.25, according to the following equations Following suggestions of Tononi et al. (1992), the product of equation (2) was set equal to 0 when a membrane potential (ui) was less than a voltage threshold VT (set equal to 0.5). In that case, only a voltage-independent stimulus-related input determines the net input to the neuron, without any amplification. As can be inferred from Fig. 4b, the attentional bias (attentional input for four out of eight items) in the voltage dependent condition is expressed after the stimulus-input onset, in terms of an amplification effect. Also in this simulation all four biased (and one unbiased) items were maintained after stimulus offset. We believe that voltage-dependent input to high-level areas of the visual system may enable an effective top-down control mechanism, since it produces strong amplification effects, but only in the presence of relevant stimulus-input (see also Hirsch and Gilbert 1991; Tononi et al. 1992, about voltage-dependent signaling in the visual system). This multiplicative signaling effect would prevent spurious activation of representations in the visual system in the absence of any bottom-up sensory evidence. A series of follow-up simulations showed that the probability of item retention depends on the relative biasing top-down input for a given item. A strong attentional input to one or two items may ‘narrow’ visual working memory capacity to one or two highly focused elements, due to the higher firing rate (e.g. bursting) of the neurons coding for the focused items. Mechanisms for integration in working memory Any domain-specific account of functions or representations in the brain ultimately implies a binding problem, since the specialized neural representations need to be dynamically bound to enable the creation (and a conscious readout) of coherent and integrated representations of complex events. These binding processes may occur in terms of the selective synchronization of reverberating cell assemblies, as shown in our simulations. In this framework, segregated neural representations may be bound via reciprocal synchronizing connections that originate within the prefrontal cortex (for a discussion of this and other forms of binding, see Murre et al. 2006; see also Engel et al. 2001; Varela et al. 2001). In our model, dlPFC can play a crucial role in what we call “associative control”, i.e. in selecting meaningful or currently salient conjunctions of otherwise separated representational elements held in a temporarily active state within vlPFC-IT circuits. We have shown that reciprocal fast signaling between a dlPF neuron and its corresponding vlPFC neurons may “entrain” all these vlPFC neurons coding for different features or units to fire in a nearly synchronous manner. This synchronous firing is propagated to IT neurons, which in turn fire almost coincidently before the action of the mutual inhibitory signals within the IT module (Fig. 5a, b). We suggest that such a top-down controlled selective synchronization of independent representations is a likely candidate of the process underlying integration in PFC. This mechanism ensures the simultaneous and coherent activation of independent representations that may be widely dispersed over various brain areas. It may cause, for instance, the large-scale task-dependent synchronization in the gamma band, as observed by Rodriguez et al. (1999). Fig. 5Associative control on vlPFC neuronal activities by dlPFC neurons. The panels a, c, and e shows the dynamic behavior of IT neurons, and the panels b, d and f the responses of respectively matched vlPFC neurons. As shown in panels a, b, inter-neuron synchronization of the four vlPFC neurons is induced by a dlPFC neuron (activity not shown), and is then propagated to the matched IT neurons. As shown in panels c, d, the joint effect of dlPFC feedback and Hebbian associative signals in IT may induce a higher firing rate of neurons in IT and vlPFC. Panels e, f show that a stable synchronization of IT and vlPFC neurons is observed after Hebbian strengthening of IT chunking synapses, with feedback from the dlPFC neuron being switched-off The same mechanism may allow Hebbian learning to occur between meaningful or behaviorally-salient conjunctions of independent elements (see Miltner et al. 1999). In order to account for the retention of integrated units in visual working memory, in previous simulations we defined units by assuming already established chunking fields, i.e. pre-existing associative connections between IT neurons, coding different features of the same unit. A major problem, however, is how ‘chunking fields’ or integrated representations in long-term memory are formed, i.e. how associative Hebbian learning may take place in the simulated cortical circuits of (visual) working memory. This learning process should operate against the phase-segregation tendency due to mutual desynchronizing inhibition. In the present model, the dlPFC module may perform this temporary synchronization control processes. In a series of simulations, we used the entrainment of vlPFC and IT neurons by a top-down dlPFC signal and combined it with a timing-dependent Hebbian learning rule in IT (simulating timing-dependent plasticity of synapses, see e.g., Körding and König 2000; Song et al. 2000, see Fig. 5c, d). These simulations showed that the synchronous firing in IT, induced by a top-down recurrent signal, causes the weights between the IT neurons to increase, resulting in a ‘chunked’ representation of initially independent features. As synaptic weights increased according to the timing-dependent learning rule, the synchronization among related IT neurons becomes gradually less dependent on the feedback action of dlPFC neurons on related vlPFC neurons, ultimately leading to an “automatic” within-chunk integration within the IT module, in the absence of any control feedback from the dlPFC module (Fig. 5e, f). Control simulations with the same model and learning parameters, showed that a relatively high number of simultaneous presentations of initially independent elements, may slowly give rise to novel chunks with a variable degree of stability, even in the absence of feedback from dlPFC to vlPFC. This slow chunk-learning process may be related to an implicit learning process, rather than the more rapid process of controlled chunk formation by feedback from dlPFC. Discussion Miller and Cohen (2001) concluded that PFC is critical when top-down control of processing is needed for the guidance of behavior by internal states or intentions. In this paper we conjecture that such control is possible by assuming that the PFC is a brain system that is reciprocally connected to posterior parts of the brain and therefore capable of modulating direct perception-action relations. We argued that three interdependent functions are required for such control: maintenance of activation patterns even in the face of distraction, large scale integration of information from different sources, and top-down selective attention. We also showed that these functions can be simulated, at least in principle, with a biologically plausible model assuming recurrent connections and a hierarchical structure of PFC. Importantly, the functions were simulated using a single framework consisting of a hierarchy of recurrently connected modules. This framework proved capable of maintaining a limited number of mutually desynchronized patterns, of controlling selective attention by top-down modulatory signals, and of high level integration by top-down synchronization of the activity of independent patterns. Of course, the model suggested here is extremely simplified and it does not capture many of the complexities and intricate details of the real system. For instance, we did not try to incorporate modulatory effects of neurotransmitters and we have not discussed the many possible interactions of the PFC with subcortical systems, like the basal ganglia and amygdala, and the cerebellum. It may also be noted that we discussed, but did not try to incorporate, the possibly important role of a dopamine gating system for maintaining high level integrative activation states (O’Reilly et al. 1999, 2002). One of our aims to develop the model was to try to find out how the structural and functional characteristics of the PFC may explain the central executive component of working memory assumed by Baddeley (2003). In the latest version of this model an additional component, an episodic buffer, was proposed. This component was thought necessary to explain interactions between working memory and LTM, for example to explain how chunking may supplement the capacity for immediate serial recall. A continuous interaction between working memory and LTM is a basic assumption in our model, and we have shown that chunking is a natural consequence of this interaction. According to Baddeley (2003), the episodic buffer allows information from different systems to be integrated, and it may be regarded as the ‘storage component of the executive’. Moreover, Baddeley suggested that the buffer is a separate temporary store in which long-term memory information is downloaded in order to be manipulated and used for creating new representations. In our view, the structure of such an episodic buffer would actually be a hierarchy of PFC modules maintaining and integrating information at successively higher levels. Information is not copied and downloaded in PFC, but selected task-relevant perceptual and memory information in posterior cortical areas is kept in an active state by recurrent loops. We assume that at the highest integration levels, presumably located in anterior and dorsolateral areas of PFC, information can be integrated over space and time. At this level, general goals, and plans to achieve them, would be generated. We suggest that this may be the implementation of an episodic buffer and even of a central executive. The creation of short-term and long-term goals at high levels of integration, allows to control behavior by top-down regulation of activation in subordinate modules and eventually by modulating perception-action systems. Current goals and tasks would have to be maintained here in order to control ongoing behavior. More distant goals, however, may be stored in LTM to be retrieved later on. We do not imply that control is always hierarchical. Because the PFC is assumed to be a modulatory system, simple tasks may be executed automatically without PFC involvement. Only controlled processing requires modulation of automatic processes by PFC feedback. Also here, however, control does not need to be completely hierarchical. Depending on the type and complexity of a task, control may be executed by specialized modules at subordinate levels. Automatic processing and controlled processing under the guidance of subordinate levels in PFC, is a requirement to free higher integrative areas to engage in the kind of simulated actions and perceptions, and using anticipated outcomes, that are the contents of conscious thought and planning. So there are things we can do automatically, using previously established associations between perceptual events and effective responses, and there are things we can do in a controlled way, using the possibility of PFC to modulate and thus control perception and action. The possibility of PFC to work off-line, to use and integrate all present and past knowledge for creating virtual worlds and for making plans and carry them out if conditions are appropriate, has generated a tremendously flexible potential for control. The model we propose is vastly insufficient to simulate such cognitive feats that we engage in daily. We believe, however, that the principles for a controlling system we have suggested may be a first step towards a better understanding of what is until now a cognitive Terra Incognita.

Eur_Spine_J-2-2-1602195

Preadolescent presentation of a lumbar chordoma: results of vertebrectomy and fibula strut graft reconstruction at 8 years

Chordoma is a tumour of notochordal origin which usually involves the sacrum or skull base presenting in adulthood. Chordoma in a mobile spinal segment is infrequent and the authors report an extremely rare presentation of L3 chordoma in a child aged 7 years. Although a benign tumour, mobile segment chordoma is more locally aggressive, more likely to metastasise and has a poorer 5 year survival than sacral and clival lesions. Wide surgical excision and reconstruction is the treatment of choice in vertebral chordoma. This case was treated with staged vertebrectomy and fibular strut graft reconstruction and the results of clinical and radiological follow up at 8 years are presented. Case report A previously well 7-year old child presented with a 6 month history of right anterior thigh and knee pain. Examination revealed a marked list to the right, painful restriction of spine movements in all directions, an absent right knee jerk with reduced sensation in the right L3 distribution. There was no motor weakness or sphincter disturbance. Back pain restricted straight leg raise bilaterally to 30°. No history of weight loss, systemic symptoms and no past medical and family history of relevance. Blood tests were within normal limits. Inflammatory markers were not elevated. Plain radiographs revealed an osteolytic lesion in the posterior body of L3. MRI of the whole spine identified an isolated lesion at L3 with right pedicle involvement and tumour extension anterior to the dural sac at L4 (Fig. 1). Isotope bone scan showed no increased uptake. A staging CT of the chest showed no evidence of metastases. Fig. 1Pre-operative MRI scan of lumbar spine showing the lesion involving L3 vertebra Transpedicular biopsy at L3 was performed and histology revealed fragments of remodelling bone, haemorrhage and a myxoid tissue containing strips and ribbons of cohesive, vacuolated cells. These cells had the appearance of notochordal-type tissue. Immunohistochemistry was positive for Cytokeratin S-100. In view of the histological findings and atypical young age of presentation a radical 360° L3 vertebrectomy with pedicle screw stabilization and fusion was planned. At operation the L3 tumour had extended through the lamina and transverse process on the right compromising the right L3 nerve root. There was both cranial and caudal extradural extension of the tumour anterior to the dural sac with tenting from L2 to L4. A wide decompression was performed with resection of the posterior elements and partial posterior vertebrectomy at L3. To ensure that the tumour within the spinal canal was completely excised it was necessary to resect the inferior lamina of L2 and most of the lamina at L4 bilaterally. Hence it was felt appropriate to extend the instrumentation inferiorly to L5. Pedicle screw instrumentation was performed at L2 and L5 bilaterally. Free fat graft was placed over the dura and nerve roots. Bone graft was placed posterolaterally from L2 to L5 (Fig. 2). An epidural infusion was used postoperatively for pain relief. Initial post operative neurological assessment was normal. Fig. 2AP and lateral X-rays following first stage of the procedure The following day deterioration in her lower limb neurological status was noted with no active movement or sensation in her lower limbs. That this was not recognised immediately was in part attributed to her epidural and to some difficulty with neurological assessment in a young child with systemic analgesia after major surgery. She was returned to theatre where a large haematoma was found and drained. Although subsequent sensory recovery was good, her motor recovery was slow with grade 1 power in her lower limbs. Two weeks later, once progressive neurological recovery was established, she underwent second stage anterior vertebrectomy of L3 through a retroperitoneal approach. Unilateral L2, L3 and L4 segmental vessels were ligated to expose L3. A complete excision of the remaining L3 and posterior longitudinal ligament was performed in order to adequately decompress the dural sac anteriorly. Three fibular strut autografts from L2 to L4 were used to reconstruct the anterior column. Posterior iliac crest graft was used for further augmentation (Fig. 3). Fig. 3AP and lateral X-rays following second stage of the procedure showing the fibular strut grafts used Neurological recovery continued post-operatively and she had regained normal sensation and grade 5 motor power in both lower limbs by 6 months. Bladder sphincter dysfunction has persisted, requiring intermittent self catheterisation and anti-cholinergic medication. She was nocturnally continent. Bowel function remained normal throughout. Radiographic fusion was evident at 6 months. Metal work was removed at 12 months and intra-operative biopsy showed no tumour recurrence (Fig. 4). Fig. 4AP and lateral X-rays following removal of metal work showing consolidation of the strut grafts She remained under annual follow-up with MRI and plain radiographs. At 8 years she has no evidence of tumour recurrence. Her neurological status is unchanged. Imaging has shown excellent remodelling of the graft (Figs. 5, 6). She is skeletally matured with no lower limb malalignment and experienced no fibular donor graft site morbidity. Fig. 5Chronological X-rays showing graft incorporationFig. 6Eight year follow-up MRI Scan Discussion Chordoma is a primary bone tumour arising from notochord remnants. It is a rare tumour accounting for only 1–4% of primary bone tumours [1]. Chordoma is usually a slow growing tumour, often with local recurrence and it may metastasise late in its course [1]. Typically it presents in late middle age and may occur anywhere along the spinal column. Chordoma is found most commonly in the clivus, sacrum and infrequently in the mobile vertebrae [2]. The optimum treatment for vertebral chordoma is wide surgical excision. Radiotherapy may provide short term benefit when adequate excision is not possible and in cases of local recurrence [3]. Surgical treatment has become more aggressive in recent years evolving from intralesional debulking to en bloc resection, as documented by Boriani et al., over a 45-year period in their treatment of 21 cases of chordoma located in the spine above the sacrum [4]. Chordomas are considered to be low grade tumours and metastases are infrequent at presentation [1, 5]. The prognostic value of staging is therefore diminished, compared to the value of staging in other bone tumours, because all these tumours are stage IA or IB according to the Musculoskeletal Tumor Society Staging System [6]. Mobile segment chordomas have a 50% survival rate at 5 years and a 28% survival rate at 10 years [4, 7]. Sacral chordomas on the other hand are reported as having 86% 5 year survival [3]. Treatment outcome is significantly influenced by the size and site of chordoma [1, 4]. Chordomas found in the vertebral bodies appear to be more aggressive than those arising in the clivus or the sacrum [3]. Metastases have been reported in 80% of the vertebral body chordomas, as compared to a rate of 43% of all chordomas [3]. However, the survival rate appears to be affected more by local tumour progression than by metastases [1]. Conclusion This case is unusual given the young age at presentation. She is the youngest patient reported in world literature. Despite the poorer published results of mobile segment chordomas, radical surgical excision and reconstruction has resulted in an 8 year disease free survival. Fibular strut graft reconstruction proved an excellent reconstructive option in the growing skeleton with potential for remodelling. It is likely that the young age has contributed to the near complete neurological recovery.

Eur_J_Clin_Pharmacol-4-1-2254473

Prior outpatient antibiotic use as predictor for microbial aetiology of community-acquired pneumonia: hospital-based study

Objective The causative micro-organism in community-acquired pneumonia (CAP) is often difficult to predict. Different studies have examined chronic morbidity and clinical symptoms as predictors for microbial aetiology of pneumonia. The aim of our study was to assess whether prior outpatient antimicrobial treatment is predictive for determining the microbial aetiology of CAP. Introduction Community-acquired pneumonia (CAP) remains a major reason for hospital admission and a common cause of death in developed countries [1, 2]. The initial management of patients hospitalised with pneumonia consists mostly of empirical antimicrobial treatment [3, 4]. Determination of the appropriate antimicrobial treatment is essential as inadequate antimicrobial treatment, generally defined as microbial ineffective therapy against the causative pathogen, can negatively influence patient outcome [5]. Because it is not always possible to identify the causative pathogen, especially during the first days following hospitalisation, many studies have focussed on other parameters suggestive of the causative pathogen. The most frequently studied parameters for this purpose are patient characteristics (age and co-morbidities) and clinical signs [6, 7], although nonresponsiveness to prior outpatient antimicrobial treatment has also been suggested as a predictor of the aetiology of pneumonia. However, to date, this latter parameter has not been comprehensively quantified and documented. The aim of the study reported here was to assess whether prior outpatient antimicrobial treatment is a predictor of microbial aetiology in patients admitted to hospital for CAP. Methods The study was conducted in the St. Antonius Hospital, a 600-bed teaching hospital in Nieuwegein, The Netherlands. Patient population This was a prospective observational study of patients with confirmed pneumonia admitted between October 1, 2004 and August 1, 2006. Pneumonia was defined as a new or progressive infiltrate on a chest X-ray plus at least two of the following criteria: cough, sputum production, temperature >38°C or <35°C, ausculatory findings consistent with pneumonia, leucocytosis or leucopenia (>10 g/l, <4 g/l or >10% rods in leucocyte differentiation), C-reactive protein >3 times the upper reference value for normal. Patients who were immune compromised (systemic steroid use at admission (prednisone equivalent >20 mg/daily for more than 3 days), haematological malignancies and other immunosuppressive therapy) were excluded. The study was approved by the local Medical Ethics Committee, and informed consent was obtained from each patient. Microbial aetiology workup At least two blood cultures were performed, and sputum was taken for Gram-staining and culture and subsequently analysed by Taqman real-time PCR for Mycoplasma pneumoniae, Legionella pneumophila and Chlamydophyla psittaci [8]. Pharyngeal samples were taken for viral culture. Urine was sampled for antigen testing on Streptococcus pneumoniae and L.pneumophila (Binax NOW; Binax, Scarborough, ME) [9, 10]. In addition, serum samples taken on the day of admission and on day 10 were analysed in pairs for detection of a fourfold rise of antibodies to respiratory viruses, Coxiella burnetii, M. pneumoniae, and C. psittaci by complement fixation assay [11]. For each patient, the total workup was completed, and the microbiology department was blinded to data on outpatient antibacterial drug use. When both viruses and bacteria were identified in a patient, the search prevailed for a definite aetiology for the bacteria. Exposure to antimicrobial therapy Data on outpatient antimicrobial drug use were acquired through community pharmacy dispensing records that captured all drug exposures the year prior to hospital admission. A patient was considered to be exposed to an antimicrobial drug when a prescription was filled within 14 days prior to hospitalisation. The name, dosage and amount of antimicrobial drug dispensed were also ascertained. The prescribed antimicrobial drug was classified as appropriate or inappropriate in accordance with current Dutch guidelines on the initial treatment of patients with suspected pneumonia [12, 13]. Co-morbidity assessment In addition to outpatient antimicrobial drug use, co-morbidities and other relevant patient characteristics were identified to address factors related with the aetiology of CAP. Co-morbidities were defined based on the presence of conditions for which the patient was under active medical supervision or was receiving treatment at the time of hospital admission. The co-morbidities evaluated were pulmonary diseases (chronic obstructive pulmonary disease or treated asthma), congestive heart failure, diabetes (both type I and type II), history of stroke and end-stage renal disease (serum creatinine >150 μmol/l). The patients were also classified according to the Pneumonia Severity Index (PSI) developed by Fine et al. [14]. This index classifies patients into five categories of predicted mortality risk (with the fifth category being that of highest mortality risk). The outpatient use of oral corticosteroids and gastric acid-suppressing drugs was also ascertained. Statistical analysis The SPSS statistical package (ver. 12.0.1 for Windows; SPSS, Chicago, IL) was used for the statistical analyses. Continuous data were expressed as the mean ± SD or as the median (interquartile range) where appropriate. To study the association between prior outpatient antimicrobial treatment and the aetiology of pneumonia, we applied multivariate logistic regression analyses. The analyses were conducted for overall aetiology and relevant pathogens separately. All baseline characteristics were considered potential predictors. Potential predictors were included in the multivariate model when they were retained after backward stepwise elimination. Significance was set at a p value < 0.05. The model’s performance (goodness-of-fit and discriminative ability) was tested by performing the Hosmer and Lemeshow test and calculating the area under the receiver operator characteristic (ROC) curve. Results In total, 201 patients with pneumonia were included in the study. The mean age of the patients was 63 ±17 years, and 124 patients were male. Three patients (1%) were admitted from a nursing home. Pneumonia severity as well as co-morbid illnesses are summarised in Table 1. The overall median duration of the hospital stay was 10 days [7 –14], and 21 patients were admitted to the intensive care ward. During the hospital stay, ten patients died, all due to pneumonia. The overall 28-day mortality rate was 5%. Forty-seven patients (23%) had received antimicrobial treatment in the 14-day time-window prior to hospital admission. The antimicrobial drugs dispensed to these patients are summarised in Table 2. Table 1Demographics, co-morbidities and pneumonia severity index of 201 patients with community-acquired pneumonia (CAP)Characteristicn (%)Age (Years)  <6074 (37)  60–6939 (19)  70–7950 (25)  >8038 (19)Gender  Male124 (62)  Female77 (38)Co-morbidities  Pulmonary diseases71 (35)  Heart failure18 (9)  Diabetes35 (17)  History of stroke17 (9)  End-stage renal disease10 (5)  Nursing home resident3 (1)Co-medication  Oral corticosteroids58 (29)  Gastric acid suppressing drugs61 (30)Fine score at admissiona  I30 (15)  II34 (17)  III53 (26)  IV56 (28)  V28 (14)aFine et al. [14]Table 2Outpatient antibiotics utilization profile prior to hospitalisation for CAPType of antimicrobial drugNumber of users (%)aAppropriatebClavulanic acid18 (38)YesAmoxycillin12 (26)YesDoxycycline7 (15)YesClarithromycin5 (11)YesCo-trimoxazole4 (8)NoCiprofloxacin2 (4)NoNorfloxacin1 (2)NoAzithromycin1 (2)YesaTotal percentage exceeds 100% because some patients (n = 3) had two prescriptionsbBased on current Dutch guidelines NVALT and SWAB [12, 13] The majority of the patients (79%) had their prescription filled within 4 days prior to hospital admission, and 85% of the prescribed antimicrobial drugs complied with current Dutch guidelines [12, 13]. A microbial aetiology could be determined in 128 (64%) of the patients. Table 3 shows the results of different tests used to determine the aetiological diagnosis of CAP. Table 3Results of of different tests used to determine the aetiology of CAP Sputum cultureSputum PCRAntigen testingBlood cultureSerologyViral cultureNumber of samples1487818318213088Number of positive samples781436193814Percentage of positive samples531820102916Streptococcus pneumoniae33–3017––Gram-negative strain  Heamophilus influenzae19––0––  Other10––1––Atypical  Mycoplasma pneumoniae–7––8–  Legionella spp.156–7–  Other–2––1–Viral––––2214Other  S. aureus6––1––  Gram-positive other2––0–– Fewer causative pathogens were found in the population hospitalised following prior outpatient antimicrobial treatment than in patients without prior antimicrobial treatment [57 vs. 66%; crude odds ratio (OR) 0.71, 95% confidence interval (CI) 0.36–1.38]. In patients with prior beta-lactam treatment, aetiology in the group comprising atypical bacterial pathogens was more probable (eight of 29 cases, 28%; crude OR 4.66, 95% CI 1.73–12.56). The aetiology of S. pneumoniae was less prevalent in patients with prior beta-lactam antimicrobial treatment (four of 29 cases, 14%; crude OR 0.33, 95% CI 0.11–0.99). In multivariate analyses, these associations remained significant (OR 3.51, 95% CI 1.25–9.99 and OR 0.30, 95% CI 0.10–0.90, respectively) (Table 4). Additional significant predictors included in these models were heart failure, pulmonary comorbidity, pneumonia severity index and the use of gastric acid suppressing drugs. The goodness-of-fit of both multivariate models was good, with a p value of 0.899 (Hosmer and Lemeshow test) for the model predicting pneumococcal pneumonia and a p value of 0.995 for the model predicting pneumonia of an atypical aetiology. The corresponding areas under the ROC curve were 0.62 and 0.67, respectively. Table 4Odds ratios (OR) for aetiology and prior outpatient beta-lactam treatment in patients admitted to hospital for CAPAetiologyPrior outpatient beta-lactam treatment (n = no. of patients)OR (95% CI)Yes (%)No (%)Total no. of samples29 (100)172 (100)Univariate  Pneumococcal4 (14)56 (33)0.33 (0.10–0.99)  Atypical8 (28)13 (8)4.66 (1.72–12.56)  Viral2 (7)14 (8)0.84 (0.18–3.89)  Gram-negative strains2 (7)21 (12)0.53 (0.12–2.40)  Other1 (3)7 (4)0.84 (0.10–7.11)  Unidentified12 (41)61 (36)1.28 (0.58–2.87)Multivariate  Pneumococcal––0.30 (0.10–0.90)a  Atypical––3.51 (1.25–9.99)bOR, Odds ratio; CI, confidence IntervalaAdjusted for heart failure and use of gastric acid suppressing drugsbAdjusted for pulmonary diseases and pneumonia severity index In patients aged <60 years without co-morbidities, the aetiology of atypical bacterial pathogens was more prevalent (OR 4.64, 95% CI 1.72–12.56). Pulmonary co-morbidity was associated with the finding of S. pneumoniae and H. influenzae as causative pathogens (OR 1.87, 95% CI 1.00–3.47; OR 3.72, 95% CI 1.20–11.57, respectively). Patients who received macrolides prior to hospitalisation had an increased probability of viral pneumonia (two of five cases, 40%) (crude OR 8.67, 95% CI 1.34–56.23). Discussion The results of our study show that among patients admitted for CAP, a threefold decreased chance of having a S. pneumoniae infection and a threefold increased probability of having pneumonia of atypical aetiology were associated with the patient having received an initial beta-lactam treatment as an outpatient. These findings indicate that information on prior outpatient antimicrobial therapy has a predictive value in the diagnostic workup aimed at identifying the causative pathogen and planning the corresponding treatment in patients with pneumonia. The initial management of patients hospitalised with pneumonia has been under constant study in different settings during the past decades. The choice of antimicrobial treatment, time to first antimicrobial drug administration and route of administration have all appeared to be relevant factors associated with the outcome of pneumonia [15–17]. A knowledge of the predominant microbial patterns in CAP is therefore essential when choosing an essential empirical antimicrobial treatment. Previous studies have found that S. pneumoniae, H. influenzae, Influenza virus A and B, Legionella spp. and C. pneumoniae are the most frequent pathogens in CAP [13], which is in accordance with our results. Because S. pneumoniae is the most frequently appearing pathogen, the administration of beta-lactam antibiotics is the initial empirical antimicrobial treatment of choice in the treatment guidelines on CAP [3, 12, 13]. Beta-lactam antibiotics, however, do not cover Legionella spp., C. pneumoniae and M. pneumoniae, the so-called atypical pathogens. Therefore, patients with pneumonia of atypical aetiology who are treated with beta-lactam antibiotics as an outpatient will probably not respond to treatment, with the possible consequence being a deterioration of the situation and subsequent hospital admission. Our finding of an increased prevalence of atypical pathogens in patients with prior outpatient beta-lactam treatment supports such an explanation, but also confirms what has already been suggested in the different guidelines for the management of community-acquired pneumonia in adults [13, 18]. These guidelines state that following the failure of the initial empirical treatment with beta-lactam antibiotics, the microbiological examination should be reassessed with a view to excluding the less common pathogens, such as atypical pathogens, and that antimicrobial treatment covering atypical pathogens should be considered. Our study supports the choice of antimicrobial treatment covering atypical pathogens (e.g. macrolides) for all patients with CAP who are admitted to hospital after prior treatment with beta-lactam antibiotics. However, whether this protocol will result in improved clinical outcome should be subject to additional study. The observed reduction in the frequency of S. pneumoniae in patients who received prior outpatient antimicrobial treatment could also be due to a failure to detect the organisms in cultures. This could mask S. pneumoniae as the causative pathogen. However, we believe that this explanation is less plausible, especially since we also used antigen testing to identify the causative pathogen [19]. In addition, such a mechanism can not explain the finding of an increased probability of pneumonia caused by atypical pathogens. To the best of our knowledge, our study is the first to specifically document and quantify the failure of initial outpatient antibiotic treatment as a predictor of the microbial aetiology of CAP. This study was conducted in a single teaching hospital in The Netherlands, but we believe that the conclusions drawn will apply to other settings. First, the percentage of identified aetiology (64% in this study) is in agreement with that of other studies using similar microbiological techniques [20–22]. Second, our patient characteristics comply to a great extent with a previous nationwide study on prior outpatient antibacterial therapy as a prognostic factor for mortality in patients hospitalised for pneumonia [23]. In that large database study, the percentage of patients hospitalised after initial outpatient antimicrobial treatment was almost identical to that observed in our study (27 vs. 23%, respectively). In addition, age distribution, co-morbidities and the antibiotic utilisation profile of the outpatients were very similar as were the median duration of hospital stay and in-hospital mortality. Unfortunately, due to limited numbers, we were unable to study an association between prior outpatient antimicrobial treatment and mortality in our study. A very reassuring finding was that 85% of all outpatient antibiotic prescriptions complied with national guidelines on the initial treatment of adults with suspected pneumonia. This reduces the possibility that the findings, rather than being associated with antibiotics, might reflect the diagnostic acumen of the physicians who saw the patients in primary care. However, we cannot rule out the possibility that typical signs of infection with atypical pathogens may have been missed by primary care physicians. As well as finding a relation between prior antimicrobial treatment and aetiology, we also found an association between aetiology and age and pulmonary co-morbidity. Patients aged <60 years without co-morbidities were more likely to have an aetiology comprising viral or atypical bacterial pathogens, and pulmonary co-morbidity was independently associated with S. pneumoniae and H. influenzae as causative pathogens. These findings confirm the results of previous studies on the impact of age and co-morbidity on the microbial aetiology of CAP [6]. A limitation of the present study, however, is that we were not able to adjust for smoking habits and alcohol intake of the patients. Previous studies on determinants for pneumonia aetiology found that these factors are significant predictors of pneumococcal infection [6, 7]. On the other hand, we do not expect prior antimicrobial therapy and smoking and alcohol intake to coincide in such a way that this would result in a null effect when all the information is available. In conclusion, among patients admitted for pneumonia, whether or not a patient has received prior antimicrobial therapy as an outpatient provides relevant information in the diagnostic workup, in particular in terms of identifying the causative pathogen and planning the initial treatment at the time of hospital admission. This finding supports a further strengthening of the continuity of care at the interface between the extramural and hospitalised settings.

Histochem_Cell_Biol-4-1-2248610

Retrograde traffic in the biosynthetic-secretory route

In the biosynthetic-secretory route from the rough endoplasmic reticulum, across the pre-Golgi intermediate compartments, the Golgi apparatus stacks, trans Golgi network, and post-Golgi organelles, anterograde transport is accompanied and counterbalanced by retrograde traffic of both membranes and contents. In the physiologic dynamics of cells, retrograde flow is necessary for retrieval of molecules that escaped from their compartments of function, for keeping the compartments’ balances, and maintenance of the functional integrities of organelles and compartments along the secretory route, for repeated use of molecules, and molecule repair. Internalized molecules may be transported in retrograde direction along certain sections of the secretory route, and compartments and machineries of the secretory pathway may be misused by toxins. An important example is the toxin of Shigella dysenteriae, which has been shown to travel from the cell surface across endosomes, and the Golgi apparatus en route to the endoplasmic reticulum, and the cytosol, where it exerts its deleterious effects. Most importantly in medical research, knowledge about the retrograde cellular pathways is increasingly being utilized for the development of strategies for targeted delivery of drugs to the interior of cells. Multiple details about the molecular transport machineries involved in retrograde traffic are known; a high number of the molecular constituents have been characterized, and the complicated fine structural architectures of the compartments involved become more and more visible. However, multiple contradictions exist, and already established traffic models again are in question by contradictory results obtained with diverse cell systems, and/or different techniques. Additional problems arise by the fact that the conditions used in the experimental protocols frequently do not reflect the physiologic situations of the cells. Regular and pathologic situations often are intermingled, and experimental treatments by themselves change cell organizations. This review addresses physiologic and pathologic situations, tries to correlate results obtained by different cell biologic techniques, and asks questions, which may be the basis and starting point for further investigations. Introduction The biosynthetic-secretory route is traveled by newly synthesized luminal and membrane proteins and glycoproteins from their sites of synthesis at bound ribosomes of the rough endoplasmic reticulum (RER) to their final destinations inside and outside the cells (for review e.g. Farquhar and Hauri 1997; Mellman and Warren 2000). It involves complex and highly dynamic organelles, such as the Golgi apparatus, pre-Golgi intermediates, and post-Golgi organelles, which all have central roles in transport regulation, sorting, and targeting. The route that leads out of the RER at special ER-export sites, involves complicated ER-Golgi intermediate compartments, enters the Golgi apparatus stacks at their cis side, leads across the stacks of Golgi cisternae, where the traversing molecules are subjected to major modifications, and continues at the trans Golgi side. The trans Golgi network (TGN) is a central place of molecule sorting, and packaging of cargo molecules into vehicles for transport to the final intra or extracellular destinations, where the molecules eventually exert their specific functions, such as organelles of the lysosomal system, diverse domains of the plasma membrane, and the extracellular space. At all levels of the biosynthetic-secretory route, anterograde transport of membranes and cargo is counterbalanced by retrograde traffic (for review Sannerud et al. 2003). This is necessary for several reasons, which include the retrieval of molecules that have escaped from the sites of their specific functions, for membrane balance and the maintenance of the functional integrities of organelles and compartments along the secretory route and repeated use of molecules, and molecule repair. Major antero-retrograde traffic cycles are located at the pre-Golgi junction between the ER, the intermediate compartment, and the Golgi apparatus, as well as at the post-Golgi junctions between the TGN, early and late endosomes, and plasma membrane. Within the Golgi apparatus stacks, anterograde transport is assumed to be counterbalanced by retrograde traffic as well. Internalized molecules have been shown to travel along retrograde biosynthetic pathways (e.g. Gonatas et al. 1983; Pavelka et al. 1998; Vetterlein et al. 2002; Volz et al. 1995), and physiologic retrograde routes are misused by harmful substances to reach the sites of their specific toxic effects (e.g. Pelham et al. 1992; Rapak et al. 1997; Sandvig et al. 1991, 1992, 2002; for review Sandvig and van Deurs 2002, 2005). Since the early report in 1992 by Kirsten Sandvig and colleagues, who for the first time showed that along such pathways toxins could be transported en route from the plasma membrane into early compartments of the biosynthetic-secretory system, such retrograde trails of toxins are at the center of cell biologic and medical research. For both the cell physiologic retrograde routes, and misused pathways, knowledge about molecular machineries, and insights into regulatory mechanisms are increasing rapidly (e.g. Gokool et al. 2007; Mari et al. 2008; Popoff et al. 2007; Wälchli et al. 2008; Yamane et al. 2007; for review Bonifacino and Rojas 2006; Sannerud et al. 2003). In immuno and fine structural analyses, multiple fine details of the compartments, structures and complex architectures involved have been visualized (e.g. Mari et al. 2008; Marsh 2005; Mogelsvang and Howell 2006; Mogelsvang et al. 2004; Pavelka 2007; Pavelka et al. 1998; Vetterlein et al. 2002, 2003). However, there still exists a considerable lack in understanding the mechanisms involved and uncertainties in the interpretations of individual results obtained by molecular biologic and genetic investigations on one hand, and morphologic findings on the other hand. Well known functional processes yet cannot be exactly attributed to concrete defined compartments, and vice versa, the functional implications of multiple structures and architectures need to be clarified. There is a strong need to bring together the biochemical, molecular biological, genetic, morphologic, immuno-cytochemical, and fine structural results. Retrograde traffic in cell physiology Anterograde transport in the secretory route is accompanied and counterbalanced by retrograde traffic at each section of the secretory route, including pre- and post-Golgi areas, and presumably at the level of the stacks of Golgi cisternae as well. The diverse retrograde routes and cycles cannot be seen separately. There exist similarities in the mechanisms and machineries, close relationships, and mutual influences. Retrograde Golgi-to-ER traffic and pre-Golgi circuits By retrograde Golgi apparatus-to-ER traffic, ER-resident luminal and membrane proteins are retrieved (Munro and Pelham 1987; Semenza et al. 1990), and membrane proteins, including the cargo receptor ERGIC-53 (Appenzeller et al. 1999; Schweizer et al. 1988), and Golgi-resident proteins, such as glycosyltransferases, recycle (Lee et al. 2004; Lippincott-Schwartz et al. 1990; Storrie et al. 1998; Storrie 2005). It is evident that multiple retrograde traffic routes work in parallel in order to retrieve molecules to the ER, and cycle proteins and lipids between ER and Golgi apparatus. The first retrograde Golgi-to-ER pathway described was the KDEL-receptor mediated transport of luminal ER proteins (Lewis and Pelham 1990; Munro and Pelham 1987; Semenza et al. 1990), which following the export out of the ER together with the flow of multiple other proteins destined for transport to other destinations have to be retrieved to their sites of functions in the ER. This traffic, as well as the transport of type-I transmembrane proteins bearing the double-lysin motif signal (Letourneur et al. 1994) is mediated by COPI-coated vesicles (Nickel and Wieland 2002). The retrieval of escaped ER-proteins takes place most efficiently from the ER-Golgi intermediate compartment (ERGIC) but can occur from successive Golgi apparatus compartments up to the TGN as well. Luminal acidification, and calcium concentration in the lumina of the respective compartments have been suggested to be regulatory factors (reviewed in Sannerud et al. 2003). Furthermore, COPI-independent transport machineries act in Golgi-to-ER retrograde traffic (Johannes and Goud 2000). Such routes are known to involve Rab6 GTPases (Girod et al. 1999; Storrie 2005; White et al. 1999), key regulators of intracellular membrane traffic (Hammer and Wu 2002; Stenmark and Olkkonen 2001; Zerial and McBride 2001). Three different isoforms are known, Rab6A, Rab6A′, an alternatively spliced variant of Rab6A (Echard et al. 2000), and the brain-specific Rab6B (Wanschers et al. 2007). Recently, it has been shown that Rab6A and Rab6A′ perform different non-overlapping functions in cells, and the Rab6A′ isoform is shown mainly regulating the COPI-independent retrograde pathway to the ER (Del Nery et al. 2006). Rab6A GTPases interact with a subunit of the dynein–dynactin complex, and transport compartments move between Golgi apparatus and ER bidirectionally along the microtubules (Matanis et al. 2002; Short et al. 2002). COPI-independent Rab6A-dependent retrograde pathways are used by Golgi apparatus-resident proteins, such as glycosyltransferases, which cycle continuously between the Golgi apparatus and ER (Martinez et al. 1997; Rhee et al. 2005; Storrie 2005; Storrie et al. 1998). Retrograde traffic within the Golgi apparatus In the anterograde biosynthetic-secretory traffic, newly synthesized molecules arriving from the ER and ERGIC are taken up into the Golgi apparatus, visit cisternae of the stacks to be modified in well known subsequent steps, and are sorted to different further destinations at the trans side and TGN. Since decades, it is well established that the stacks of Golgi cisternae are subcompartmentalized into separate functional spaces, in each of which defined sets of enzymes are active, and collaborate in performing specific modifications of the molecules that visit the respective cisternae, e.g. perform changes of the sugar chains of the glycoconjugates, such as secretory and plasma membrane glycoproteins, enzymes of the lysosomal system, and lysosomal membrane constituents (Berger 1985; Farquhar and Palade 1998; Glick 2000; Pavelka 1987; Puthenveedu and Linstedt 2005; Rambourg and Clermont 1997; Roth 1997; Storrie 2005). It is assumed that the cisternae of the Golgi apparatus stacks are visited by the molecules to be modified, and there occurs a flow of membranes and contents across the stacks. However, at present, it is not clear, how in the dynamics of the flow the Golgi apparatus subcompartmentalization, which ensures effective glycosylation and other processing in the secretory pathway, is maintained. Conflicting results make interpretations complicated but there exist indications that retrograde transport of membrane constituents, such as the subcompartment-specific glycosyltransferases, is involved. The question how, and even whether, newly synthesized membrane and cargo proteins traverse the stacked Golgi cisternae, is a major point of debate. Several models are discussed, suggesting either an anterograde transport via vesicles, or anterograde flow by progression of the cisternae themselves, or by tubular connections between the cisternae, or by a signal-mediated temporary opening of channels connecting adjacent cisternae (Kartberg et al. 2005; Malhotra and Mayor 2006; Marsh and Howell 2002; Marsh et al. 2004; Mironov et al. 2005; Pelham and Rothman 2000; Puthenveedu and Linstedt 2005; Rodriguez-Boulan and Müsch 2005; Sallese et al. 2006; Storrie 2005, Trucco et al. 2004). Contradictory results also concern the retrograde traffic, which at least in the “cisternae progression model” is required for maintenance of the specific subcompartments. COPI-coated vesicles, also being involved in the retrograde Golgi-to-ER traffic (see the previous chapter), have been proposed as the candidates for retrograde intra-Golgi traffic but the results are conflicting (summerized by Rabouille and Klumperman 2005), and alternative mechanisms are considered as well, such as traffic across tubular connections between the Golgi cisternae (Marsh et al. 2004; Trucco et al. 2004). The retrograde transport and the cycling of Golgi-resident glycosyltransferases between Golgi apparatus and ER have been shown to be functionally connected with the maintenance of the structure of the Golgi apparatus (Starr et al. 2007; Storrie 2005). Recently, it has been shown that the Rab6-binding protein TMF (TATA element modulatory factor), which is involved in the Rab6-dependent retrograde transport processes both from endosomes to the Golgi apparatus, and from the Golgi apparatus to the ER, via its cytoplasmic region is implicated in the retention of N-acetylgalactosaminyltransferase-2 (Yamane et al. 2007). Retrograde plasma membrane-to-Golgi traffic and post-Golgi circuits In part comparable with the pre-Golgi retrograde routes, post-Golgi retrograde traffic on one hand represents a backward flow that counterbalances forward flow by anterograde traffic to the final destinations of newly synthesized molecules to the plasma membrane, extracellular space, secretory granules, endosomes, or lysosomes. Post-Golgi retrograde trafficking concerns TGN-resident proteins, such as TGN38, and receptors involved in the sorting of newly synthesized molecules of the trans Golgi apparatus and TGN to their specific sites of action. This transport is bidirectional; the receptors recycle to the TGN to be reused in further rounds, possibly also being modified or repaired (Bonifacino and Rojas 2006; Rohn et al. 2000; Snyder and Rogers 1985; Volz et al. 1995). Best-studied examples are the cation-independent and the cation-dependent mannose-6-phosphate receptors (CI- and CD-MPR; Ghosh et al. 2003). Other recycling proteins are the multi-ligand receptor sortilin (Mari et al. 2008), TGN38 (Banting and Ponnambalam 1997; Ghosh et al. 1998; Stanley and Howell 1993), GPP130 and GP73 (Puri et al. 2002), processing enzymes, such as the transmembrane endoproteases furin and carboxypeptidase D (Molloy et al. 1999; Varlamov and Fricker 1998), and SNAREs (soluble N-maleimide-sensitive fusion protein /NFS/ attachment protein receptor; Hettema et al. 2003; Hong 2005). At least, two independent retrograde routes from early endosomes to the TGN exist involving specific components of the Rab and SNARE machineries. One leads to the TGN via late endosomes (Barbero et al. 2002; Carrol et al. 2001; Lombardi et al. 1993; Mallet and Maxfiled 1999); the other one is a direct route from early and/or recycling endosomes to the TGN, thus bypassing late endosomes (Mallard et al. 1998, 2002). During the past years, insights into the regulatory mechanisms of retrograde traffic from endosomes to the TGN considerably increased. Multiple constituents of the molecular transport machineries have been characterized and their roles specified; these include Rab9 and TIP47 (tail-interacting protein 47 kDa; Diaz et al. 1997), PACS1 (phosphofurin acidic cluster sorting protein 1) and AP-1 (Crump et al. 2001; Meyer et al. 2000), EpsinR (Saint-Pol et al. 2004), the t-SNAREs syntaxin 16 and 5 (Amessou et al. 2007), and constituents of the retromer complex with the two subcomplexes, a membrane-bound coat that consists of the phosphinositide binding proteins Sorting nexin 1 and possibly Sorting nexin 2, and the cargo-binding proteins Vps26, Vps29, and Vps35 (Arighi et al. 2004; Gokool et al. 2007; Popoff et al. 2007; Restrepo et al. 2007; Rojas et al 2007; Seaman et al. 1998; Seaman 2004, 2005; for review Bonifacino and Rojas 2006). Clathrin and the retromer complex are suggested to function in consecutive retrograde sorting steps on early endosomes (Popoff et al. 2007). There exist several different carrier compartments for retrograde endosome-to-TGN traffic that can form at the same vacuolar early endosome (Mari et al. 2008): tubular sorting endosomes (Peden et al. 2004), and tubular endosomal networks (Bonifacino and Rojas 2006), and the endosome-to-TGN carriers (Mari et al. 2008). The tubular endosomal network (TEN) is connected to vacuolar early endosomes, which exhibit bilayered coats composed of clathrin and Hrs (hepatocyte-growth-factor-regulated tyrosine kinase substrate) thought to contain the ESCRT machinery that targets proteins to the intraluminal vesicles, thus being sorted to the multivesicular-late endosomal-lysosomal route (Hurley and Emr 2006; Sachse et al. 2002). The vacuolar endosomal part is suggested to receive endocytic cargo from the plasma membrane on one hand, and biosynthetic cargo, e.g. lysosomal enzymes, from the TGN on the other hand; it progressively becomes acidic, leading to the release of the cargo from the receptors. It is proposed that the vacuolar endosomal region matures into late endosomes, whereas from the TEN-region, cargos are sorted to different destination, such as to the TGN and Golgi apparatus along the retrograde route but also to different regions of the plasma membrane for recycling and transcytosis, and to specialized storage compartments, such as melanosomes. The TEN is equipped with different transport machineries including the retromer. It is considered that the pericentriolar endocytic recycling compartment (Ghosh et al. 1998; Maxfield and McGraw 2004; Ullrich et al. 1996) could be a specialized subdomain of TEN. With respect to its central sorting role, TEN is proposed to represent a “mirror image” of the TGN (Bonifacino and Rojas 2006). A possibly comparable endocytic trans-Golgi network has been shown to develop after internalization of wheat germ agglutinin (WGA; Pavelka et al. 1998; Vetterlein et al. 2002; Figs. 1, 2). Recently, a novel class of carriers, the endosome-to-TGN carriers (ETCs), has been characterized (Mari et al. 2008). These carriers transport mannose-6-phosphate receptors and sortilin; they are dependent on the presence of sorting nexin 1 (SNX1), and show unique structures appearing as non-branched tubules and vesicles, clearly different from the tubular sorting endosomes (Peden et al. 2004) and tubular networks (Bonifacino and Rojas 2006). Fig. 1The Golgi apparatus reorganizes during endocytosis of WGA within a period of 30 min. ainsert: The insert shows WGA-HRP reaction products concentrated in a clathrin-coated pit, and a coated vesicle (arrows), as well as lining the membrane of a small uncoated vesicle (arrowhead). ×35,000. a Small endosomes (arrowheads), and a large vacuolar endosome (arrow) are apparent in the cytocentre (c), and close to a small Golgi apparatus stack (Golgi). A distinct TGN is not visible. ×30,000. b Globular early endosomes are accumulated close to the trans side of Golgi apparatus stacks (Golgi); some of the endosomes are covered with clathrin coats (arrowheads); partly, they exhibit WGA reaction products attached to the membranes, partly contained within the lumina. Some of endosomes appear to contact each other, and fine bridges are visible (short arrows). WGA reactions are apparent within the cisternae of a small Golgi apparatus stack (large arrow). ×31.500. c A network, an endocytic trans Golgi network (endocytic TGN), is apparent consisting of interconnected globular pieces (arrowheads) that resemble the earlier globular endosomes. Again here, the WGA reaction products are either attached to the membranes, or fill the lumina. Parts of this endocytic TGN are attached to stacks of Golgi cisternae at their trans sides (arrows), and are associated with trans Golgi ER. At the cis side, ER-Golgi-intermediate compartments (ERGIC) are visible. ×28,000. d Endocytic TGN consisting of interconnected globular pieces (arrowheads) and filled with WGA reaction products are partly attached to Golgi apparatus stacks (arrows), and associated with trans Golgi ER. ×32,000Fig. 2Portions of the endocytic TGN are integrated in the Golgi apparatus stacks a The WGA-labeled endocytic TGN (long arrows) represents an integrated part of this Golgi apparatus stack, and is closely associated to the transmost Golgi cisterna at one side, and with trans Golgi ER at the opposite side. One large vacuolar part (V) of the endocytic TGN is visible; some of the globular pieces of the endocytic TGN are covered with clathrin coats (short arrow). ×35,000. b The endocytic TGN (arrows) consists of globular pieces, and a large vacuole part (V), which here is attached to the Golgi apparatus stack. ×44,000Fig. 3a–d This 3D-model of a Golgi apparatus stack containing integrated endocytic TGN, has been formed according to the data of an electron tomography tilting series. The model is shown from different sides. The endocytic TGN (green) consists of a free portion (long arrows), and another portion integrated in the stack (short arrows), and is closely associated with trans Golgi ER (red). In both the free and the Golgi-associated portions, globular elements are visible. Stacked Golgi cisternae are shown in yellow. In c, the stacked Golgi cisternae are removed. Spikes mark clathrin coats Retrograde traffic of internalized molecules The plasma membrane-endosome-TGN pathway used by plasma membrane proteins for reentry into the biosynthetic system for reuse, or modification and repair (Snyder and Rogers 1985; Volz et al. 1995; reviewed in Bonifacino and Rojas 2006) also opens the secretory pathway for entry of extracellular ligands, including harmful substances, such as bacterial and plant toxins (for review Sandvig and van Deurs 2005), and provides tracks for drug delivery to the interior of cells (for review Tarrago-Trani and Storrie 2007). Retrograde routes of lectins and toxins Since more than two decades, it is known that lectins, and bacterial and plant toxins, such as the Shiga and Cholera toxins, the Pseudomonas exotoxin A, and ricin, utilize the retrograde routes within the cells to travel to the TGN and Golgi apparatus, ER, and cytosol (e.g. Gonatas et al. 1983, Sandvig and Brown 1987; van Deurs et al. 1986, 1987). Sandvig et al. (1992) were the first, who showed that Shiga toxin is transported from the cell surface en route to the endoplasmic reticulum. Shiga toxin consists of an enzymatically active A-subunit that is noncovalently linked to a pentamer of B-chains. The latter binds to the glycosphingolipid Gb3, and mediates the toxin’s transport. Following internalization, the toxin is transported to early endosomes, sorted to the TGN and Golgi apparatus, and further transported to the ER, from where after cleavage the A-subunit is retrotranslocated to the cytosol, the site where it exerts its toxic action by inactivation of ribosomes, and inhibition of protein synthesis (Johannes and Goud 1998, 2000; Sandvig and van Deurs 1994, 1996, 2002, 2005). It is well established that different uptake mechanisms are used, and there exists not only one route to the endoplasmic reticulum but different retrograde pathways may be traveled by different toxins, and trafficking of one class of toxin is not limited to one route, e.g. Pseudomonas exotoxin A has been shown to travel to the endoplasmic reticulum along multiple pathways (Smith et al. 2006). There also appears to be fundamental differences between the endosomal sorting into the retrograde pathway to the TGN of Shiga and Cholera toxins (Bujny et al. 2007; Chinnapen et al. 2007; Feng et al. 2004; Massol et al. 2004; Torgersen et al. 2001). Multiple details concerning different uptake mechanisms, and retrograde traffic of Shiga toxin and ricin have been elucidated during the past years (e.g. Amessou et al. 2007; Garred et al. 1995; Grimmer et al. 2005, 2006; Johannes et al. 1997; Lauvrak et al. 2004, 2006; Mallard et al. 1998; Rapak et al. 1997; Römer et al. 2007; Skanland et al. 2007; Slominska-Wojewodzka et al. 2006; Tai et al. 2004; Torgersen et al. 2007; Utskarpen et al. 2006, 2007; Yoshino et al. 2005). A key compartment is the early endosome (see previous chapter), from where internalized molecules, are either recycled to the plasma membrane to be reused, or sorted to the late endosomal pathway and to the lysosomes to be degraded, or transferred to a direct pathway to the TGN, travelled by endogenous proteins, like the mannose-6-phosphate receptors (for review Bonifacino and Rojas 2006). It is the latter route that is travelled by Shiga toxin and ricin. For efficient transport of Shiga toxin from early endosomes to the TGN and Golgi apparatus, the retromer complex is required (Bujny et al. 2007; Popoff et al. 2007; Utskarpen et al. 2007). The ricin endosome-to-TGN and Golgi traffic has been shown to be facilitated by depletion of sphingolipid (Grimmer et al. 2006), and to be dependent on Rab6A and Rab6A′ (Utskarpen et al. 2006), which are also regulators in pre-Golgi circuits (see previous chapter). The syntaxins 5 and 16 being involved in retrograde transport of mannose-6-phosphate receptors, also are necessary for efficient retrograde traffic of Shiga toxin, as well as for trafficking into the cells of ricin and cholera toxin (Amessou et al. 2007). In the center of interest are the regulatory roles of phosphoinositides (for review De Matteis and Godi 2004), and the importance of signaling (Pelkmans et al. 2005; Perret et al. 2005; von Zastrow and Sorkin 2007). Recently, it has been demonstrated that retrograde traffic of Shiga toxin and ricin are phosphoinositide-regulated (Skanland et al. 2007; Utskarpen et al. 2007). The phosphoinositide-binding proteins sorting nexins 1 and 2, being part of the retromer complex (Seaman 2004, 2005), are necessary for efficient transport of Shiga toxin to the Golgi apparatus (Bujny et al. 2007; Popoff et al. 2007; Utskarpen et al. 2007). Sorting nexins have been shown to be in crosstalk with the phosphatidylinositol (PI) 3-kinase hVps34; it is proposed that hVps34 produces a specific PI(3)P pool needed for the localization of sorting nexins on endosome vesicles, which in turn is required for retrograde endosome-to-Golgi traffic of ricin (Skanland et al. 2007). Shiga toxin is an active player in its own transport mediating both internalization, and intracellular transport. Upon binding to the plasma membrane or entry into the cells, it is able to trigger signaling cascades (Ikeda et al. 2000; Lauvrak et al. 2006; Wälchli et al. 2008). Shiga toxin activates the tyrosine kinase Syk, by which clathrin phosphorylation and uptake of Shiga toxin is induced (Lauvrak et al. 2006). Protein kinase Cδ is specifically activated by Shiga toxin regulating endosome-to-Golgi transport (Torgersen et al. 2007). Evidence is provided for activation of a signaling cascade that involves a crosstalk between Ca2+ and the MAP kinase p38; it is suggested that Shiga toxin, by modifying Ca2+ homeostasis, recruits p38 to endosomes for regulation of transport to the Golgi apparatus (Wälchli et al. 2008). The entry of internalized molecules into the secretory pathway at the TGN and Golgi apparatus level is a major event influencing further retrograde traffic to the ER and cytosol. Although it is well established, and has been shown ultrastructurally very early for several toxins, that transport to the TGN and Golgi apparatus is followed by uptake into cisternae of the Golgi apparatus stacks (e.g. Sandvig et al. 1992; van Deurs et al. 1987), the involved mechanisms are still poorly understood. A detailed ultrastructural analysis has been undertaken using a HepG2 hepatoma cell model, and uptake of wheat germ agglutinin labeled with horseradish peroxidase (WGA-HRP; Pavelka et al. 1998; Pavelka 2007; Vetterlein et al. 2002). Wheat germ agglutinin is an N-acetyl-glucosamine and sialic acid-specific lectin, which is known for many years to be transported to the Golgi apparatus in retrograde direction (e.g. Gonatas et al. 1977; Stieber et al. 1984); it is also used in connection with the developments of drug delivery systems (Lochner et al. 2003; Weissenböck et al. 2004). WGA reacts with numerous binding sites at the cell surface, and is taken up in large amounts, thus mimicking normal situations in cells, and here particularly reflecting the functions of hepatocytes in the liver tissue. HepG2 hepatoma cells are especially well suited for these studies, since prior to, and in the initial phases of WGA endocytosis, the Golgi apparatus mainly consists of small inconspicuous stacks of cisternae (Fig. 1a), and the changes and reorganizations of the Golgi apparatus during WGA-endocytosis are clearly visible (Figs. 1, 2, 4). Briefly, the results show that internalized WGA, following internalization via clathrin-coated vesicles (Fig. 1a insert) and possibly other mechanisms as well, is rapidly transported to the Golgi apparatus, and induces dramatic Golgi reorganizations. Early globular endosomes accumulate at the trans Golgi side (Fig. 1b) and a network is formed, an endocytic trans-Golgi network (endocytic TGN) that consists of interconnected globular pieces (Fig. 1c), which in dimensions and shapes resemble the earlier globular endosomes (Fig. 1b). In these compartments, the WGA reaction products in part detach from the limiting membranes, and fill the lumina (Fig. 1b, c, d) indicating that the luminal milieu is changing. Portions of the endocytic TGN in close association with trans-Golgi ER, attach to trans Golgi cisternae (Fig. 1c, d), thus becoming integrated parts of the Golgi apparatus stacks (Figs. 2a, b, 4a; various views of a 3D-model are presented in Fig. 3a–d). The trans-Golgi attachment of the endocytic TGN leads to interconnections of the small Golgi stacks (Figs. 1c, d, 4b), and causes the formation of Golgi apparatus ribbons. Concomitantly, and in part prior to the formation of an endocytic TGN, internalized WGA appears within cisternae of Golgi apparatus stacks (Figs. 1b, d, 4a). Stacks that contain high amounts of WGA within the lumina of all the cisternae are mainly found at later periods of WGA endocytosis, e.g. after 45–60 min (Fig. 4b). Very seldom, internalized WGA can be localized within the ER cisternae. At all periods of WGA endocytosis, internalized WGA appears within multivesiculated bodies (Fig. 4a, b), indicating that internalized WGA not only trafficks into the Golgi apparatus but is sorted to the late endosomal–lysosomal pathway as well. Fig. 4WGA is taken up into the stacked Golgi cisternae a Free portions (long arrows), and Golgi-integrated portions (short arrows) of the WGA-reactive endocytic TGN are visible. Fine WGA reactions also are apparent at confined regions of some of the stacked Golgi cisternae (double arrow). Several multivesiculated bodies (MVB) are densely filled with WGA-reaction products, and show domains, where tubular and vesicular transport carriers are formed. ×20,000. b A Golgi ribbon is visible, in which the individual stacks of cisternae are interconnected at the cis, as well as at the trans side. Most of the stacked Golgi cisternae are densely filled with WGA reaction products; others show reactions at limited regions (arrow). At the right lower corner, connections of medial cisternae with the cis-most cisterna of the same stack, and of other stacks are shown. In a vacuolar endosome (V), WGA reaction products are attached to the limiting membrane, and to an intra-vacuolar vesicle (arrowhead). ×23,000 The detailed knowledge of these endocytosis-induced Golgi reorganizations has been used to develop a precise time schedule for regulated retrograde transport of WGA into the endoplasmic reticulum by treatment with Brefeldin A (Vetterlein et al. 2003). However, multiple questions are open, which concern the mechanisms of formation of the endocytic TGN, and its attachment to the Golgi stacks, the delivery of cargo, and uptake of membranes, interactions of transport carriers with Golgi subcompartments, mechanisms of retrograde transport within the Golgi apparatus stacks, signaling and the role of contact points, the importance of lipid transfer (De Matteis et al. 2007; Hanada et al. 2007), and possible direct endocytic TGN-to-ER traffic via trans-Golgi associated ER. Of particular interest are questions, as to whether secretory anterograde flow influences retrograde traffic, and vice versa, as to whether retrograde flow influences Golgi apparatus size and localization, and whether the formation of Golgi ribbons by retrograde flow is comparable with the formation of Golgi ribbons by input of membrane derived from the ER (Marra et al. 2007). Concluding remarks In future studies, it will be important to correlate the detailed fine structural findings with the molecular biologic and genetic results on the machineries and regulatory mechanisms of retrograde traffic (e.g. Amessou et al. 2007; Bonifacino and Rojas 2006; De Matteis et al. 2007; Hanada et al. 2007; Levine and Loewen 2006; Marra et al. 2007; Missiaen et al. 2007; von Zastrow and Sorkin. 2007; Wälchli et al. 2008). The crucial roles of retrograde traffic in the biosynthetic-secretory route for cellular homeostasis and intoxication of cells, the importance for assessment of effects and side effects of drugs (e.g. Sandoval and Molitoris 2004), and the development of strategies for targeted drug delivery to the interior of cells (El Alaoui et al. 2007; Johannes and Decaudin 2005; Kreitman 2006; Smith et al. 2002; Tarrago-Trani and Storrie 2007; Weissenböck et al. 2004) will be the driving forces.

J_Mol_Evol-3-1-1894752

The Mechanisms of Codon Reassignments in Mitochondrial Genetic Codes

Many cases of nonstandard genetic codes are known in mitochondrial genomes. We carry out analysis of phylogeny and codon usage of organisms for which the complete mitochondrial genome is available, and we determine the most likely mechanism for codon reassignment in each case. Reassignment events can be classified according to the gain-loss framework. The “gain” represents the appearance of a new tRNA for the reassigned codon or the change of an existing tRNA such that it gains the ability to pair with the codon. The “loss” represents the deletion of a tRNA or the change in a tRNA so that it no longer translates the codon. One possible mechanism is codon disappearance (CD), where the codon disappears from the genome prior to the gain and loss events. In the alternative mechanisms the codon does not disappear. In the unassigned codon mechanism, the loss occurs first, whereas in the ambiguous intermediate mechanism, the gain occurs first. Codon usage analysis gives clear evidence of cases where the codon disappeared at the point of the reassignment and also cases where it did not disappear. CD is the probable explanation for stop to sense reassignments and a small number of reassignments of sense codons. However, the majority of sense-to-sense reassignments cannot be explained by CD. In the latter cases, by analysis of the presence or absence of tRNAs in the genome and of the changes in tRNA sequences, it is sometimes possible to distinguish between the unassigned codon and the ambiguous intermediate mechanisms. We emphasize that not all reassignments follow the same scenario and that it is necessary to consider the details of each case carefully. Introduction: Distinguishing Possible Mechanisms of Codon Reassignment Now that many complete genomes of organisms and organelles are available, there is ample evidence that the genetic code is not as universal (Knight et al. 2001a; Yokobori et al. 2001; Santos et al. 2004) as previously believed (Crick 1968). Many cases are now known where a codon (or related group of codons) has been reassigned from one amino acid to another, from a stop to an amino acid, or from an amino acid to a stop. If a change in the translation system occurs in an organism such that a codon is reassigned, most of the occurrences of this codon will still be at places where the old amino acid was preferred. We would expect the changes causing the codon reassignment to be strongly disadvantageous and to be eliminated by selection. It is possible for mutations to cause disappearance of the codon in its original positions and reappearance in positions where the new amino acid is preferred. Mutations throughout the genome are required for it to readjust to the change in the genetic code. The problem is therefore to understand how codon reassignments can become fixed in a population despite being apparently deleterious in the intermediate stage before the genome has time to readjust. Various mechanisms have been proposed to explain the process of codon reassignment. We have recently shown (Sengupta and Higgs 2005) that these mechanisms can be described within a framework that we call the gain-loss framework. “Gain” refers to the gain of a new tRNA gene that is able to translate the reassigned codon as a different amino acid, or the gain of function of an old tRNA gene (e.g., by base modification in the anticodon) so that it translates the reassigned codon in addition to the codons it previously interacted with. “Loss” refers to the deletion of an existing tRNA for the reassigned codon or the loss of function of the tRNA so that it can no longer translate this codon. We identified four mechanisms in the gain-loss framework, as described below. Codon disappearance (CD) mechanism This was originally proposed by Osawa and Jukes (1989, 1995). For an amino acid (or stop) with more than one codon, it is possible for all occurrences of a codon to be replaced by synonymous codons, so that the first codon disappears entirely from the genome. After this, the gain and loss in the translation system are neutral changes that do not affect the organism. After the gain and loss occur, the codon may reappear in the genome by mutations at sites where the new amino acid is preferred. The distinguishing element of this mechanism is that the codon disappears first, and the gain and loss occur during a period in which the codon is absent. For the other three mechanisms described here, the codon does not need to disappear before the change. Ambiguous intermediate (AI) mechanism This was proposed by Schultz and Yarus (1994, 1996). They argued that a codon does not need to disappear in order to be reassigned and proposed that there is a transient period when the codon is ambiguously translated as two distinct amino acids. In terms of our gain-loss framework, this corresponds to the case where the gain occurs before the loss, i.e., there are two different tRNAs specific to the amibiguous codon during the intermediate period, and the new code becomes established when the old tRNA is lost. Unassigned codon (UC) mechanism This mechanism arises as a natural possibility in our gain-loss framework (Sengupta and Higgs 2005). It corresponds to the case where the loss occurs before the gain. There is an intermediate period where there is no tRNA available that can efficiently translate the codon; hence we say the codon is unassigned. The new code becomes established when the gain in function of the new tRNA occurs and the codon is reassigned to the new amino acid. If a codon were truly unassigned, and no tRNA could translate it at all, then the loss of the original tRNA would be lethal if the codon had not previously disappeared. However, several cases are known where an alternative tRNA is able to translate a codon (albeit less efficiently) after the tRNA that was specific to that codon has been deleted (Yokobori et al. 2001). Deletion of tRNAs appears to be frequent in mitochondrial genomes and we argue below that deletion of a tRNA is the primary event that instigated several of the mitochondrial codon reassignments. Compensatory change mechanism The final mechanism that occurs in the gain-loss framework is referred to as compensatory change because of its analogy with compensatory mutations in molecular evolution. Kimura (1985) considered a pair of mutations such that each is deleterious when it occurs alone, but when both occur together they are neutral, e.g., in the paired regions of RNA secondary structures (Higgs 1998, 2000; Savill et al. 2001). The gain and the loss in codon reassignment are changes in two different parts of a genome that form a compensatory pair. It is possible that one of these changes occurs but remains infrequent in the population until the second change occurs in an individual that already has the first change. Once the gain and loss are present in the same individual, they can spread simultaneously through the population, although they did not occur at the same time. In the compensatory change case there is no point at which individuals with ambiguous codons or UCs are frequent in the population.We showed using a population genetics simulation (Sengupta and Higgs 2005) that all four mechanisms can occur within the same model depending on the parameter values. Here, we consider which mechanisms occur in real cases. We limit ourselves to mitochondrial changes because most of the observed changes are in mitochondria and because the availability of substantial numbers of complete mitochondrial genomes makes it possible to pinpoint the changes and to study codon usage and tRNA gene content in the genomes before and after the reassignment. Studies of this type were carried out several years ago by Knight et al. (2001a, 2001b) and more recently by Swire et al. (2005). We compare our results with these previous surveys in the discussion section.In order to interpret the changes in the genetic code, phylogenetic trees are required. We split our set of species into groups that are consistent with previous studies. Figure 1 shows fungi and related species, and Fig. 2 shows plants/algae and related species. These were obtained by our own phylogenetic analysis of mitochondrial genes. Figure 3 shows alveolates/stramenopiles and related species, and Fig. 4 shows metazoa. These were obtained using combined information from other sources. Details of phylogenetic methods are given in the Supplementary Information. Fig. 1.Phylogeny of fungi and related species derived from mitochondrial proteins.Fig. 2.Phylogeny of plants and algae derived from mitochondrial proteins.Fig. 3.Phylogeny of alveolates, stramenopiles, and haptophytes according to published sources.Fig. 4.Phylogeny of metazoa according to published sources. Reassignments That Can Be Explained by Codon Disappearance Reassignments of UGA from Stop to Trp The UGA Stop-to-Trp change is the most frequently occurring reassignment known. The review by Knight et al. (2001a) lists six of these in mitochondria (and four in nuclear genomes). Our updated analysis of the mitochondrial data identifies the following 12 cases in mitochondria.Metazoa, Monosiga, and Amoebidium: This change is shared by all Metazoa and by their two closest known relatives (Fig. 1 and Lang et al. 2002).Acanthamoeba: Fig 1. and Burger et al. (1995).Basidiomycota (Crinipellis and Schizophyllum): Fig. 1.Ascomycota (group containing Penicillium and relatives): Fig. 1.Ascomycota (group containing Yarrowia and relatives): Knight et al. (2001a) list a single change in the ancestor of Metazoa, Acanthamoeba, and Fungi. However, Fig. 1 shows that UGA remains a stop codon in Chytridiomycota, Zygomycota, and Dictyostelium. Therefore, it is likely that cases i, ii, and iii are separate reassignments. The reassignment has also occurred in almost all the Ascomycota, with the exception of the Schizosaccharomyces group. This implies that cases iv and v in the Ascomycota are also separate reassignments. These conclusions depend on the argument that a reversal of this change (from Trp to Stop) is very unlikely, which we discuss below.Rhodophyta (Chondrus, Porphyra): Fig. 2 and Burger et al. (1999).Pedinomonas: Fig. 2 and Turmel et al. (1999).Haptophytes: This change was reported in a subgroup of haptophytes including Phaecocystis and Isochrisis by Hayashi-Ishimaru et al. (1997). The complete genome of Emiliana (Sanchez-Puerta et al. 2004) shows that it also possesses the reassignment. This is consistent with our phylogeny (Fig. 3).Ciliates (Paramecium, Tetrahymena): Knight et al. (2001) place this change at the base of the alveolates. However, our analysis of codon usage and sequences from Plasmodium species discussed below shows that UGA remains a Stop codon in Plasmodium. Thus the reassignment is not shared by all alveolates.Cafeteria: Fig. 3.Bacillariophyta (Skeletonema, Thalassiosira): Fig 3 and Ehara et al. (2000)Kinetoplastida (Trypanosoma, Leishmania): Inagaki et al. (1998) (not shown in figures). In species in which only the UGG codon codes for Trp, the tRNA-Trp has a CCU anticodon. Mutation of the wobble position C to a U creates a UCU anticodon that can pair with both UGA and UGG: hence the reassignment of UGA to Trp. We have analyzed the tRNA-Trp sequences of all available mitochondrial genomes to determine in which species this mutation has occurred. We find that all species in which UGA is Stop have a CCU anticodon, as expected. Almost all species in which UGA is Trp have a UCU anticodon, but we find some exceptions below. Table 1 shows the number of occurrences of each of the standard Stop codons together with the tRNA-Trp anticodon. When UGA is Stop it is usually used less frequently than the preferred stop codon UAA. When the reassignment occurs, many UGG Trp codons mutate synonymously to UGA; hence UGA becomes frequent. This is seen in all species having a UCA anticodon. However, UGA is also frequent in Amoebidium, Crinipellis, and Schizophyllum, which have a CCA anticodon. Therefore there must be a posttranslational modification of the C base in the anticodon that permits translation of UGA, rather than a mutation in the gene. Another special case is the kinetoplastids. These have no mitochondrial tRNAs, and import all the required tRNAs from the nucleus. The tRNA-Trp from the nucleus has CCA anticodon because the canonical code is used in the nucleus. The same tRNA is imported to the mitochondrion and then undergoes a base modification so that it can translate both UGA and UGG (Alfonzo et al. 1999). Table 1.Codon usage data relevant to reassignments of Stop codons UGA and UAGUGA istRNA-Trp anticodonCodon usageUGAUAAUAGSpecies in Fig. 1  Amoebidium castellaniiTrpCCAa782812  Dictyostelium discoideumStopCCA2308  Monosiga brevicolisTrpTCA124284  Homo sapiensbTrpTCA9283  Rhizophidium sp. 136StopCCA532117c  Spizellomyces punctatusStopCCA1220140c  Hyaloraphidium curvatumNot usedCCA0162  Monoblepharella sp. JEL 15StopCCA1205  Harpochytrium sp. JEL105Not usedCCA0122  Harpochytrium sp. JEL94Not usedCCA0131  Allomyces macrogynusNot usedCCA01614  Mortierella verticillataRareCCA2e214  Rhizopus oryzaeNot usedCCA0204  Crinipellis perniciosaTrpCCAa1247316  Schizophyllum communeTrpCCAa25191  Penicillium marneffeiTrpTCA61161  Hypocrea jecorinaTrpTCA89145  Schizosaccharomyces japonicusNot usedCCA061  Schizosaccharomyces octosporusNot usedCCA080  Schizosaccharomyces pombeRareCCA1e70  Yarrowia lipolyticaTrpTCA57122  Candida stellataTrpTCA3480  Candida albicansTrpTCA4958  Saccharomyces cerevisiaeTrpTCA124190Species in Fig. 2  Malawimonas jakobiformisStopCCA1462  Cyanidioschyzon merolaeStopCCA2293  Chondrus crispusTrpTCA101223  Porphyra purpureaTrpTCA117274  Chaetosphaeridium globosumStopCCA7309  Chara vulgarisStopCCA8308  Prototheca wickerhamiiNot usedCCA0351  Pseudoendoclonium akinetumStopCCA114021  Pedinomonas minorTrpTCA62110  Scenedesmus obliquuseStopCCA12141b  Chlamydomonas eugametosNot usedCCA0122  Chlamydomonas reinhardtiiNot usedCCA062Species in Fig. 3  Emiliana huxleyiTrpUCA73192  Rhodomonas salinaStopCCA1349  Naegleria gruberiNot usedCCA0379  Plasmodium reichenowiNot usedNone030  Plasmodium falciparumNot usedNone030  Paramecium aureliaTrpTCA832917  Tetrahymena pyriformisTrpTCA228440  Tetrahymena thermophilaTrpTCA228450  Caferteria roenbergensisTrpTCA190322  Phytophthora infestansStopCCA1390  Saprolegnia feraxNot usedCCA0421  Chrysodidymus synuroidesNot usedCCA0343  Ochramonas danicaNot usedCCA03014  Laminaria digitataStopCCA4296  Pylaiella littoralisStopCCA7387aThese species have a CCA anticodon but UGA is Trp. This suggests that the C base is modified in the tRNA.bIn vertebrates AGA and AGG are also Stop codons. Each is used once in Homo sapiens.cUAG is reassigned to Leu in these species.dThese codons are presumed to be inefficiently translated as Trp by the tRNA-Trp(CCA).eIn S. obliquus, UCA is a novel Stop codon, occurring 17 times. The predictions from our theory and simulations (Sengupta and Higgs 2005) were that Stop codons are most likely to be reassigned via the CD mechanism because they are rare in the first place (and chance disappearance is therefore relatively likely) and, also, because if the codons do not disappear, the penalty for read-through of a stop codon is likely to be larger than the penalty for mistranslation of an amino acid. (However, read-through of a stop codon is not necessarily lethal, especially if the number of additional codons translated until the next random occurrence of a stop codon is not too large.) The figures in Table 1 make a strong case that the CD mechanism is responsible. UGA is rare in almost all species where it is used as a Stop codon. Many of these genomes have high AU content, which is probably the reason UAA is preferred over UGA as a Stop codon. Many of the closest relatives to the species where UGA is reassigned have particularly low usage of UGA, e.g., Allomyces and Rhizopus, close relatives of the Basidiomycota, have zero; Cyanidioschyzon, a close relative of Chondrus and Porphyra, has only two; Scenedesmus and Chlamydomonas, close relatives of Pedinomonas, have one and zero respectively; and Plasmodium, a close relative of Paramecium and Tetrahymena, has zero. This last example shows that UGA is not reassigned in Plasmodium, as mentioned in case ix above. These examples clearly show that disappearance of UGA is possible and that, in many of the cases, there is good evidence that UGA was absent or almost absent at the time it was reassigned. A mutation pressure from GC to AU, which is implicated in the disappearance of UGA, will also tend to cause rapid mutations from UGG Trp codons to UGA after the Trp tRNA gains the ability to decode UGA. This is one reason that reversal of the change is unlikely. There are often around 100 UGAs in genomes where the reassignment has occurred, and it would be very difficult for this large number to disappear by chance because this would act against the mutation pressure. A second reason is that the reassignment to Trp would be associated with the loss of function of the release factor that originally interacted with the UGA. A reversal would also require regaining of the function of the release factor. The unlikeliness of the reversal of the reassignment is important in our interpretation of cases i to v above. The codon usage and tRNA-Trp anticodon data (Table 1) show that UGA is not established as a Trp codon in Dictyostelium, Chytridiomycota, Zygomycota, or Schizosaccharomyces. This means that multiple reassignment events are required to explain the observed pattern. It appears that UGA was rare and prone to disappearance right from the base of the Metazoa/Fungi tree in Fig. 1, and the release factor may already have lost its function. Therefore all that is required for the reassignment to occur is the simple mutation of the tRNA anticodon. In groups where the release factor was lost, but the tRNA mutation did not occur, the UGA codon would be effectively unassigned and would be selected against. Seif et al. (2005) argue that this has occurred in Mortierella and Schizosaccharomyces, where there is evidence of a small number of UGA acting as Trp codons that are translated very inefficiently by the standard Trp tRNA with CCA anticodon. A genetic code change might easily become established in these species in the future if the mutation occurred in the tRNA. The Probability of Disappearance of UGA Codons It is possible to calculate the probability Pdis that a codon will disappear using a simple model of the mutation process. Swire et al. (2005) have used this method to show that in most cases of stop codon reassignment, the probability of disappearance of the codon was relatively large, whereas in many cases of sense codon reassignments, the probability was extremely small. Hence they argue that stop codons were reassigned via CD but sense codons were not. We agree with this conclusion in almost all cases, but there are a few cases where we argue for CD in sense codon reassignments as well. In this section, we consider only UGA stop codons. The probability of UGA CD can be calculated as follows. Let the equilibrium frequencies of the bases under the mutational process be πA, πC, πG, and πU. Let fUGA, fUAA, and fUAG be the relative frequencies of the three stop codons (fUGA + fUAA, + fUAG = 1). If there is no selective preference of one stop codon over another, we expect these frequencies to be in equilibrium under mutation. Therefore Hence . In mitochondrial genomes, the two strands of the genome are not equivalent, the four base frequencies are all different, and it is not true that πC = πG and πA = πU (see Urbina et al. 2006). The values of the frequencies can be estimated from the frequencies of the bases at fourfold degenerate (FFD) sites. These are calculated by summing over all third-position sites that are FFD. If the total number of stop codons in the genome is Nstop, then the probability of disappearance of UGA is . We do not know the values of Nstop or fUGA at the point where the codon reassignment occurred, but we can use the species that are close to the reassignment point on the tree as an estimate, as shown in Table 2. Nstop is the the sum of UAA and UAG from Table 1. UGA codons do not contribute to Nstop because these are now Trp codons in these species. We used Monosiga as a proxy for case i. The resulting probability is high (Pdis = 0.17) and is consistent with the CD mechanism. Amoebidium was not used in this case because it has a derived multichromosome structure of the mitochondrial genome, and metazoa were not used because they have a much reduced genome size. Both of these would be poor estimators of what the genome was like at the reassignment point. In all these species, πG is much less than πA, but the bias is more extreme in some species than others. In case ix, if either of the Tetrahymena species is used as proxy, a fairly high Pdis is obtained, but if Paramecium is used, Pdis is much lower. Similarly in case iii, Pdis is high if we use Schizophyllum, but low if we use Crinipellis. These two species differ in base frequencies and also in the number of genes on the genome (i.e., Nstop). Although Nstop = 89 in Crinipellis (also known as Moniliophthora perniciosa), this genome has many genes labeled as “hypothetical protein” that do not have homologues in related species. These genes may be recent insertions or may not even be expressed sequences. Thus, Nstop was probably much less than 89 at the time of the reassignment, and Pdis would be correspondingly higher. In both these cases, it is reasonable to conclude that the properties of the genome at the point of reassignment were such that CD was not too unlikely. Table 2.Estimates of stop codon disappearance probabilitiesCaseSpeciesNstop%G%Afcodon (%)PdisUGA CDiMonosiga brevicolis322.8547.525.360.17iiAcanthamoeba castellanii409.8926.3521.446.4E-05iiiCrinipellis perniciosa899.1932.8417.942.3E-08iiiSchizophyllum commune203.4842.876.990.23ivPenicillium marneffei172.3835.645.890.36ivHypocrea jecorina194.6840.149.460.15vYarrowia lipolytica141.5549.012.970.65vCandida stellata80.9441.572.170.84viChondrus crispus256.0836.6112.463.6E-02viPorphyra purpurea3110.9835.8019.011.4E-03viiPedinomonas minor111.6021.316.520.48viiiEmiliana huxleyi219.2636.5016.832.1E-02ixParamecium aurelia4614.8719.6030.146.8E-08ixTetrahymena pyriformis443.2242.276.624.9E-02ixTetrahymena thermophila452.6340.465.757.0E-02xCafeteria roenbergensis326.1732.5613.746.6E-03UAG CDiRhizophidium373.3334.058.184.3E-02iSpizellomyces punctatus326.0330.6914.107.7E-03iiScenedesmus obliquus36.6435.7313.550.65 The only other case where Pdis is low is case ii, where Pdis = 6.4 × 10−5 if Acathamoeba is used as proxy. There is no other related species for which data are available. We have already seen that fluctuations in base frequencies can change Pdis by several orders of magnitude. Therefore a value of order 10−5 does not appear ridiculously small, and the true value could well have been much higher at the point the reassignment actually occurred. Finally, we comment on case 17 of Swire et al., Table 2a, where they estimate the low probability of Pdis = 2.7 × 10−11. This corresponds to our case vi, where we estimate Pdis = 3.6 × 10−2 if Chondrus is used as proxy and Pdis = 1.4 × 10−3 if Porphyra is used. Both of these are much higher than the estimate of Swire et al. We also note that, according to our phylogeny, these two species are related and we assume a single reassignment in their ancestor, whereas Swire et al. treat them as independent (cases 16 and 17). The change in the phylogeny would not affect our estimate of Pdis. Thus, we conclude that all these examples of UGA codon reassignment are consistent with the CD mechanism. (Cases xi and xii are not included in Table 2 because there are insufficient data to perform the calculation.) Reassignments of UAG Stop Codons Two cases of reassignment of UAG from Stop to Leu are known: one in the two chytrids, Rhyzophidium and Spizellomyces (Fig. 1; Laforest et al. 1997) and the other in Scenedesmus (Fig. 2; Hayashi-Ishimaru et al. 1996; Kück et al. 2000). In Table 1, UAG is also seen to be rare in general and rare particularly in the close relatives of the species reassigned: only two in the other chytrid, Hyaloraphidium; and zero in Pedinomonas, relative of Scenedesmus. These reassignments show the same pattern as the UGA Stop-to-Trp examples and can also be attributed to the CD mechanism. Disappearance of UAG would also be favored by mutation pressure increasing the AU content. The reason UAG is reassigned less frequently than UGA may be because of the relative difficulty of the required change in the tRNA. In the case of UGA, the existing tRNA-Trp can simply mutate its anticodon. In the UAG case it is necessary to create a new tRNA-Leu with a CUA anticodon to pair with UAG. This can be done by making a mutation in the second position of a tRNA-Leu with a CAA anticodon. Such a mutation in the anticodon would not change the amino acid specificity of the tRNA because the anticodon of the tRNA-Leu does not act as an identity element for recognition by the leucyl-tRNA synthetase (Asahara et al. 1993), in contrast to most other amino acyl-tRNA synthetases, which recognize the anticodon. However, this can only be done after duplication of the tRNA-Leu, because one copy of this gene is still required to translate UUG Leu codons. Thus, this reassignment is relatively rare because it requires a prior tRNA duplication. It is interesting to note that in most genomes, both UUA and UUG would be Leu, and there would be a single tRNA with UAA anticodon to translate both these. Another peculiarity of Scenedesmus is that UUA codons are not used (Kück et al. 2000), but UUG codons are frequent. This is clearly related to the fact that the tRNA-Leu has CAA anticodon in this species. UAG has been reassigned to Ala in a group of green algae that are closely related to Scenedesmus (Hayashi-Ishimaru et al. 1996). Complete genomes are not available for this group, therefore they are not included in our Fig. 2. It seems likely that UAG disappeared in the ancestor of Scenedesmus and the other algae, and that the codon was captured by Leu in Scenedesmus and by Ala in the other species. Table 2 also shows the probability of disappearance of the UAG codon for these two cases of UAG reassignment. This is calculated in the same way as for UGA above. These values do not appear unreasonably small in view of the uncertainty in the genome properties at the point of reassignment. Therefore, we conclude that these changes are consistent with the CD mechanism. Sense Codon Reassignments Linked to Codon Disappearance It is not only Stop codons that can be reassigned by the CD mechanism. The following examples show sense codon changes occurring via this mechanism. In the canonical code, both CUN and UUR code for Leu and these two codon blocks are accessible to one another by a synonymous mutation at the first position. Similarly CGN and AGR both code for Arg and are also accessible to one another via a synonymous mutation at first position. In AU-rich genomes, CUN and CGN codon families are subject to disappearance and replacement by UUR and AGR. Table 3 gives the number of occurrences of the codon groups for Leu and Arg in several Fungi. Base frequencies at FFD sites are also shown. C and G frequencies are very low in all the species in Table 3, indicating a strong mutation pressure toward A and U. Table 3.Codon usage in some Fungi lineages: intronic ORFs excluded (E) or included (I)Leu codonsArg codonsIle and Met CodonsFrequency at FFD sitesCUNUURCGNAGRAUUAUCAUAAUG%U%C%A%GS. japonicus (E)7919873213340324876.03.419.21.3S. octosporus (E)68236234161340d5761.11.934.82.2S. pombe (E)5319273311339495156.41.340.51.8Y. lipolytica (E)446180c751748727711948.31.149.01.5C. stellata (E)3279122912381565457.30.241.60.9C. albicans13239747261198122910055.44.837.72.1C. parapsilosis (E)6654739453033219311768.60.729.80.9C. parapsilosis (I)137728601024104929914365.33.029.02.7P. canadensis (E)2571418672741856210549.20.749.40.7P. canadensis (I)2774620742982058610950.00.948.40.6A. gossypii80a2910c40215795d,e3457.20.042.80.0K. lactis (E)0b2860c48213167d6344.01.653.31.1K. lactis (I)0b3120c552561627d6543.42.652.11.9K. thermotolerans (E)16a304244204172d5647.70.951.00.4K. thermotolerans (I)42a44010722982330d7848.32.147.62.0C. glabrata (E)11a2941c452072116d,e7346.70.652.00.7C. glabrata (I)28a4151c603182585d,e7848.60.949.80.7S. cerevisiae(E)33a3337492393160d,e7348.62.347.12.0S. castellii (E)19a2740c402037101d,e5647.51.649.91.0S. servazzii (E)22a3000c462181195d,e7037.20.959.52.4aCUN is reassigned to Thr and an unusual tRNA-Thr gene is present in these species.bThere is no tRNA for the CUN block in this species.cThe tRNA-Arg is deleted in these species.dThe tRNA-Ile(CAU) is deleted in these species.eAUA is reassigned to Met in these species. The species in the top half of Table 3 (P. canadensis and above) use CUN for Leu, as in the canonical code. These species possess a tRNA-Leu with anticodon UAG for this family. In all these cases the CUN family is less frequent than the UUR family. The six species followed by a superscript a in Table 3 have undergone a reassignment of the CUN family to Thr. These species possess an unusual tRNA-Thr with anticodon UAG (Sibler et al. 1981; Osawa et al. 1990) and the usual tRNA-Leu(UAG) is not found in the genome. In K. lactis, the codon family is not used at all. Significantly, this is the only species in which no tRNA exists that pairs with this codon family. The fact that the tRNA-Thr(UAG) appears only in species where the tRNA-Leu(UAG) is absent suggests that the new gene evolved directly from the old tRNA-Leu(UAG) gene. This requires no change in the anticodon, but changes elsewhere are required in order that the tRNA is recognized by the threonyl-tRNA synthetase and not the leucyl-tRNA synthase. Although many amino acyl-tRNA synthetases recognize the anticodon, recognition of tRNA-Leu by leucyl tRNA-synthetase is exclusively determined by the bases in the large variable arm (Asahara et.al. 1993). Changes in the variable arm could have prevented the tRNA acting as a tRNA-Leu and would have left it open to evolving a new function as a tRNA-Thr. The codon usage figures in Table 3 indicate that the change is attributable to the CD mechanism. The number of occurrences of CUN was most likely driven to zero prior to the branching of A. gossypii (see Fig. 1). A major change in the tRNA-Leu(UAG) gene was then possible, allowing it to be charged with Thr. CUN codons later reappeared with a new meaning. In the context of our gain-loss framework (Sengupta and Higgs 2005), this example is interesting in that the gain and loss of function occur in the same tRNA. Although it is likely that the unusual tRNA-Thr in these species evolved directly from the old tRNA-Leu, the alternative scenario is that there was a duplication of the tRNA-Thr(UGU) gene that decodes the normal ACN Thr codons. One of these genes might then have changed its anticodon to UAG (two mutations required). In order to determine the origin of the tRNA-Thr(UAG) gene, we constructed a phylogeny of all tRNAs from P. canadensis, K. thermotolerans and S. castellii, i.e., from representative species before and after the codon reassignment. The tRNAs of each type formed monophyletic triplets, with the exception of the genes with UAG anticodons. The tRNA-Thr(UAG) genes in K. thermotolerans and S. castellii showed no close relationship to either the tRNA-Thr(UGU) genes or the tRNA-Leu(UAG) gene in P. canadensis. Thus, neither of the two possible scenarios was supported. Despite this, it stills seems most likely that the new tRNA-Thr gene evolved from the old tRNA-Leu, a conclusion also reached by Sibler et al. (1981) and Osawa et al. (1990). One possibility that cannot entirely be ruled out from the codon usage data is that the changes in the tRNA-Leu(UAG) occurred when the CUN codons were very rare but not entirely absent. The changes might have been such as to immediately cause the tRNA to be charged by Thr, or might have happened more gradually, so that the same tRNA could be ambiguously charged by Leu and Thr during the changeover period. We would then have to count this as an example of the AI mechanism. Nevertheless, the low CUN number resulting from mutation pressure is clearly a major factor in this case, and we therefore feel comfortable in classifying it under the CD mechanism. The tRNA gene has undergone considerable modification, including an unusual insertion that makes the anticodon loop larger than the standard seven bases. In our opinion, it is unlikely that such a large change could have happened while the tRNA remained simultaneously functional for both amino acids. Table 3 also shows disappearances in the Arg codons that happen in parallel with those in the Leu codons. Mutation pressure away from C causes replacement of CGN codons by AGR. Once CGN codons have disappeared, there is no penalty to the deletion of the tRNA-Arg(ACG) gene. This gene is absent in the species followed by a superscript c in Table 3, and CGN is absent in these species (or apparently there is just one in C. glabrata). On the other hand, the gene is still present in S. cerevisiae and K. thermotolerans, and a handful of CGN codons still remain in these species. The tRNA-Arg(ACG) genes in S. cerevisiae and K. thermotolerans show clear sequence homology to that in P. canadensis. From the phylogeny of these species in Fig. 1, we see that it requires four independent deletions of the tRNA-Arg(ACG) to explain the observed pattern of absences of the gene, indicating that deletion of redundant genes is a rapid phenomenon. Figure 1 shows several separate reassignments of CGN from Arg to unassigned and only one reassignment of CUN from Leu to Thr. This is because the symbols label points where the tRNAs were deleted or mutated, not points where the codon disappeared. Our interpretation is that both CUN and CGN codons were absent in the ancestor of the group of yeast species prior to the branching of A. gossypii, and that change in the tRNA-Leu occurred at this point, whereas the deletions of the tRNA-Arg genes did not occur till after the split between the species. This would permit small numbers of CGN codons to reappear in some lineages where the gene was not deleted (e.g., S. cerevisiae and K. thermotolerans). The final twist in this story is that the CUN codons are absent in K. lactis. This means that there must have been a further reassignment of CUN from Thr to unassigned in this species only. Clearly the codons have disappeared in this case, which could be due to drift, or because of poor functioning of the tRNA-Thr(UAG) gene in this species, which would have caused selection against this codon block. We have little evidence in this case, but for the purposes of our summary table in the Discussion below (Table 6), we classified it as CD. Table 3 shows another example of the disappearance of the CGN block in Y. lipolytica. The corresponding tRNA-Arg is also deleted in this species (Kerscher et al. 2001). In S. pombe and C. stellata, the species that branch before and after Y. lipolytica, the tRNA-Arg is still present and the CGN codons are still used. CGN has thus become unassigned in this single species. This also appears to be a result of mutation pressure causing disappearance of CGN before deletion of the tRNA-Arg. The Probability of Disappearance of CUN and CGN Codons The number of Leu and Arg codons in these genomes is much larger than the number of stops. Therefore, it is more difficult for these sense codons to disappear than it is for stops. Nevertheless, the probability of CD is not unreasonably low in these cases. If we assume that all six Leu codons are in equilibrium, we find from which the sum of the relative frequencies of all the CUN block codons is The probability of disappearance of the CUN block is , where NLeu is the number of Leu codons in the genome. However, Table 3 shows that this calculation is hardly necessary. In A. gossypii, the species branching immediately after the proposed point of the codon reassignment, %C = 0 at FFD sites, i.e., the mutational bias against C is so strong that there are no C bases at all in any FFD site. Thus fCUN = 0, and Pdis = 1. Presumably the chance of mutating to a C is not precisely zero in A. gossypii, but it is small enough that all the FFD C-ending codons disappear. Therefore it is not unreasonable that the CUN block should have disappeared in the ancestor of this group. As we do not know that %C was exactly zero in the ancestor, we would like to know how small it must have been in order for there to be a significant probability of disappearance of CUN. It can be seen that %C is low in all species in Table 3, but variations of ±1% make a big difference in estimation of Pdis. Estimation of base frequencies is further complicated by the presence of intronic ORFs in many of these species. It has previously been shown that codon usage patterns differ between the intronic ORFs and the other genes (Bullerwell et al. 2003; Talla et al. 2005). For several species in Table 3, we have listed separate codon information for the genes excluding the intronic ORFs (E) and for the full set of genes including the intronic ORFs (I). In every case considered, %C is slightly higher if the intronic ORFs are included. These ORFs vary in number between species, and are not present in some species, including A. gossypii. This makes it difficult to know the number of genes in the ancestral genome at the point of CUN reassignment. It also suggests that these sequences may be recent insertions and that the base composition may not have adjusted to the equilibrium base composition of the rest of the genome. Furthermore, the presence of rare codons in the intronic ORFs suggests that these sequences may not be expressed at a high level or that their expression may not be important for the organism. For example, in K. lactis and K. thermotolerans, AUA is avoided because of deletion of the corresponding tRNA (to be discussed below), but there are substantially more AUA codons in the intronic ORFs than the regular genes. For these reasons, it is likely that the base frequencies in the genes excluding the intronic ORFs are better indicators of the equilibrium base frequencies of the mutational process affecting the most important genes in the genome. We also note that the presence of the intronic ORFs is sometimes associated with alternative splicing, so that the same exon is used in several proteins (e.g., S. cerevisiae and C. stellata). Clearly we do not want to count the same exon twice when counting codons. A full investigation of all these effects on intronic ORFs is beyond the scope of this paper. Given all the caveats above, it is clear that any estimate of Pdis is subject to a very large uncertainty. However, we wish to give a fair estimate for a case where %C is not exactly zero. We use the figures for C. glabrata because the base frequencies with and without the intronic ORFs differ less than in most species. To be conservative, we use the figures including the ORFs. From Table 3, we obtain fCUN = 0.0354, and Pdis = (0.9646)415 = 3.2 × 10−7. Slight reductions in %C would increase this figure by orders of magnitude. We conclude the bias against C must have been at least as strong as it is in C. glabrata (i.e., %C < 0.9) in order for the CUN codons to have a reasonable chance of disappearing. The fact that %C becomes exactly zero in A. gossypii and is less than 0.9% in several other cases in Table 3 shows that these extreme biases do occur. Swire et al. (2005) have calculated many examples of CD probabilities. The CUN reassignment corresponds to case 3 of Swire et al. (2005, Table 2b). For some reason, they estimated a vanishingly small probability of disappearance of the CUN codons (Pdis = 6.8 × 10−427), which is the lowest of any of the cases they list. They thus argued that CD was impossible in this case. We disagree, since we have shown that Pdis is high due to the extreme bias against C bases. This case is exceptional, however. In all the other examples of sense-to-sense codon reassignments discussed by Swire et al., we agree that the probability of CD is indeed unrealistically low and that the reassignment must have occurred without CD. Using an argument similar to that above, the relative frequency of CGN codons is , and the probability of disappearance of the CGN block is . We have already seen that %C = 0 in A. gossypii and is very low in related species. NArg is much lower than NLeu, so disappearance of CGN in the ancestor of this group is therefore quite likely, and is easier than disappearance of CUN. For the case of Y. lipolytica, if we use the figures from this species as a proxy for the ancestor, we obtain Pdis = (1 – 0.0426)75 = 0.038, which is not unreasonably small. Clearly it is much more difficult for sense codons to disappear than stop codons, as the number of codons involved is much larger. For this reason, we need an extremely strong base frequency bias in order to cause disappearance. However, in the above cases we have shown that the bias is indeed extreme and that the CD mechanism seems likely. A reviewer of this paper pointed out that Osawa and Jukes (1989), the original proponents of the CD model, insisted that, according to their theory, every single occurrence of the codon must disappear from the genome before the reassignment occurred. From current genome data it will never be possible to tell whether every occurrence of the codon disappeared at the time the reassignment occurred. Therefore, if we insist on this strict definition, it is impossible to assign a mechanism to any of the reassignment events. However, to be more constructive and more practical, we wish to make a clear distinction between the examples discussed above and those that we discuss in the remainder of the paper. In the above examples, there is a clear case that CD is likely due to the direction of the mutation pressure and the best estimates of the CD probabilities are not too low. The extreme rarity (and probably absence) of these codons was clearly a major factor in allowing the codon reassignment to occur. In contrast, in the examples below, there is clear evidence that the codon was not rare at the time of the reassignment and that the codon reassignment was initiated by some other event not related to CD. Reassignments That Cannot Be Explained by Codon Disappearance Reassignment of AUA from Ile to Met Is Initiated by Loss of a tRNA-Ile Gene In the canonical code, the AUN codons are unusual in that AUU, AUC, and AUA code for Ile, and only AUG codes for Met. The tRNAs need to distinguish between codons ending in A and those ending in G, which is not usually necessary (except for the UGG Trp codon). The translation system in bacteria uses two tRNAs for Ile: one with anticodon GAU, which translates AUY codons, and one with anticodon CAU, which translates AUA only. It is known from studies on E. coli (Muramatsu et al. 1988) that the C at the wobble position undergoes a posttranscriptional base modification to K2C (lysidine), which allows this tRNA-Ile to pair with the AUA codon. The same gene also is present in many mitochondrial genomes in which AUA is translated as Ile, and it is presumed that the same base modification occurs. However, the tRNA-Ile(CAU) gene is absent in mitochondrial genomes in which AUA is reassigned to Met. Before codon reassignment, tRNA-Met also has a CAU anticodon, which pairs with AUG only, but modification of the wobble-position C allows it to pair with both AUA and AUG (Tomita et al. 1999b). In order to locate and understand the cases of AUA codon reassignment in mitochondria, it is crucial to establish which genomes contain the tRNA-Ile(CAU) gene. This is complicated by the fact that some genomes contain three tRNAs with CAU anticodons (a Met initiator, a Met elongator. and an Ile), and the Ile tRNA is often mistakenly annotated as Met. To uncover misannotations, we constructed a phylogeny of all the tRNAs with CAU anticodon from all the fungi genomes in our data set. The genes fell into three groups that could be reliably identified. Supplementary Table S3 lists the position of the three tRNAs in each genome and, hence, shows in which genomes there have been gene deletions. From this, we note that the tRNA-Ile is missing from the seven species from A. gossypii downward in Table 3. Hence, this gene must have been deleted prior to the branching of A. gossypii, as shown in Fig. 1. Before the deletion of the tRNA-Ile(CAU) gene, AUA and AUU are frequent codons for Ile but AUC is rare (e.g., C. albicans, C. parapsiplopsis, and P. canadensis). This is to be expected from the FFD base frequencies. Mutation pressure is therefore in the wrong direction to cause disappearance of AUA, and in fact AUA is the most frequent Ile codon in P. canadensis, the closest relative to the species concerned. We therefore argue that this reassignment was initiated by loss of the tRNA-Ile while the AUA codon was frequent in the genome. Hence, it is an example of the UC mechanism, according to our classification. Immediately after the loss of this tRNA, the AUA codon had no specific tRNA to pair with it, and it may be said to have been unassigned. The tRNA loss could not have been lethal, however, so there must have been another way of decoding AUAs. In Metazoa, some genomes are known where the tRNA-Ile(CAU) gene has been deleted, but the AUA codon has not been reassigned. Yokobori et al. (2001) have shown that in such cases, AUA is translated as Ile by the tRNA-Ile(GAU), even though this requires GA mispairing at the wobble position. It is presumed that this is tolerated if there is no better alternative available, i.e., if the tRNA-Ile(CAU) has been deleted but no modification to the tRNA-Met has occurred. This may be what is happening in the two Kluyveromyces species, where usage of AUA is very low, despite the fact that A is also frequent at FFD sites in these species. This demonstrates a definite avoidance of AUA, probably due to the fact that there is no tRNA that translates the AUA codon efficiently. Talla et al. (2005) argue that AUA is assigned to Met in K. thermotolerans based on only two occurrences of this codon in the VAR1 gene. It may be more reasonable to say that AUA is unassigned in both K. lactis and K. thermotolerans. In C. glabrata, Koszul et al. (2003) conclude that AUA is associated with Met based on comparison of C. glabrata sequences with S. cerevisiae sequences from which they find that 8 of 16 AUA codons in C. glabrata appear at conserved Met sites encoded by AUA in S. cerevisiae. There is still some avoidance of AUA codons in C. glabrata nevertheless. In A. gossypii and the Saccharomyces species, AUA is frequent, and we presume that it is efficiently translated as Met. This may imply that there have been two independent adaptations of the tRNA-Met to deal with AUA in these two lineages. We wish to underline the difference between the AUA reassignment and the reassignments of CUN and CGN in the same species (see previous section). For CUN and CGN, mutation pressure drives the disappearance. In both these cases there is only one tRNA for the block of four codons, and if this were deleted (in the CGN case) or drastically mutated (as probably happened in the CUN case), there would be no plausible alternative tRNA that could step in. Hence these changes can only occur after the codon disappears. In contrast, mutation pressure leads to frequent usage of AUA for Ile. Loss of the tRNA-Ile(CAU) is possible without CD because the tRNA-Ile(GAU), and possibly also the tRNA-Met(CAU), has some ability to pair with this codon. A calculation of the AUA CD probability also serves to highlight the difference with the previous section. If the three Ile codons are in equilibrium, , and . Using C. glabrata (I) as an example, as we did for the CUN disappearance, we obtain Pdis = (1−0.502)343 = 1.4 × 10−104, which is indeed vanishingly low. As described in the previous paragraph, the adaptation to using AUA for Met seems to have been gradual. A possible reason for this could be that the change in the tRNA-Met is due to a base modification of the wobble position C, not a mutation (as in the tRNA-Trp, where the wobble position C mutates to U when the UGA codon is reassigned to Trp). The difference between the avoidance of AUA seen in Kluyveromyces and the fairly frequent use of AUA as Met in Saccharomyces may lie in the fraction of tRNAs in which this base gets modified. The other species of fungus in which tRNA-Ile(CAU) is missing is Schizosaccharomyces octosporus, although the gene is still present in S. pombe and S. japonicus. Table 3 shows that AUA is absent altogether in S. octosporus (excluding intronic ORFs), whereas comparison with the other two species suggests that there were 30 or 40 AUA codons present before the loss of the tRNA. Once again, it is unlikely that these codons would have disappeared by chance, and A is a frequent base at FFD sites in all three Schizosaccharomyces species. Hence the disappearance of AUA in S. octosporus is due to selection against this codon after the loss of the tRNA. Bullerwell et al. (2003) show that AUA has not disappeared entirely in the intronic ORFs of S. octosporus. This suggests that there is weaker selection on codon usage in the intronic ORFs because these genes are less strongly expressed than the standard genes. It also confirms that selection must have driven the disappearance of AUA in the standard genes, because if mutation pressure were responsible, it would have affected the intronic ORFs equally. Thus, we conclude that the initial event must have been the deletion of the tRNA-Ile(CAU), and hence this corresponds to the UC mechanism. In this case, the codon has not been reassigned to Met, but there is the potential for this to occur in the future if there is a gain of function in the tRNA-Met. From Supplementary Table S3, we see that C. albicans, C. parapsilosis, and Y. lipolytica have only one tRNA-Met. Possibly this plays the role of both initiator and elongator. We do not consider it further because this deletion is not associated with any codon reassignment. AUA is also reassigned to Met in some Metazoa. The tRNA-Ile(CAU) gene is present in close relatives of the Metazoa, Amoebidium, and Monosiga (Burger et al. 2003) and, also, in Porifera (Lavrov et al. 2005). AUA is Ile in these species. The gene is deleted in all remaining Metazoa (see Fig. 4), which suggests that it was lost after the divergence of the phylum Porifera from other metazoan phyla. Cnidaria are unusual in that they have lost almost all tRNAs from the mitochondrial genome and require import of tRNAs from the nucleus. AUA remains Ile in this group, presumably because the tRNA-Ile(CAU) is imported. There are also two other groups in which AUA remains as Ile: Platyhelminthes, or more specifically Rhabditophora (Telford et al. 2000), and Echinoderms/Hemichordates. These phyla retain sufficient tRNAs to translate the full code and there is no suggestion of tRNA import. It therefore appears that AUA is translated as Ile by the tRNA-Ile(GAU). This is consistent with the argument of Yokobori et al. (2001) that GA pairing at the anticodon wobble position is tolerated if there is no better alternative anticodon-codon pairing solution available. In the remaining metazoan phyla, AUA is reassigned to Met. The usual tRNA-Met anticodon is CAU. This can be modified in several different ways to pair with AUA in addition to AUG. In Drosophila, modification from C to 5-formylcytidine (f5C) at the wobble position occurs (Tomita et al. 1999b). These authors also show that the unmodified CAU anticodon can translate both codons if the base in the 37th position of the tRNA is t6A (N6-threonylcarbamoyladenosine) instead of A (see also Kuchino et al. 1987). Note that the 37th position is the nucleotide immediately 3’ of the anticodon. In urochordates, the tRNA-Met has a U*AU anticodon (where U* is an unknown derivative of U). This is a third means by which the tRNA-Met has gained the ability to translate AUA. There is insufficient evidence to say exactly where these modifications to the tRNA-Met evolved, and there are many species for which the details of the tRNA-Met are not known. Nevertheless, the existence of these different modifications to the tRNA-Met suggests that the reassignment of AUA to Met has occurred independently at least three times in different phyla after the loss of the original tRNA-Ile(CAU) gene (as shown in Fig. 4). Note that the AUA codon is common in Porifera and Cnidaria, and we did not find any example of disappearance of this codon in any of the Metazoa (see Table 4). Hence, there is no suggestion that the codon disappeared before the loss of the tRNA-Ile(CAU). As the tRNA loss occurred first, this is an example of the UC mechanism, very similar to the case of AUA reassignment in Fungi. Table 4.Codon usage in some metazoan lineagesIle and Met codonsSer and Arg codonsAsn and Lys codonsFrequency at FFD sitesAUUAUCAUAAUGAGUAGCAGAAGGAAUAACAAAAAG%U%C%A%GAxinella corrugate222452051307423 35b1710714 862835.911.730.621.8Geodia neptuni207182181187915 55b1510418 892440.77.135.716.4Metridium senile190591101385821 43b128431 792242.517.525.015.0Acropora tenuis182441151136811 34b169015 723143.511.517.627.4Limulus polyphemus241107171a43228 661210344 661935.717.138.88.4Daphnia pulex18793 90a524920 68 1e7543 484136.220.224.918.7Drosophila melanogaster35516216a13300 74 0e19310 81550.52.443.63.6Caenorhabditis elegans25723134a446161263913912 951450.24.137.58.2Trichinella spriralis136114193a964620 71346880 653429.615.242.213.0Katharina tunicate22649125a454636 883513447 793043.112.332.012.5Lumbricus terrestris194105194a642014 63137066 702228.224.336.910.5Terebratulina retusa137147152a34623 78184667 801922.635.334.47.7Fasciola hepatica1198 3996928 741502 18f4468.25.46.919.4Schistosoma mansoni1341515512310210 6250544 39f6052.84.323.319.5Taenia crassiceps17421261011041 4329895 59f4964.31.419.914.5Paracentrotus lividus146531651021114 66154450 88f5425.022.942.69.6Asterina pectinifera12966178782223 54194058110f4828.825.135.410.6Balanoglossus carnosus97148 69771835 19 0e2098 04522.437.829.110.6Saccoglossus kowalewskii18572 96703422 4 0e6068 471240.723.031.25.1Halocynthia roretzi17012112a601254 68c88785 294253.03.626.117.2Cionia intestinalis29719237a34737221c37143161032256.54.833.35.4Branchiostoma lanceolatum18647139a637830 12 0e8626 403437.79.132.820.5Branchiostoma floridae18746138a637333 12 0e8724 403338.58.533.020.0Epigonichthys maldivensis15970122a594935 31 2e7040 373534.713.629.422.3Myxine glutinosa210154182a442328 2d 169771021529.227.434.98.5Homo sapiens124196167a401439 1d 132132 851014.540.438.76.4aAUA is Met in these species.bAGR remains Arg.cAGR is Gly.dAGR is Stop.eAGG is avoided, probably because there is an unmodified G in the wobble position of tRNA-Ser(GCU).fAAA is reassigned to Asn. We interpret the decoding of AUA as Ile in Platyhelminths and Echinoderms/Hemichordates as indicating that an appropriate change in the tRNA-Met did not occur, and that AUA has always been Ile in these lineages. An alternative is that AUA was assigned to Met once at the point of the loss of the tRNA-Ile(CAU) and that there were reversals of the change in these two groups (as shown in Fig. 2 of Knight et al. [2001a] and Fig. 1 of Yokobori et al. [2001]). However, this scenario does not explain why more than one type of modification of the tRNA-Met should exist. Also, once the AUA codon is captured by Met and mutations have occurred throughout the genome (so that AUA appears in positions where a Met rather than an Ile is required), then the organism is better adapted than it would be if AUA were inefficiently translated by the tRNA-Ile(GAU). Thus we see no reason why a reversal of this change should occur. Telford et al. (2000) have also discussed these two alternative scenarios in a parsimony treatment, however, they did not consider the position of the tRNA-Ile(CAU) gene loss or the fact that there are several independent modifications to the tRNA-Met, both of which are essential points in our interpretation. Reassignment of AUA from Ile to Met has also been observed (Ehara et al. 1997) in some species of yellow-green algae (Xanthophytes). The entire genome of these species is not yet available and the lack of information about the tRNA genes and codon usage makes it impossible to deduce the mechanism of change. However, it is clear that the change occurred independently of similar changes observed in Metazoa and Fungi. Reassignments Involving the AGR Block Are Initiated by Loss of a tRNA-Arg Gene The AGR block, assigned to Arg in the canonical code, has undergone multiple reassignments in Metazoa. The tRNA-Arg(UCU) is present in Porifera but is absent in all other Metazoa. In Cnidaria, almost all tRNAs are missing, and tRNA import from the cytoplasm is required. This is not associated with codon reassignment. In the remaining Metazoa (i.e., Bilateria) the loss of the tRNA-Arg(UCU) gives rise to subsequent changes in the genetic code. Loss of this gene leaves the tRNA-Ser(GCU) as the only candidate for translating these codons. Hence, AGR is reassigned to Ser at the base of the Bilateria (Fig. 4). This is classed as the UC mechanism, as it is initiated by a tRNA loss. Note that AGR codons are frequent in Porifera and Cnidaria; therefore there is no suggestion of CD (Table 4). The GCU anticodon of tRNA-Ser usually interacts with only AGY codons prior to the loss of the tRNA-Arg. However, pairing of the tRNA-Ser with AGR codons is also possible to some extent, once there is no longer any competition from the tRNA-Arg (Yokobori et al. 2001; Yokoyama et al. 1995). In Drosophila it is known that the tRNA-Ser(GCU) translates AGA (Tomita et al. 1999), i.e., there is a GA mispairing at the wobble position in the same way as in the mispairing of AUA with tRNA-Ile(GAU) after the loss of the tRNA-Ile(CAU). In Drosophila, AGG is absent, which may indicate that this codon is avoided because of selection against an unfavorable GG mismatch that would occur at the wobble position. It should also be noted that %G and %C are very low in Drosophila (Table 4), so the absence of AGG may simply be a result of mutation pressure. In Daphnia there is only one AGG codon, even though %G is high. This suggests definite avoidance of this codon. However, not all arthropods show this pattern. As an example, there are 12 AGG codons remaining in Limulus. Two other groups in which AGG is absent or very rare are the hemichordates and cephalochordates. Here the tRNA also has a G at the wobble position, which we presume is unmodified. In contrast, Table 4 shows many other cases where AGA and AGG are both frequent. In several invertebrate groups, the G wobble position is mutated to a U, which allows it to pair with all the AGN codons. This is the case with C. elegans and T. spiralis in Table 4, where we see relatively high usage of AGG. Alternatively, in some echinoderms (Matsuyama et al. 1998) and mollusks (Tomita et al. 1998) a base modification from G to m7G also permits it to decode the entire AGN block as Ser. The AGR block has undergone further changes in urochordates, where it is reassigned to Gly, and craniates, where it is reassigned to Stop. The most likely state prior to these changes would be that AGA was moderately frequent and AGG was rare or absent (as with the current hemichordates and cephalochordates). Urochordates contain an additional tRNA-Gly with anticodon UCU (Gissi et al. 2004). This has arisen by duplication of the standard tRNA-Gly(UCC), followed by anticodon mutation. This new tRNA pairs efficiently with AGR codons and outcompetes the tRNA-Ser(GCU). AGR codons become very frequent after the reassignment (Table 4) because mutation from GGR glycine codons occurs. Note that %A is greater than %G in all these genomes, so mutation pressure favors synonymous mutations from GGR to AGR. Since AGA was a moderately frequent codon for Ser prior to the reassignment, and since %A is high, we cannot argue that mutation pressure caused the disappearance of AGA and replacement by AGR prior to the change. Thus, it seems unlikely that CD is responsible for this change. This argument also applies for the reassignment of AGR to Stop in the craniates. AGA is rare in craniates after the change because Stop codons are always rare. These genomes have very high %A, so unless there was a short period when mutation pressure was in the opposite direction to that seen in all the current genomes, then we must suppose that this reassignment occurred without disappearance of the codon. This is puzzling since it implies premature termination of translation of genes containing AGA Ser codons. Spruyt et al. (1998) argued that AGA is also read as Gly in Branchiostoma lanceolatum. They found a putative tRNA with anticodon UCU that may have arisen by duplication and mutation of the tRNA-Gly(UCC), as occurred in the urochordates. However, the codon usage numbers suggest that this is not the case and that the sequence identified does not function as a tRNA-Gly. The pattern seen for AGA and AGG in B. lanceolatum resembles that in the hemichordates and some of the arthropods, where there is a single tRNA-Ser(GCU). Also, other cephalochordate genomes are now available, including B. floridae (Boore et al. 1999) and Epigonichthys maldivensis (Nohara et al. 2005), which also show no evidence of reassignment of AGA to Gly. The distinction between B. lanceolatum and B. floridae (e.g., Knight et al. 2001a) therefore seems unfounded. Reassignment of AAA from Lys to Asn May Proceed Via an Ambiguous Intermediate The reassignment of AAA from Lys to Asn occurred independently in echinoderms and platyhelminthes. Furthermore, the AAA codon is absent in the hemichordate Balanoglossus (see Table 4) but present in the related species, Saccoglossus, where it is translated as Lys, as in the canonical code (M.J. Smith, personal communication). Most metazoans in which AAA retains its standard assignment of Lys have one tRNA-Lys(UUU), which pairs with AAR codons, and one tRNA-Asn(GUU), which pairs with AAY codons. Mispairing between a G in the anticodon wobble position and an A-ending codon is implicated in both the reassignment of AUA from Ile to Met and the reassignment of AGR from Arg to Ser, as discussed above. In the case of the tRNA-Asn(GUU), mispairing with AAA is prevented by posttranscriptional modification of the wobble position G to queuosine (Q), which allows pairing with U and C but inhibits pairing with A (Morris et al. 1999). In echinoderms and platyhelminthes, the tRNA-Lys anticodon is mutated to CUU (Tomita et al. 1999a), which pairs only with AAG. Several changes to the tRNA-Asn occur that increase its ability to pair with AAA. The modification of G to Q does not take place (Yokobori et al. 2001), which may allow some degree of GA mispairing. In echinoderms, modification of the second anticodon position from U to Psi (pseudouridine) enhances the ability of the GPsiU-Asn anticodon to recognize the AAA codon. Also in some echinoderms, the mutation of the base immediately adjacent to the 5’ end of the anticodon from a U to a C seems to help the GUU anticodon to recognize AAA (Castrasena et al. 1998). In Balanoglossus the tRNA-Lys has also changed to CUU, but there seem to be no modifications in the tRNA-Asn that would allow AAA to be reassigned to Asn. AAA is therefore unassigned. In all groups in which AAA is still Lys, it is a relatively frequent codon and it is usually more frequent than AAG. %A is usually quite high at FFD sites. Thus, even though Castrasena et al. 1998 argue for the CD mechanism, it seems unlikely to us that AAA disappeared in either the ancestor of the platyhelminthes or the echinoderms. Note that the absence of AAA in Balanoglossus is not shared with Saccoglossus, so we cannot use this to argue that AAA disappeared in the common ancestor of echinoderms and hemichordates. The AI mechanism seems the most likely mechanism for the reassignments in platyhelminthes and echinoderms. The process would have begun by a gain in function of the tRNA-Asn that gave it some ability to pair with AAA, and hence made AAA ambiguous, e.g., the Q modification ceased to occur. Then further changes to the tRNA-Asn would have increased the fraction of AAA codons translated as Asn, and a straightforward mutation of the tRNA-Lys could then have removed this ambiguity. Alternatively, it is possible that the changes occurred via the UC mechanism. If the tRNA-Lys mutation happened first, this would leave the tRNA-Asn able to pair inefficiently with the AAA, and subsequent changes to the tRNA-Asn would allow it to recognize AAA more easily. This argument is analogous to the case of the AUA:Ile to Met change or the AGR:Arg to Ser. However, in those two cases the loss of function is the deletion of the original tRNA for the codon in question, which is irreversible. In the AAA case, the loss of function is just the mutation of U to C in the anticodon. This would be a deleterious mutation that could easily reverse, so it is difficult to see why the change would go to fixation. This makes the UC mechanism seem less plausible for AAA. We conclude, therefore, that of all the reassignments considered in this paper, these two examples of AAA reassignment are the best candidates for the AI mechanism. In the case of Balanoglossus, the AAA codon cannot disappear by mutation pressure, and therefore we require a selective reason why this codon should be absent. This can be attributed to either UC or AI mechanisms. If a gain of function of the tRNA-Asn(QUU) brought about by the loss of Q modification happened first, the resulting tRNA-Asn(GUU) would have acquired some ability to decode AAA as Asn and consequently AAA would be ambiguously translated. The negative selection against such ambiguous translation may have driven the AAA codon to disappear. Subsequently, the loss of function of tRNA-Lys(UUU) occurred, brought about by a mutation from U to C at the first anticodon position, and this removed the ambiguity in AAA decoding. Further experimental information on the state of modification of the tRNA-Asn in Balanoglossus would be useful to confirm this. Alternatively, the loss of function of tRNA-Lys that prevents it from translating AAA may have occurred first. In the gain-loss framework, this would be a loss without a gain. AAA would be unassigned, and selection against this codon would drive its usage down to zero. This is analogous to the situation in Schizosaccharomyces octosporus, where we argued that the AUA codon was driven to zero after the loss of the tRNA-Ile(CAU) gene. However, this case is again made less plausible by the fact that the change in the tRNA-Lys would be reversible, whereas the loss of the gene from S. octosporus is irreversible. Indeed, the current situation in Balanoglossus seems unstable. A mutation of the tRNA-Lys(CUU) back to UUU would be neutral in the absence of AAA codons. If this occurred, then synonymous mutations of AAG codons to AAA would rapidly occur (since %A is greater than %G). This would drive the system back to the canonical code. The fact that AAA remains absent in Balanoglossus suggests that there is some reason why the reversal of the tRNA-Lys back to UUU would not be neutral. One possibility may be that CUU pairs more efficiently than UUU with the AAG codons, so translational efficiency would favor retention of the CUU anticodon once AAG became the dominant codon for Lys. Another possibility is that unmodified wobble position U bases can often pair with all four bases in four codon families. For this reason, wobble position U’s are often modified when they occur in tRNAs for two-codon families only (Yokobori et al. 2001). Thus prior to the codon reassignment, the tRNA-Lys anticodon would be U*UU, where U* is a posttranscriptional modification of U that permits it to pair with A and G only. If the ability to modify the U were lost by Balanoglossus when the anticodon mutated to CUU, then a reversal of this mutation back to UUU would leave an unmodified U in the wobble position that could pair with both Asn and Lys codons. This would be selected against due to the ambiguity created in the Asn codons. Both these explanations are rather speculative. Introduction of New Stop Codons In addition to the AGR block, which becomes Stop in vertebrates, there are two further cases where a sense codon has been reassigned to Stop. According to the GenBank entry AF288091 (G. Burger, unpublished), UUA has been reassigned from Leu to Stop in the stramenopile Thraustochytrium aureum (Fig. 3). One tRNA-Leu gene in T. aureum has a CAA anticodon instead of the usual UAA anticodon, hence UUA is no longer translated as Leu. %A is high in this and related species, and UUA is frequently used as Leu in related species. This makes it unlikely that UUA disappeared due to mutation pressure. The mechanism cannot be assigned with certainty, but the most likely explanation is that UUA was driven to become rare after the mutation in the tRNA-Leu occurred because of inefficient pairing with the CAA anticodon. Gain of function of the release factor could then have occurred. This scenario corresponds to the UC mechanism (loss before gain). An alternative is that changes in the release factor began to occur in such a way that UUA became ambiguously translated as Leu and Stop. In this case, UUA would be selected against as a Leu codon and would be driven to low frequency, and selection would favor the mutation to C in the anticodon, so that UUA was no longer translated as Leu. This would correspond to the AI mechanism (gain before loss). It is difficult to decide between these scenarios. A second similar case is the reassignment of UCA from Ser to Stop in Scenedesmus obliquus (Kück et al. 2000). In most species, a tRNA-Ser(UGA) would decode UCN codons, but in S. obliquus, the tRNA-Ser has a GGA anticodon that can only pair with the UCY codons. UCA has become the standard Stop codon, used in the 13 standard mitochondrial protein genes. Alongside this, UCG is absent, presumably because it no longer pairs with the tRNA-Ser. The Stop codons UAA and UGA are not used, while UAG has been reassigned to Leu (as discussed above under ‘reassignments of UAG stop codons’). Table 5 shows that %A is fairly high in S. obliquus and related species, and that UCA is a fairly common codon in both P. minor and C. eugametos, where it retains its assignment to Ser (we consider C. reinhardtii separately in the following section). It seems unlikely that UCA would disappear due to mutational pressure. Here again, it is possible to explain the result in terms of a UC or an AI scenario. In the UC scenario, the tRNA-Ser anticodon would mutate to GGA, which would drive UCA and UCG to very low values and permit the release factor to change so that it interacts with UCA. In the AI scenario, the release factor would change first, making UCA ambiguous and causing it to be selected against and driven to low frequency, and allowing the mutation in the tRNA-Ser to occur. Table 5.Codon usage in some green algaeLeu codonsIleSer codonsThr codonsGlu codonsArg codonsFFD sitesUUAUUGAUAUCAUCGACAACGGAAGAGAGAAGG%A%GPedinomonas minor41184115538270581135221.31.6Scenedesmus obliquus038401301115731409335.76.6Chlamydomonas eugametos3533817680398653423048.511.7Chlamydomonas reinhardtii0251000000550026.13.7Note. Intronic and other nonstandard ORFs are excluded. Unusual selective effects influence S. obliquus and C. reinhardtii. The full codon usage table of S. obliquus is given by Kück et al. (2000) and Nedelcu et al. (2000). In Table 5 we have chosen selected codons in order to illustrate that S. obliquus is unusual in other ways. The usual Leu UUA codon is absent, and the tRNA-Leu has a CAA anticodon, which translates UUG only (the same as in T. aureum except that UUA is not used as a Stop codon). AGA is also absent, but there is a tRNA-Arg with CCU anticodon that translates AGG. Nedelcu et al. (2000) identified a tRNA-Arg(UCU) that should interact with AGA and a tRNA-Ile(UAU) that should interact with AUA. However, AGA and AUA are both absent. They state that these two tRNAs are redundant because their corresponding codons are not used. However, we argue that these two tRNAs must be nonfunctional or not expressed, otherwise the corresponding codons would be used. Note that %A > %G in S. obliquus, so if there were functional tRNAs with U at the wobble position, the A-ending codons would be frequent. As an illustration, the Glu codons in S. obliquus do follow the expected trend from the mutational frequencies: GAA is more frequent than GAG because these are both translated by a tRNA-Glu(UUC) with U at the wobble position. The Thr codons shown in Table 5 also have ACA more frequent than ACG, but one final peculiarity of the S. obliquus genome is that it possesses no tRNA-Thr. It is presumed that a tRNA-Thr(UGU) is imported from the cytoplasm. Importation of tRNAs from the Cytoplasm to the Mitochondria Although there are many mitochondrial genomes that possess sufficient tRNAs to translate the full genetic code, there are also many genomes in which this is not the case, and where import of nuclear-encoded tRNAs from the cytoplasm to the mitochondrion must be occurring. Supplementary Table S1 lists the amino acids for which tRNAs are present/absent in the mitochondrial genome of each species. In the figures, we have used a white arrow symbol to indicate positions where import of multiple tRNA genes must have evolved. We estimate at least nine independent origins of tRNA import: Acanthamoeba, Dictyostelium, Cnidaria, Chytridiomycota, land plants, Pedinomonas, Chlamydomonas, Naegleria+Alveolata, and kinetoplastids (the latter are not shown but are known to use tRNA import). This count depends on the details of the phylogeny: it is possible that Acanthamoeba and Dictyostelium have a common ancestor, which would decrease the count by one, and it is possible that Naegleria is not a sister group to the alveolates, which would increase the count by one. In all the above groups, at least four tRNAs are absent, and it is clear that the remaining tRNAs are insufficient, thus there is no doubt that import must occur. However, there are also many other groups where only a small number of tRNAs are missing. We have already discussed several cases above where the loss of a tRNA leads to the corresponding codon becoming unassigned or reassigned to a new amino acid. These cases leave clear signals in the codon usage patterns. This means that there is no import of a replacement tRNA from the cytoplasm in the cases discussed in previous sections. In contrast, we now discuss several cases where a small number of tRNAs are absent from the mitochondrial genomes but there is no change in the genetic code or unusual codon usage pattern. In these cases it appears that import of one or a few specific tRNAs is occurring. There is no tRNA for Thr in any of the stramenopile genomes. In addition, tRNA-Thr is absent from Reclinomonas, the Rhodophyta, Mesostigma, Scenedesmus, and all the groups where multiple tRNAs are missing. This makes tRNA-Thr the most frequently deleted tRNA. We consider this significant, but we have no plausible explanation why. The tRNA-Thr attracts attention because there are many genomes in which this is the only tRNA missing. Thr codons are nevertheless frequent in these genomes, so there must be a way of translating them. Import of tRNA-Thr from the cytoplasm is quite possible, but we do not know why this specific tRNA should be imported. Alternatively, in the genomes where tRNA-Thr is one of the only ones missing, it is possible that another tRNA is post-transcriptionally edited to become a tRNA-Thr. Two types of tRNA would thus arise from the same gene. This point has been suggested previously (Saks et al. 1998), but as far as we are aware, there is no direct evidence that this occurs. There is no tRNA for the Arg CGN block in several of the stramenopiles (Cafeteria, Thraustochytrium, Chrysodidymus, Ochromonas, Laminaria, Pylaiella) but it is present in Phytophthora and Saprolegnia. Complete genomes are not available for the remainder of the stramenopiles shown in Fig. 3, hence we cannot tell if the gene is present. If the phylogeny is correct, this implies three independent losses of the tRNA-Arg. Either tRNA import or some other tRNA change must be compensating. Recall that this same tRNA-Arg is also absent in several fungi and that this leads to the CGN block becoming unassigned (as shown in Fig. 1). So tRNA-Arg is not imported in the fungi. Supplementary Table S1 lists several other genomes where just one or two tRNAs are missing. It is possible that tRNA import occurs or that there are unknown tRNA modifications that compensate for these individual losses. However, we consider this as somewhat uncertain in cases where only one genome is known with the missing gene. This could simply be a failure to locate the gene on the genome or a problem of misannotation. The codon usage of Chlamydomonas reinhardtii is markedly different from C. eugametos (Table 5 and Denovan-Wright et al. 1998). Both of these species have only three tRNAs (Supplementary Table S1), hence tRNA import is required. C. eugametos uses almost all codons, and the codon usage is in line with what we would expect from the FFD site frequencies. However, there are many codons that are absent entirely from the C. reinhardtii genome. This suggests that a restricted set of tRNAs is being imported in C. reinhardtii, which causes selection against codons that cannot be translated, whereas a much broader set is being imported in C. eugametos, which allows a more standard pattern of codon usage. There are several parallels between the codon usage in C. reinhardtii and Scenedesmus obliquus (a sister species; see Fig. 2). UUA (Leu), AUA (Ile), UCG (Ser), and AGA (Arg) are all absent from both. In S. obliquus, this is due to peculiarities in the tRNAs encoded by the mitochondrial genome (see previous section). The correspondence between the codon usage and tRNA set in S. obliquus shows that there is no tRNA import in this species. It therefore appears that the tRNAs that were first imported in C. reinhardtii correspond to those that were needed according to the codon usage pattern of the ancestor of C. reinhardtii and S. obliquus. In C. eugametos, the evolution of tRNA import has proceeded further. A more complete set is imported and the codon usage has relaxed back to a more standard pattern due to mutation. Discussion Distinguishing the Mechanisms The gain-loss framework that we use to classify mechanisms of codon reassignment was introduced in discussion of our computer simulations (Sengupta and Higgs 2005). Here we wish consider to what extent this classification is applicable to the real events. The CD mechanism is defined by the fact that the codon disappears from the genome and is replaced by synonymous codons prior to any changes in the tRNAs and release factors. The reasons for the disappearance are random drift and directional mutation pressure. In simulations it is obvious whether a codon has disappeared. In real examples, there is clear evidence if the codon usage is zero or close to zero in the species close to the reassignment point, and there is also clear evidence if mutation pressure is in the right direction to cause CD. It is, of course, never possible to say that the codon usage was exactly zero in some genome in the past. A practical definition of the CD mechanism for real cases is that absence or extreme rarity of the codon is the major factor that initiates the codon reassignment. There should be evidence that the codon was rare and that the probability of CD was not too low. We have shown that this is true in the cases where we argue that the CD mechanism occurred. Therefore, the distinction of CD from the alternatives is one that we feel can be reliably made in real examples. In cases where we have determined that the codon did not disappear, the central question is to distinguish between UC and AI mechanisms. The UC mechanism is defined by the fact that the loss occurs before the gain, whereas in the AI mechanism, the gain occurs before the loss. Although this distinction is obvious in simulations, it can be more difficult in real cases, as we did not observe the order of events. Nevertheless, we gave several examples where we feel a reliable classification of UC can be made. In particular, the reassignments of AUA from Ile to Met and AGR from Arg to Ser are both associated with the deletion of a tRNA from the genome. The loss event is an irreversible gene deletion that leaves the organism in a deleterious state. The gain event is then positively selected in response to this loss. In contrast, in the case of the reassignment of AAA from Lys to Asn, the loss event is a mutation in the anticodon of the tRNA-Lys. This is reversible, so selection would favor mutation back to the original state, which seems more likely to occur than making a codon reassignment. The gain in this case is probably due to the cessation of the Q modification process, which could occur due to deleterious mutations in the modifying enzyme or disruption of the transport of the enzyme to the mitochondria. These changes seem less easily reversible than a single mutation in the anticodon. Therefore we argue that the gain occurred first and that the loss occurred in response. This makes the AAA reassignment the most likely example of the AI mechanism. The cases where both loss and gain seem easily reversible are the most difficult to classify, and both AI and UC scenarios can be proposed. (These arguments are based on the assumption that the ambiguous and unassigned states are deleterious with respect to the original code. We deal with the alternative suggestion that ambiguous translation can be positively selected later in the Discussion.) We just argued that the AUA Ile-to-Met case is initiated by deletion of the tRNA-Ile(CAU) and that this change can be reliably classed as UC. Nevertheless, it is possible that ambiguous translation plays a role in this reassignment in the following sense. After the deletion, the AUA codon would be translated inefficiently by the tRNA-Ile(GAU), but it is possible that there might also be some interaction with the tRNA-Met(CAU). When the tRNA-Met becomes modified, it definitely gains the ability to translate AUA. So, ambiguous translation of AUA as both Ile and Met at some points in this process is not unlikely. Before the original tRNA-Ile(CAU) is deleted, any slight ability of the other two tRNAs to translate AUA would be irrelevant. Also, after the modification of the tRNA-Met, any slight ability of the tRNA-Ile(GAU) to translate AUA would be irrelevant. The ambiguity, if it existed, would only be relevant when there is a competition between two poorly adapted tRNAs. Despite all this, the gain in function of the tRNA-Met only occurs after the loss of the tRNA-Ile(CAU), so this reassignment counts as UC not AI. We recommend the use of the term AI only when ambiguous translation occurs as a result of a gain occurring before a loss, and where competition occurs between two well-adapted tRNAs. The reassignment of AGR from Ser to Gly in urochordates involves a gain due to the duplication and anticodon mutation of the tRNA-Gly. In this case, too, it is plausible that an AI stage for AGR decoding may have resulted after the gain of the new tRNA-Gly. The ambiguity would eventually be removed due to the selection of the superior recognition of AGR by the new tRNA-Gly. However, this only happens after the deletion of the original tRNA-Arg(UCU). We have counted the changes from Arg to Ser and Ser to Gly as separate events because the amino acid changes twice. However, we may think of the duplication of the tRNA-Gly as the gain that finally compensates for the loss of the tRNA-Arg considerably earlier in evolution. The AGR:Ser to Gly case is one of the examples where the number of tRNAs increases (see Table 6 and discussion below), but this only returns the number back to its value in the canonical code; it does not represent an increase with respect to the canonical code. Table 6.Summary of mechanisms of codon reassignment in mitochondriaCodon reassignmentNo. of timesCan this be explained by GC→AU mutation pressure?Change in no. of tRNAsIs mispairing important?Reassignment mechanismUAG: Stop → Leu2Yes. G → A at 3rd position.+1NoCDUAG: Stop → Ala1Yes. G → A at 3rd position+1NoCDUGA: Stop → Trp12Yes. G → A at 2nd position.0Possibly. CA at 3rd position.CDCGU/CGC/CGA/CGC: Arg → unassigned5Yes. C → A at 1st position.–1NoCDCUU/CUC/CUA/CUG: Leu → Thr1Yes. C → U at 1st position.0NoCDCUU/CUC/CUA/CUG: Thr → unassigned1No–1NoCDAUA: Ile → Met or unassigned3/5aNo–1Yes. GA at 3rd position.UCAAA: Lys → Asn2No0Yes. GA at 3rd position.AIAAA: Lys → unassigned1No0Possibly. GA at 3rd position.UC or AIAGA/AGG: Arg → Ser1No–1Yes. GA at 3rd position.UCAGA/AGG: Ser → Stop1No0NoAIbAGA/AGG: Ser → Gly1No+1NoAIbUUA: Leu → Stop1No0NoUC or AIUCA: Ser → Stop1No0NoUC or AINote. CD, codon disappearance; UC, unassigned codon; AI, ambiguous intermediate.aThe loss of the tRNA-Ile has occurred three times (once in metazoa, once in yeasts, and once in S. octosporus) but the reassignment of the codon to Met has occurred at least five times (three times in metazoa and twice in yeasts).bIf AGR was fully reassigned to Ser prior to these changes, then they are probably AI. However, they may also be considered as a final response to the loss of the original tRNA-Arg, in which case we might count them as UC. A slow response to a tRNA deletion occurs for both the tRNA-Ile(CAU) and the tRNA-Arg (UCU), which are deleted at roughly the same time in the ancestral Bilateria. This would leave AUA inefficiently translated as Ile and AGR inefficiently translated as Ser. Both these situations remain in some phyla today. However, in other groups, AUA has been captured by Met, and AGR either has become a useful Ser codon (due to mutation of the tRNA-Ser) or has been reassigned to Gly or Stop. These secondary changes may have occurred considerably later than the deletions of the original two tRNAs. The fourth mechanism occurring in the gain-loss framework is the compensatory change mechanism. This would correspond to a case where the loss and gain become fixed in the population at the same time. Organisms with either the gain or the loss but not both would only occur as rare variants, and would never be frequent in the population. This mechanism is expected in the theory due to the analogy of the problem with compensatory mutation theory (Kimura 1985; Higgs 19982000). In the simulations of codon reassignment (Sengupta and Higgs 2005) it was possible to distinguish a few cases where this occurred. It may well be that the compensatory change mechanism occurs in real cases, and it could be an explanation in cases where both AI and UC seem equally plausible. However, it would be very difficult to distinguish this from either the UC or the AI mechanism after the event. Comparison with Previous Surveys Recent papers by Knight et al. (2001b) and Swire et al. (2005) have also carried out surveys of changes in the mitochondrial genetic code. Knight et al. (2001b) highlighted the fact that variation in GC content plays a major role in the CD mechanism. Most mitochondrial genomes are subject to mutation pressure from GC to AU. Since third codon positions are often synonymous, they argued that the CD mechanism would predict that the codons that are reassigned should have C or G at the third position. As this is not always the case, they argued against the CD mechanism. However, it is not the third codon position per se that matters; it is the position at which the synonymous mutation occurs. Table 6 gives examples where synonymous mutations driven by GC-to-AU mutation pressure have occurred at all three positions. Knight et al. (2001b) came down strongly in favor of the AI mechanism. We wish to redress the balance here. Our data show the CD mechanism is the best explanation in over half the cases. We agree with Swire et al. (2005) that CD applies well to stop codon reassignments, but does not usually apply to sense codon reassignments. The cases of CUN and CGN reassignments are exceptions to this, where we argue for CD with sense codons. In their proposal of the AI mechanism, Schultz and Yarus (1994) identified three specific types of codon-anticodon mispairing that might be important in allowing ambiguous translation of a codon: GA at the third codon position, CA at the third codon position, and GU at the first codon position. Knight et al. (2001b) considered whether these mispairing events were implicated in the reassignments in mitochondrial genomes. In Table 6, we also do this (cf. Table 3 of Knight et al.). We agree that GA mispairing at the third codon position is important in the reassignments of AAA and AGR. For the AUA Ile-to-Met transition, we argue that GA mispairing between tRNA-Ile(GAU) and the AUA codon is important, otherwise deletion of the tRNA-Ile(CAU) would not be possible. Mispairing of the unmodified tRNA-Met(CAU) with AUA (as in the Knight et al. table) might also be relevant, but we do not think this is significant prior to the deletion of the tRNA-Ile(CAU). The reverse transition of AUA from Met back to Ile listed by Knight et al. does not occur in our interpretation (Fig. 4). Also, the reassignment of UAA from Stop to Tyr listed by Knight et al. is no longer thought to occur. Knight et al. argue that CA mispairing is important for the UGA Stop to Trp transition. There is evidence for this in Bacillus subtilis (Lovett et al. 1991, Matsugi et al. 1998), where UGA is ambiguous. It is not clear whether this is a key feature in the UGA reassignments in mitochondria. There are many species where the canonical system works fine (UGA is Stop, UGG is Trp, and the tRNA-Trp has CCA anticodon). It is possible that if CA mispairing started to occur due to some change in the tRNA-Trp outside the anticodon, UGA would become ambiguous, and this would provide a selective pressure against UGA codons that would help to drive them to disappear. However, we are not aware of any evidence for this in mitochondria. The disappearance of UGA is likely to occur in any case due to mutation pressure and drift, and this seems to be a more important factor. The other cases listed by Knight et al. where mispairing is implicated do not occur in mitochondria, and we did not consider them. From Table 6, we would argue that mispairing is important in some but not all reassignments. However, the existence of mispairing does not necessarily imply the AI mechanism occurred, and mispairing is not a diagnostic feature of AI, as previously envisioned. If translation of a codon becomes inefficient due to tRNA loss, then selection will act against it and it will tend to be replaced by synonymous codons. If this situation remains for some time without any compensating gain of function in another tRNA, then the codon may in fact disappear. We have argued that this is the case in S. octosporus, where the tRNA-Ile(CAU) has been deleted and AUA has disappeared, and also in the two Kluyveromyces species, in which AUA is very rare (Table 3). A similar thing happens with the AGR reassignment: there are several invertebrate groups in which AGG becomes rare after the deletion of the tRNA-Arg because of inefficient pairing with the tRNA-Ser when it retains its original GCU anticodon (Table 4). In cases where there is a full gain in function of the new tRNA, the codon in no longer selected against. This is seen when modifications occur in both the tRNA-Met and the tRNA-Ser. Given that the loss of the tRNA in the first step of the UC mechanism is likely to be deleterious, anything that offsets the deleterious effect to some extent will increase the likelihood of the UC mechanism occurring. It has been suggested that deletion of a tRNA gene might be favored since it reduces the total length of the genome and hence allows more rapid replication (Andersson and Kurland 1991, 1995). We considered this effect in our simulations (Sengupta and Higgs 2005) and showed that selection for reduction in genome length does indeed increase the likelihood of the UC mechanism. This factor will be more important in smaller genomes because the relative change in length for deleting a tRNA will be larger. For the cases of deletion of the tRNA-Ile and tRNA-Arg that we are considering, we do not have an experimental measurement of the selective disadvantage to the translation system of loss of the gene, so it is not possible to say whether the potential advantage of shortening the genome is large or small compared to this. We do not wish to argue that these reassignments are “caused” by selection to reduce the genome size, but we do wish to say that it is the chance deletion of these tRNAs that initiated the reassignment in these cases, and that the fixation in the population of the genome with the deletion could well be aided by selection for more rapid genome replication. However, there is no general trend for tRNA deletion at codon reassignment. We have given the changes in tRNA number associated with our interpretation of the reassignment events in Table 6. There are three cases of tRNA gain, each of which occurs once. These are due to a tRNA duplication followed by an anticodon mutation. There are four cases of tRNA deletion, which together occur 10 times. Changes in release factors are not included in this table because these genes are not coded by the mitochondrial genome. In Table 6 in the supplementary information of Swire et al. (2005), a distinction is made between the theoretical and the observed changes in tRNA number. Given the genome information now available, this distinction is no longer necessary. The AGR Arg to Ser reassignment in Bilateria is classed as an observed gain by Swire et al. (their case 9) because they are comparing with Cnidaria. However, it is clear that tRNA import has evolved separately in Cnidaria and that it does not occur in either Porifera or the ancestral Bilaterian. The proper comparison for Bilateria is with Porifera, where the tRNA-Arg is present; therefore it is evident that the gene has been deleted in Bilateria at the point of the codon reassignment. Case 6 of Swire et al. is the AUA Ile to Met that occurs in yeasts. We have shown in this paper that the tRNA-Ile(CAU) gene is present in the fungi prior to this reassignment and is deleted at the point of the reassignment. Once again, this is a clear deletion and there is no difference between the observed and the expected change. An important issue in the AI mechanism is the nature of the selective effect on an ambiguous codon. In our model (Sengupta and Higgs 2005), we assumed that ambiguous translation would be unfavorable because the wrong amino acid would be inserted part of the time. However, it has been suggested that ambiguous translation may be favorable (Santos et al. 1999). Swire et al. (2005) also gave an argument related to the cost of amino acid biosynthesis, whereby ambiguous translation might be favorable because the savings in the cost of using an energetically cheaper amino acid might outweigh the penalty of the ambiguity. The ambiguity penalty may be very small at particular sites where the amino acid substitution is neutral or almost neutral; hence, the cost savings would make this change advantageous. This mechanism is possible and we are currently in the process of studying it using simulations of the same type as those used by Sengupta and Higgs (2005). A problem we see with this argument is that, if an amino acid substitution were favored by cost saving, it would be easier to achieve this by a straightforward mutation in the gene without changing the genetic code. Such mutations could occur in places where it was favorable and not in places where it was not. There must be some sites in the protein where the substitution is deleterious, even if there are also some neutral sites. On the other hand, if the codon were reassigned, this would force changes in all sites in the protein, whether favorable or not, which would act against the postulated benefit of the change. We do not think that positive selection for ambiguous translation can be a general explanation of sense-to-sense reassignments. We have already given one case where a sense-to-sense reassignment is better explained by the CD mechanism and several where it is better explained by UC. Swire et al. (2005) did not distinguish between AI and UC. However, the cost-savings argument is less relevant for UC than AI. In AI, if the cost-savings scenario occurred, it would be central to the argument because it would provide a reason why the gain occurred first, i.e., it would be the driving force for the reassignment. In the UC mechanism, the cause of the transition is the chance loss event, which is presumed to be deleterious and is not driven by selection. The gain event would then be positively selected because it would get rid of the penalty from the slow translation that occurs when there is no well-adapted tRNA. The cost-savings argument might also favor the gain event in this case. This would change the details of the selective forces favoring the gain, but this would only be relevant after the initial deletion. However, there is no need to postulate the existence of the neutral sites and cost savings in the UC case, because the gain would be driven by selection to improve translational efficiency anyway. Thus, if the cost-savings scenario occurred in a UC reassignment, it would be a minor detail, not a driving force. In any case, the best-documented example of ambiguous translation of a codon is the CUG codon in the Candida nuclear genome (Santos et al. 1999, 2004), and this example now provides evidence that ambiguous translation is deleterious. It has been shown that the reassignment is driven by the gain of a new tRNA-Ser with anticodon CAG that translates the CUG codon, which is normally Leu. This case was formerly cited as an example of positive selection for ambiguous translation. However, Massey et al. (2006) have now shown that there is selection against the ambiguous CUG codons that causes them to become very rare in their original Leu positions. Most occurrences of CUG in Candida are newly evolved in Ser positions. This shows that there is negative selection on the ambiguous codon. (Note that this is the third reason for a codon to become rare: in the CD mechanism, the codon disappears due to mutation pressure; in the UC mechanism, the codon is inefficiently translated and selected against and therefore might become rare; in the AI mechanism, the codon might be selected against because it is ambiguous, and therefore might become rare.) Another study suggesting that ambiguous translation is deleterious is that by Kim et al. (2000), who studied many examples of anticodon mutations in tRNA-Phe from E. coli. In cases where the cells expressing the mutant tRNA were viable, it was found that the mutant tRNA was charged with the amino acid corresponding to the new anticodon, i.e., there was no ambiguous translation. However, in other cases, expression of the mutant tRNA was lethal. It was presumed that the mutant tRNA was still charged with Phe, although its anticodon corresponded to another amino acid. This would cause ambiguous translation, which was apparently deleterious in the experiment. The arguments in this paper apply only to codon reassignments that have occurred since the establishment of the canonical code. They do not apply to the origin and buildup of the canonical code. We believe that there is a good case that positive selection was important in the establishment of the canonical code (Freeland et al. 2003; Ardell and Sella 2002) and that this accounts for the nonrandomness and apparent optimization of the code. Adding a new amino acid to the code opens up a whole new realm of protein sequences that can be formed using the new amino acid. Therefore, there is ample scope for positive selection to act in determining the codons to which each amino acid is initially assigned. Translational error rates and mutation rates may have been large in early organisms, so minimizing the effects of these errors may have been important. However, we do not think that positive selection guides the codon reassignments that have occurred after the establishment of the canonical code. There is little to be gained from a codon reassignment in modern organisms because no new amino acid is being added, and the effects of minimizing error will be very small when both translational error and mutation rates are as small as they are today. tRNA Evolution Although Table 6 shows no strong trend to reduce the number of tRNAs at codon reassignment events, we feel that this obscures the very large role that tRNA deletion has had in the evolution of the mitochondrial translation system. A major difference between bacterial and mitochondrial systems is that four-codon families in bacteria require at least two tRNAs—one with wobble position G to translate U- and C-ending codons and one with wobble position U to translate A- and G-ending codons. Most mitochondria require only one tRNA with wobble position U for all four codons. In metazoan mitochondria the second tRNA has been deleted in all eight four-codon families. The same genes have been deleted in other groups too, and we suspect that there have been independent deletions of the same gene, although demonstrating this would be difficult because we do not have a fully resolved tree for the earliest branches of the eukaryotes. Nevertheless, the same process has occurred at least once in all eight four-codon families, which is relevant to the discussion on genome streamlining, even though there is no change in the code. Loss of a tRNA with wobble position G may have been slightly deleterious originally when it occurred in the early stages of mitochondrial evolution, but it is possible that the ribosome has become sufficiently flexible to tolerate this. It could also be that the loss still comes with a price of decreased accuracy and/or speed of translation. Selection for reduction in genome length might well play a role in these deletions also. The 22 tRNAs in animal mitochondria represent the minimal possible set. It is only possible to reduce the number further by making codons unassigned or by evolving a mechanism of tRNA import from the cytoplasm. We have seen that tRNA import has evolved multiple times. If import is possible, then there will be no deleterious effect of further tRNA deletions from the mitochondrial genome, and selection for shorter genome length may favor the fixation of variants in which tRNAs have been deleted. Thus rapid tRNA deletion might be expected in lineages in which a general import process is operating. It is known that many protein genes formerly present on the mitochondrial genome have been transferred to the nuclear genome (Blanchard and Lynch 2000). Many of these code for proteins that are required in the mitochondrion and are targeted back to the organelle after synthesis. In cases where tRNAs are imported, it is interesting to ask whether these tRNAs are in fact formerly mitochondrial tRNA genes that have been transferred to the nucleus, or whether the existing nuclear tRNAs have become functional in the mitochondria and the mitochondrial genes have simply been deleted. We attempted to answer this question with the kinetoplastids, Trypanosoma brucei and Leishmania tarentolae, which possess no mitochondrial tRNAs and where import is known to occur (Simpson et al. 1989; Hancock and Hajduk 1990; Schneider and Marechal-Drouard 2000). The full set of tRNAs in the nuclear genome of these species is known. We carried out phylogenetic analysis comparing these genes with tRNAs from other eukaryotic nuclei, bacteria, and mitochondria. For each separate tRNA, the kinetoplastid genes clustered with eukaryotic nuclear genes and not with mitochondria. We do not show these results because the trees have very low statistical support due to the short sequence length of the tRNAs. Nevertheless, we found no evidence that any of the kinetoplastid tRNAs had been transferred from the mitochondrial genome. The identity elements and antideterminants of tRNAs are key aspects of the molecules that play a crucial role in the proper aminoacylation and prevention of misacylation of a tRNA (Giegé et al. 1998). These are important factors in understanding changes in the genetic code. Most aminoacyl tRNA-synthetases use the anticodon as a key identity element for proper recognition of the tRNA. Two exceptions are Leu and Ala, where the anticodon has no role to play in the proper charging of the cognate tRNA with the amino acid. In such cases, mutations in the anticodon do not affect the aminoacylation of the tRNA. Such mutations can result in codon reassignment if the resulting anticodon acquires the ability to pair with a codon originally associated with a different amino acid. The reassignment of UAG from Stop to Leu in the mitochondrial genome of some species of green plants provides such an example (Laforest et al. 1997). Another possibility is that misacylation of a tRNA due to a change in an aminoacyl tRNA-synthetase could be a cause of a codon becoming ambiguous. The tRNA would be ambiguously charged, but there would only be one type of tRNA for the codon. However, this does not seem to occur in the mitochondrial reassignments we considered here. Conclusions We have given arguments above as to which mechanisms seem most likely in each of the codon reassignments in mitochondrial genomes. We have shown that the many reassignments of stop codons to sense codons are readily explained by CD, given the biased base composition of most mitochondrial genomes and the small total number of occurrences of stop codons in these genomes. Disappearance of sense codons is more difficult because the total number of codons for the corresponding amino acid is large. A very strong mutational bias is required for sense codons to disappear. However, in one group of yeast species, an extreme bias against C does exist, and we argue that the reassignment of CUN and CGN codons in these species is attributable to CD. In the other examples of sense codon reassignments, the mutational bias is in the wrong direction for causing CD, and the probability of disappearance is negligible. Where the codon does not disappear, we have emphasized the important distinction between the UC and AI mechanisms. The case for the UC mechanism is most clear when the reassignment is associated with a tRNA deletion. We then argue that the deletion initiated the process and the codon reassignment occurred as a response to this. The UC mechanism does not rely on selection for reducing genome length, but if such selection were significant, this would increase the likelihood of this mechanism. Many nonessential tRNAs have been deleted during mitochondrial genome evolution and these did not initiate codon reassignments because the original code was still functional after the deletion. However, this makes it clear that chance tRNA deletion is a relatively common event. We also observed several cases where a good argument can be made for the AI mechanism, i.e., where the reassignment arose because the codon first became ambiguous. In other cases, scenarios for both AI and UC seemed equally plausible, and it is difficult to distinguish them after the event. We see these genetic code changes as chance events, rather than as changes governed by positive selection. Disappearance is a chance event that occurs under drift when there is strong mutational bias in base frequencies. If a change in tRNAs or release factors happens to occur while a codon is absent, then a codon reassignment can occur. However, this is a chance event: the codon frequency could drift back to a higher level without any reassignment occurring. Our interpretation of reassignments via UC and AI mechanisms is that they too are initiated by chance events, such as the deletion of a tRNA gene or a change in the process of base modification in an anticodon. These changes are probably slightly deleterious, but efficient functioning of the translation system can be restored by making the codon reassignment. The view that AI states are driven by positive selection seems unlikely to us at present. The origin of the canonical code is outside the scope of the present paper. However, we emphasize that the situation in codon reassignments in modern organisms is different from that during the early evolution of the canonical code, where positive selection probably had an important role. We conclude that our gain-loss framework is suitable as a description of the real codon reassignment events. It emphasizes that there are several mechanisms that are alternatives within a larger picture, and that it is not always profitable to discuss these mechanisms as though they were mutually exclusive. These mechanisms can and do occur in nature, and one mechanism is not sufficient to explain all cases. Electronic Supplementary Material Supplementary material

Transgenic_Res-3-1-1829418

Effects of FVB/NJ and C57Bl/6J strain backgrounds on mammary tumor phenotype in inducible nitric oxide synthase deficient mice

The ability to genetically manipulate mice has led to rapid progress in our understanding of the roles of different gene products in human disease. Transgenic mice have often been created in the FVB/NJ (FVB) strain due to its high fecundity, while gene-targeted mice have been developed in the 129/SvJ-C57Bl/6J strains due to the capacity of 129/SvJ embryonic stem cells to facilitate germline transmission. Gene-targeted mice are commonly backcrossed into the C57Bl/6J (B6) background for comparison with existing data. Genetic modifiers have been shown to modulate mammary tumor latency in mouse models of breast cancer and it is commonly known that the FVB strain is susceptible to mammary tumors while the B6 strain is more resistant. Since gene-targeted mice in the B6 background are frequently bred into the polyomavirus middle T (PyMT) mouse model of breast cancer in the FVB strain, we have sought to understand the impact of the different genetic backgrounds on the resulting phenotype. We bred mice deficient in the inducible nitric oxide synthase (iNOS) until they were congenic in the PyMT model in the FVB and B6 strains. Our results reveal that the large difference in mean tumor latencies in the two backgrounds of 53 and 92 days respectively affect the ability to discern smaller differences in latency due to the Nos2 genetic mutation. Furthermore, the longer latency in the B6 strain enables a more detailed analysis of tumor formation indicating that individual tumor development is not stoichastic, but is initiated in the #1 glands and proceeds in early and late phases. NO production affects tumors that develop early suggesting an association of iNOS-induced NO with a more aggressive tumor phenotype, consistent with human clinical data positively correlating iNOS expression with breast cancer progression. An examination of lung metastases, which are significantly reduced in PyMT/iNOS−/− mice compared with PyMT/iNOS+/+ mice only in the B6 background, is concordant with these findings. Our data suggest that PyMT in the B6 background provides a useful model for the study of inflammation-induced breast cancer. Introduction Transgenic mouse models that develop spontaneous mammary adenocarcinomas have proven valuable in revealing molecular mechanisms underlying tumorigenesis and metastasis (Cardiff 2003; Fargiano et al. 2003). Models target specific pathways depending on the transgene being expressed under the control of the mouse mammary tumor virus long terminal repeat (MMTV-LTR) or whey acid protein (WAP) mammary gland promoters and thereby replicate genetic defects in subsets of human tumors (Cardiff and Wellings 1999; Rosner et al. 2002). Molecules from pathways known to be important in the pathogenesis of human breast cancer such as the Wnt, Ras and ErbB2/Neu pathways have been utilized, however few spontaneous mammary tumor model systems develop metastasis, a crucial aspect of cancer progression. As demonstrated by gene expression analysis (Ma et al. 2003), the polyomavirus middle T (PyMT) model in the FVB/N (FVB) background activates the same signaling pathways as erbB2, an oncogene amplified or overexpressed in approximately 30% of human breast cancers. The model has been particularly useful for genetic studies due to its short tumor latency and high incidence of pulmonary metastasis (Guy et al. 1992). These features have enabled global expression profiling of PyMT tumors from different genetic backgrounds to identify signatures of tumor virulence (Qiu et al. 2004) that are consistent with a gene set predictive of metastasis in human tumors (Ramaswamy et al. 2003). Hence, the model has been used effectively to dissect early and late stages of mammary tumor development (Guy et al. 1994; Lin et al. 2001; Webster et al. 1998; Williams et al. 2004). In women with breast cancers involving this pathway, the course of the disease varies widely. One powerful means to understand that variability is to identify useful animal models. The PyMT breast cancer model can be used in genetic crossing experiments to identify loci having modifier effects on either tumorigenesis or metastasis (Le Voyer et al. 2000). An understanding of the genetic polymorphisms underlying these effects is important in the study of human cancer as genetic variability affects host-tumor interactions in numerous ways and is an important factor in the response to therapy. Widely varied tumor responses to polyoma virus infection among different inbred strains of mice have been well documented (Freund et al. 1992; Gross 1983). Much of this difference can be attributed to the major histocompatibility complex (MHC) which results in immunologically mediated resistance associated with an H-2b haplotype such as in C57Bl strains (Law et al. 1967; Ting and Law 1965). Other inbred strains, notably those with the H-2d and H-2k haplotypes have moderate and weak tumor resistance respectively (Freund et al. 1992). The FVB strain has an H-2q haplotype, which may be an additional MHC locus that confers susceptibility to polyoma virus induced tumorigenesis. Alternatively, non-MHC genes can also result in a dominant pattern of tumor susceptibility as in the case of the C3H/BiDa strain (Freund et al. 1992). Strain variations are also relevant to the study of genetic mutations in mouse models of disease. An analysis of targeted disruptions on mammary tumorigenesis in the PyMT model requires breeding of knockout models usually developed in the 129/SvJ-C57Bl/6J background into the FVB background. It is commonly recognized that the FVB strain is susceptible while the C57Bl/6J (B6) strain is more resistant to PyMT mammary tumorigenesis and F1 crosses of these strains increase tumor latency by approximately 6 days (Lifsted et al. 1998). However, it is not clear what effects continued backcrossing has on the analysis of targeted mutations in the PyMT model. Previously we reported that inducible nitric oxide synthase (iNOS) deficiency in the PyMT model in the B6 background resulted in an increase in tumor latency that was not observed in the FVB background (Ellies et al. 2003). Since the issue of strain background is an important one for analysis of mouse models we have made iNOS knockout mice congenic in the FVB and B6 backgrounds and carried out a comparison of tumorigenesis and metastasis in both strains. Materials and methods Animals PyMT (FVB/N-Tg(MMTV-PyVT)634Mul/J) mice (Guy et al. 1992) were obtained from W. Muller (McMaster University, Ontario, Canada). We acquired PyMT mice that had been backcrossed 5 generations into the B6 background from A. Varki (UCSD, La Jolla, CA). Control (C57Bl/6J) and iNOS deficient mice (B6.129P2-Nos2tm1Lau/J; Laubach et al. 1995) were purchased from The Jackson Laboratory, Bar Harbor, ME. The PyMT B6 mice were further backcrossed until congenic in the B6 background and then bred with iNOS−/− mice. iNOS−/− mice were bred into the FVB background until congenic (>10 generations) and then crossed with PyMT FVB mice. Female mice heterozygous for the PyMT transgene and homozygous for the wild type or mutated iNOS gene were used in these studies. For clarity and brevity, we have designated FVB/N-Tg(MMTV-PyVT)634Mul/JxB6.129P2-Nos2tm1Lau/J+/+ mice PyMT/iNOS+/+ and FVB/N-Tg(MMTV-PyVT)634Mul/JxB6.129P2-Nos2tm1Lau/J−/− mice PyMT/iNOS−/−. FVB mice were palpated twice weekly from 4 weeks of age and B6 mice once weekly from 8 weeks of age to monitor mammary tumor development. Tumors were measured in 2 dimensions using calipers and tumor volume estimated using the standard calculation for a sphere 4/3 × 3.14 × a × b2 where a is the smaller diameter and b is the larger diameter. After euthanizing the mice, mammary tumors were dissected and weighed and the total tumor burden calculated (tumor weight/body weight). All studies followed the NIH guidelines for the care and treatment of experimental laboratory rodents. Mammary gland whole mounts Pairs of inguinal mammary glands were fixed in formalin or Carnoy’s fixative, dehydrated through a graded series of ethanol solutions and defatted in xylene. Following rehydration, the mammary epithelium was stained with carmine stain (Sigma Chemical, St Louis, MO) for 30 min. After removing excess stain by washing in water, samples were dehydrated and stored in methyl salicylate (Sigma Chemical, St Louis, MO). Tumor microvascular density Cryosections of OCT embedded tumor samples were cut at 7 μm and stained with anti-CD31 clone 13.3 (1:200, BD Pharmingen, San Diego, CA) as previously described (Ellies et al. 2003). Five random fields per tumor were captured on the Spot RT digital camera system (Diagnostic Instruments, Sterling Heights, MI) at 200× magnification and analyzed using the public domain NIH Image program (http://rsb.info.nih.gov/nih-image/). Microvessel areas were normalized to 1 mm2 of epithelium. Lung metastases Paraffin embedded lungs were serial sectioned and stained with hematoxylin and eosin. Both the numbers and areas of lung metastases were measured by light microscopy using images and software from the Spot RT digital camera system. Lung area per section was measured using a dissecting scope and metastatic tumor burden calculated (mm2 of lung metastases/cm2 of lung tissue) for each animal using data from 3 sections at least 100 μm apart. At least 13 PyMT/iNOS+/+ or PyMT/iNOS−/− mice were used in this analysis. Statistical analysis The percentage of mice tumor free in each group was analyzed using a Kaplan-Meier survival analysis, and the log-rank statistic was used for comparison of the curves between groups. The t-test was used for comparison of 2 groups and the analysis of variance (ANOVA) followed by the Bonferroni post-test was applied to multiple comparisons. The Mann–Whitney U test was used to compare metastastic burden in the PyMT/iNOS+/+ and PyMT/iNOS−/− groups. Results PyMT tumorigenesis is delayed in the C57Bl/6 strain The PyMT model was originally developed in the FVB strain due to its high fecundity. Tumor growth occurs rapidly and metastatic tumors form in the lungs (Guy et al. 1992). To compare tumorigenesis in the FVB strain with that in the C57Bl/6 strain onto which many genetically modified strains are bred, we made PyMT mice congenic in the B6 background. Examination of tumor latency showed a dramatic difference with a mean latency of 53 days for PyMT tumor development in FVB mice and 92 days in B6 mice, a delay of approximately 6 weeks (Fig. 1A). Whole mounts of mammary glands from the two strains also showed that even at weaning, hyperplasias were more advanced in FVB than B6 PyMT mice (Fig. 1B). Fig. 1PyMT tumorigenesis in the FVB and B6 strains. (A) The day at which the first mammary tumor was detected by palpation in each mouse was recorded and a Kaplan-Meier survival plot generated. Log rank analysis indicated a highly significant difference, P < 0.001 between the curves. FVB mice N = 44, B6 mice N = 46. (B) Whole mounts of mammary glands from 3 week old PyMT/iNOS+/+ mice of the FVB (left panel) and B6 (right panel) strains Effects of iNOS deficiency are more apparent in the C57Bl/6 background To determine how this difference in tumor latency may affect the phenotype observed following crossbreeding with genetically altered mice, we bred in iNOS−/− mice congenic in the two backgrounds. As previously reported, we observed a 3–4 week delay in palpable tumors in B6 PyMT mice lacking iNOS activity (Ellies et al. 2003). However, this difference was difficult to discern in the FVB background due to the rapid tumor onset (Fig. 2A). Mammary gland whole mounts reflected the latency data with a clear difference in tumor growth with iNOS deficiency being observed in the #4 glands of B6 PyMT mice (Fig. 2A). Fig. 2Effects of iNOS deficiency on mammary tumor development. (A) Whole mounts of PyMT/iNOS+/+ and PyMT/iNOS−/− #4 mammary glands from 11 week FVB mice or 20 week B6 mice. LN = lymph node. (B) Kaplan Meier plots of the tumor latency of PyMT/iNOS+/+ annd PyMT/iNOS−/− mice of the 2 strains. Data were analyzed by the log rank test. PyMT/iNOS+/+ FVB N = 44, PyMT/iNOS−/− FVB N = 57, PyMT/iNOS+/+ B6 N = 46, PyMT/iNOS−/− B6 N = 26 PyMT tumor growth kinetics vary in FVB/N and C57Bl/6 strains The pattern of tumor growth differed between strains and may in part account for the greater disparity in tumor development observed in the absence of iNOS in PyMT B6 mice. Once the mean tumor volume reached 1 cm3, tumor growth proceeded exponentially in the PyMT FVB mice leading to sacrifice within 3 weeks (Fig. 3A). To achieve the same final tumor volume, the PyMT B6 mice required approximately 7 weeks and growth followed a more linear pattern, suggesting more variable rates of growth of individual tumors. Indeed, when we analyzed the tumor latency of all tumors, we were able to define two distinct phases of tumor growth in the B6 strain: early, from 12–17 weeks and late, from 19–23 weeks (Fig. 3B). The major effect of iNOS deficiency appears to be a delay in the early phase of tumor development. Fig. 3Mammary tumor growth kinetics. (A) Mean tumor volume per mouse. FVB N = 32, B6 N = 44. Data are means ± SEM. (B) Tumor latency for individual B6 PyMT/iNOS+/+ and PyMT/iNOS−/− tumors. Data were derived from examining 460 B6 PyMT/iNOS+/+and 260 PyMT/iNOS−/−tumors Deficiency in iNOS-modulated tumor growth is site specific In concordance with increased tumor latency and decreased tumor growth (Ellies et al. 2003) tumor burden was significantly lower in PyMT/iNOS−/− mice (Fig. 4A). An examination of tumor burden in individual mammary glands revealed a reduction in all PyMT/iNOS−/− glands compared to controls, although the numbers were only significant in the #3 and #4 mammary glands (Fig. 4A). It is interesting to note that these are the glands with the largest fat pads and may indicate a role for the stroma in the iNOS effect. To assess a role for tumor vascularization in this effect we examined tumor microvascular density (MVD) in #1and #2–5 tumors of both genotypes. Although there was a slight reduction in the mean MVD in #1 versus #2–5 tumors in the PyMT/iNOS+/+ and PyMT/iNOS−/− genotypes, it was not statistically significant, suggesting this is not the mechanism for the difference observed in tumor growth at different locations (Table 1). Fig. 4Distribution of B6 mammary gland tumor burden. (A) Total tumor burden was significantly lower in PyMT/iNOS−/− mice, ***P < 0.001. (B) Examination of individual tumors indicated that a significant difference in genotypes was only found in the #3 and #4 mammary glands, although in each gland the mean tumor burden was lower in PyMT/iNOS−/− mice. *P < 0.05 ANOVA followed by Bonferroni post tests. PyMT/iNOS+/+ B6 N = 24 (black bars), PyMT/iNOS−/− B6 N = 25 (grey bars)Table 1Quantitative analysis of tumor microvascular densityGenotypeLocationNMVD (μm2/mm2)PyMT/iNOS+/+#19216 ± 20PyMT/iNOS+/+#2–59176 ± 12PyMT/iNOS−/−#16201 ± 11PyMT/iNOS−/−#2–511189 ± 11Data are means ± SEM This site-specific effect prompted us to examine tumor latency in individual glands. We had observed during palpations that the #1 mammary gland was usually the first to develop tumors, however we were surprised to find a clear difference between the latency of the #1 tumors compared with all other tumors (Fig. 5). This difference was more noticeable in the absence of iNOS. The greater difference in mean latency in the #4 glands compared with the #1 glands was consistent with the tumor burden results being more significant in the #4 glands. Fig. 5Tumor latency of individual B6 mammary glands. Tumor latency was lower in the #1 mammary gland in PyMT/iNOS+/+ mice compared with all other mammary glands (upper panel). A similar result was observed in PyMT/iNOS+/+ mice, with an increase in disparity between the latency of the #1 gland and all other glands (middle panel). When the #1 and #4 latencies for both genotypes were overlayed, a greater difference in mean tumor latency was observed between the #4 glands when compared with the #1 glands (lower panel) iNOS effects on tumor metastasis The PyMT model has been useful for the study of tumor metastasis due to the high penetrance of this clinically important characteristic of tumor progression. However, we found that in the FVB strain, tumors were so aggressive that no significant difference could be observed in the metastatic burden occurring in PyMT/iNOS+/+ and PyMT/iNOS−/− mice Fig 6. Although the incidence of tumor metastasis was less in the B6 strain, a statistically significant difference between the genotypes was found. We have maintained PyMT/iNOS−/− mice for up to 32 weeks and found no increase in metastatic burden (data not shown), suggesting that the decrease in metastasis is not due simply to the delayed tumor latency in these mice. Fig. 6Tumor metastasis. (A) HE stained paraffin embedded sections of PyMT/iNOS+/+ (left panel) and PyMT/iNOS−/− (right panel) lungs. (B) Comparison of metastatic tumor burden in FVB and B6 mice *P < 0.05 Mann-Whitney U test. PyMT/iNOS+/+ FVB N = 14, PyMT/iNOS−/− FVB N = 13, PyMT/iNOS+/+ B6 N = 17, PyMT/iNOS−/− B6 N = 21 Discussion Mouse models of mammary tumorigenesis have provided important insights into the molecular basis of human breast cancer and are essential for gathering preclinical data regarding therapeutic drug efficacy. The PyMT model in the FVB strain has a mean tumor latency of 53 days and has been used extensively in studies of mammary tumorigenesis (Cardiff 2003; Muller and Neville 2001). The model is frequently crossed with gene-targeted strains that have been generated in the B6/129 mixed background or backcrossed onto the B6 background. We had highly variable results examining iNOS effects in a mixed B6/FVB PyMT background and sought to characterize the phenotype of this mutation in congenic FVB and B6 strains. Our study demonstrates that the genetic background of the PyMT model can affect the ability to discern differences in phenotype due to genetic alterations that require time to influence tumor progression. Differences observed in the B6 background were obscured in the FVB background due to the signficantly shorter tumor latency and more rapid tumor growth in this strain. Further analysis of mammary tumorigenesis in the B6 background indicated that tumor formation occurred in 2 phases: early, from 12–17 weeks and late, from 19–23 weeks. It is the early phase of tumor formation that is most affected in the absence of iNOS, consistent with recent data linking iNOS expression with transcriptional upregulation of signal transducers and activators of transcription 3 (STAT3) by nuclear epidermal growth factor receptor (EGFR) (Lo et al. 2005). EGFR/ErbB1, a member of the ErbB family of receptor tyrosine kinases that signal via the same pathways as PyMT (Webster et al. 1998), is also found in the nucleus of highly proliferative cells where is functions as a transcription factor (Marti et al. 1991; Wells and Marti 2002). Structural and functional characterization of the iNOS promoter has identified binding elements for both EGFR and STAT3 (Lo et al. 2005). Immunohistochemical studies of breast carcinomas show that nuclear EGFR positively correlates with iNOS and survival analysis demonstrates that high levels of iNOS are a prognostic indicator of poor survival (Lo et al. 2005). Thus the B6 PyMT mice in the presence or absence of iNOS deficiency are likely to provide an important model for further analysis of these novel molecular interactions. Mammary tumor latency for each of the ten murine mammary glands has generally been assumed to be stoichiometric. However our data demonstrate that the #1 mammary glands have tumor initiating properties that are distinct from the other 4 pairs of mammary glands. The #1 mammary glands have the smallest fat pads and are located in the head and neck region, which is highly vascular, so we hypothesized that an increase in vascular density in the #1 tumors could be responsible for this difference. An examination of the microvacular density (MVD) of #1 tumors versus all others showed a trend in which #1 tumors had a higher mean MVD than #2–5 tumors as determined by quantitative analysis of CD31 staining. However, these values were not statistically significant, suggesting a different mechanism is responsible for the observed site-specific differences in tumor burden and tumor latency. Other factors in the stromal microevironment may also play a role in mediating this effect. The release of epidermal growth factor (EGF) from the submandibular salivary gland has been shown to promote mouse mammary tumorigenesis in virus mediated and chemical carcinogen induced models (Kurachi et al. 1985; Molinolo et al. 1998). The proximity of the #1 glands to the submandibular gland make EGF induced proliferation a plausible explanation, however further investigations will be required to determine whether the release of this growth factor primarily affects the #1 glands. Another possible explanation for this intriguing difference is that the #1 mammary gland is developmentally distinct from the other mammary glands. The #1 gland is not the first to develop since placode 3 develops first at E11.0–11.5 on the ectodermal streak between the forelimb and hindlimb (Eblaghie et al. 2004; Mailleux et al. 2002). Placodes 1 and 5 arise from the axillary and inguinal streaks respectively and placodes 4 and 2 form where the mammary line abuts these streaks (Veltmaat et al. 2004). Recent studies indicate that signaling molecules essential for mammogenesis are placode dependent. Placodes 2 and 3 do not form in the absence of Lef1 mediated canonical Wnt signaling (Boras-Granic et al. 2006) and placodes 1, 2, 3 and 5 require Fgf10 signaling from somites underlying the ectoderm (Mailleux et al. 2002; Veltmaat et al. 2006). It is clear there are developmental differences between mammary glands and it is possible that early developmental factors result in an increased propensity for malignant transformation in the #1 mammary glands, though the identity of these factors remains to be determined. Since metastasis is an aspect of tumor progression that is critical to patient survival, we examined the effect of iNOS on tumor metastasis in both the FVB and B6 backgrounds. No significant difference in metastatic burden was found in PyMT/iNOS−/− mice in the FVB background, however the tumor burden was reduced in lungs of PyMT/iNOS−/− mice in the B6 background compared with controls. Furthermore, the reduction in metastatic burden was not due simply to the delayed tumor latency as we maintained PyMT/iNOS−/− mice longer to compensate for the difference and found that this did not alter our previous findings (Ellies et al. 2003). Human clinical studies have associated increased iNOS expression with poor prognosis in breast cancer patients (Loibl et al. 2005), and the use of non-steroidal anti-inflammatory drugs (NSAIDs) with a 22–39% reduction in the risk of breast cancer (Harris et al. 2005; Khuder and Mutgi 2001). To increase our understanding of the molecular pathways involved in these effects, it is essential to have animal models that recapitulate the clinical findings. We have shown that in the PyMT B6 background loss of iNOS, an important inflammatory mediator, results in increased tumor latency and decreased tumor metastasis that is not due simply to the difference in latency. This model will be useful in determining molecular pathways underlying the role of iNOS in breast cancer.

Eur_J_Pediatr-4-1-2234440

Moebius-Poland syndrome and hypogonadotropic hypogonadism

Case report A male infant was born at 39 weeks of gestation, with birth weight (3.1 kg) and body length (49 cm) (both 50th percentile). The occipitofrontal circumference was 33 cm (25th percentile). The newborn’s face was inexpressive due to complete facial diplegia, bilateral ophthalmoplegia with impairment of the vertical gaze of the right eye, palpebral ptosis, a carp-shaped mouth, and a high-arched palate. Muscle and skeletal abnormalities were also apparent, including the absence of the pectoralis major and trapezius muscles, as well as left cubitus valgus; left hand was hypoplastic and showed 5th finger clinodactyly. These features prompted the diagnosis of Moebius-Poland syndrome. No previous cases were known in the patient’s family. Clinical examination revealed no other dysmorphic features except for micropenis. The karyotype was normal (46 XY). During follow-up, a mild psychomotor delay was confirmed by Denver Developmental Screening Test-II. Cortical-subcortical atrophy with no other structural abnormalities was observed in magnetic resonance imaging (MRI) of the brain. Delayed puberty was recorded when the patient was 15 years of age (Tanner stage I was recorded after physical examination). The testicular volume (measured using a Prader orchidometer) was 2 ml. Anosmia-tested by a standard olfactory test (CCCRC)- was not present. An endocrinological investigation revealed low baseline serum FSH (0.5 mU/ml), LH (0.1 mU/ml) and testosterone (2 ng/dl) levels. To confirm the suspected hypogonadotropic hypogonadism, a gonadotrophin releasing hormone stimulation test was performed and a very subnormal response obtained (after 180 min and 24 h FSH concentrations were 3.5 mU/ml and 0.7 mU/ml, respectively, and LH concentrations 0.7 mU/ml and 0.2 mU/ml, respectively). GH, TSH, and ACTH levels were normal. A further MRI scan of the brain showed no pathological lesions in the hypothalamus or pituitary gland. With a diagnosis of hypogonadotropic hypogonadism established, the child was treated with gonadotrophic hormone followed by depot preparations of testosterone. The patient eventually reached a normal weight and height and attained full sexual development (final testicular volume 6–8 ml). Discussion The usual diagnostic criteria for Moebius syndrome include congenital facial palsy with impaired ocular abduction [1, 2, 4, 6]. Nevertheless, in a thorough review, Verzijl et al. reported that abducens nerve palsy is not always present [6]. Vertical gaze palsy was recorded in the right eye of the present patient, suggesting involvement of the 3rd cranial nerve (Fig. 1). MRI scans often demonstrate anomalies of the posterior fossa in patients with Moebius syndrome, and it has been proposed that the latter be understood as a complex congenital anomaly affecting the rhombencephalon [6]. The present patient, however, showed no such anomalies. In addition to the main signs, more than half of all patients with Moebius syndrome have other anomalies, usually limb defects [1, 3]. The specific combination of both Moebius syndrome and Poland syndrome—a rare congenital anomaly characterised by pectoralis muscle defect and ipsilateral hand abnormalities—has been reported before, and is then referred to as Moebius-Poland syndrome [5]. Some authors suggest it may represent a single developmental condition that may be due to the variable expression of a single gene or involve contiguous genes [6]. Fig. 1Patient photo Hypogonadotropic hypogonadism is unusual in Moebius syndrome [1–3]. The present patient is only the 7th case reported to date, and the first with Moebius-Poland syndrome.

Breast_Cancer_Res_Treat-4-1-2190785

Impact of a programme of mass mammography screening for breast cancer on socio-economic variation in survival: a population-based study

Background After a systematic mass mammography breast cancer screening programme was implemented between 1991 and 1996 (attendance 80%), we evaluated its impact on survival according to socioeconomic status (SES). Introduction Mammography screening aims at early detection of breast cancer so that adequate treatment will eventually lower breast cancer mortality. In a mass screening programme, it is therefore especially important to reach women who have the highest chance of being diagnosed with advanced stage or have the lowest survival rates. Women from lower socio-economic strata are less likely to attend population screening programmes [1–4] and are also more likely to present with unfavourable stage at diagnosis, [1, 5, 6] although not all studies confirm this [7–9]. Lower breast cancer survival rates among the disadvantaged are usually attributed to advanced stage at presentation, but also to suboptimal access to adequate treatment. A recent population-based study in Switzerland found social class to be an independent prognostic factor [10]. The mass breast cancer screening programme was introduced in 1991 for women of 50–69 years and became fully implemented in 1996 in the south of the Netherlands covered by the population-based Eindhoven Cancer Registry, with a continuous high participation rate. Based on previous work [11] and a new postcode-based indicator of socio-economic status (SES) introduced by Statistics Netherlands [12] we were able to investigate survival according to SES for a sufficient period of time after introduction. We studied whether survival according to SES was affected differentially by the implementation of the screening programme. Methods The Eindhoven Cancer Registry records data on all patients newly diagnosed with cancer in the south–eastern part of the Netherlands, an area with now 2.4 million inhabitants (about 15% of the Dutch population) and only general hospitals. Trained registry personnel actively collect data on diagnosis, staging, and treatment from the medical records after notification by pathologists and medical registration offices. In the area of the Eindhoven Cancer Registry, a biennial breast cancer screening programme for women aged 50–69 years was started in 1991 and fully implemented in 1996. The attendance rate was more than 80% [13]. For our analyses we included all patients age 50–69 years diagnosed in 1983–2002 with invasive breast cancer in the eastern part of the registration area (about 1 million inhabitants). This population has been followed-up for vital status up to 1-1-2005. Information on the vital status of all patients was obtained initially from the municipal registries and since 1998 from the Central Bureau for Genealogy. These registers provide virtually complete coverage of all deceased Dutch citizens. An indicator of socioeconomic status was developed by Statistics Netherlands [12] being based on individual fiscal data from the year 2000 on the economic value of the home and household income and provided at aggregated level for each postal code (average of 17 households). Socioeconomic status was categorized according to quintiles ranging from 1 (low) to 5 (high), with a separate class for postal codes with a care-providing institution (such as a nursing home). This measure is assumed to be valid ten years before and after the basic year (2000), so for patients diagnosed before 1990 we used a measure which was also based on postal code of residence, but socio-economic status (five categories) was based on data from a marketing agency (self-reported occupation and education define 45 social classes, collapsed into a 5-level indicator based on average number of years of education), as described before [11]. We also used both SES indicators for the whole study period (1983–2002) to make sure any effect of diagnostic period was not attributable to the indicator we used. We calculated distribution of age and stage of disease according to period of diagnosis. Stage was categorized according to the TNM classification [14]. Patients with either positive lymph nodes or metastases were considered to have advanced disease. Chi-square test was performed of changes in the distribution across the three diagnostic periods. T-tests were performed of differences between two groups. Crude survival analyses were performed. The log-rank test was used to evaluate significant differences between survival curves in univariate analyses. We used Cox regression models to compute multivariate rates. The proportional hazard assumption of the predictor was evaluated by applying Kaplan–Meier Curves. The predictor satisfied the assumption of proportionality as the graphs of the survival function versus the survival time resulted in graphs with parallel curves as did the graphs of the log(−log(survival)) versus log of survival time. The independent prognostic effect of SES was investigated, adjusting for age and stage of disease, and stratified according to period of diagnosis (1983–1990, 1991–1996, 1997–2002). We also calculated the age and stage-adjusted effect of period of diagnosis stratified according to SES. Results Median age was similar for all three periods of diagnosis (59, 60, and 59 years, respectively). Patients diagnosed between 1997 and 2002 had a significantly more favourable stage at diagnosis than patients diagnosed in earlier periods (P < 0.0001): the proportion diagnosed with stage I (tumour smaller than 2 cm, no axillary lymph nodes involved) increased from 30% in 1983–1990 to 41% in 1991–1996 and 45% in more recent years (Table 1). The proportion with advanced disease, i.e. stage III or IV, was significantly lower in the most recent period (9.7%) compared to 1991–1996 (14%) and 1983–1989 (26%, P < 0.0001). Treatment varied over time, with a large proportion receiving systemic therapy in recent years (50%, vs. 36% and 29%). Table 1Characteristics of all women age 50–69 years diagnosed with invasive breast cancer between 1983–2002 in Southeastern Netherlands1983–19901991–19961997–2002Totaln%n%n%n%TNMI4653064241838451,94539II6384266543805442,10843III27818152101156.254511IV1157.5684.4653.52485.0unknown382.5312.0241.3931.9Treatment*S alone27118341222491386117S + RT7665062840658362,05242S + RT + ST3052042227673361,40028S + ST10871238231134629.4ST alone261.7231.5231.3721.5Other583.8211.4130.7921.9Socio-economic status1 (low)336222851826214883182325213192034219986203308202791835519942194154102741835819786165 (high)3022031520414221,03121institution#00.0231.5331.8561.1unknown1097.1634.0834.52555.2Total1,5341,5581,8474,939100* S = Surgery, RT = Radiotherapy, ST = Systemic therapy#institution = care-providing institution such as a nursing home The proportion of patients from the lowest socio-economic class decreased from 22% in 1983–1990 to 18% in 1991–1996 and 14% in 1997–2002 (P < 0.0001), whereas the proportion from in the higher social classes increased. Stage distribution improved significantly over time for each social class (P < 0.01). It was similar for all SES groups in 1983–1990 (P = 0.7, Fig. 1), although the proportion with stage IV was somewhat lower in the highest classes. The stage distribution was marginally more favourable for high SES compared to the lowest SES group in both 1991–1996 and 1997–2002 (P = 0.06 both periods), although the overall effect of SES on stage was not significant in the last period of time (P = 0.4). Fig. 1Stage distribution according to socio-economic status and period of diagnosis of patients age 50–69 years with invasive breast cancer in Southeastern Netherlands Survival improved for all socio-economic strata over time (Fig. 2). Survival rates did not differ among patients from each of the socio-economic classes diagnosed 1983–1990 in the period (P = 0.9), 5-year survival rates being 70%, 70%, 70%, 68% and 69% for patients from the lowest to the highest social class, respectively. For patients diagnosed in 1991–1996, survival of patients with a high SES was better than that of all other socio-economic strata (P = 0.01), 5-year survival rates being 76%, 76%, 80%, 78%, and 87%, respectively. For patients diagnosed in 1997–2002 an increasing gradient in survival was observed (P = 0.002) ranging from the lowest rates for the lowest SES group to the highest for the higher classes (80%, 84%, 83%, 85%, and 89%, respectively). Fig. 2Trend in survival according to socio-economic status for all women age 50–69 years diagnosed with invasive breast cancer in Southeastern Netherlands Multivariate analysis (Table 2) showed that patients diagnosed between 1991 and 1996 from the lower social classes had a 29% higher risk of death compared to the highest socio-economic group, after adjusting for age and stage at diagnosis (HR for the lowest versus the highest SES group: 1.29, 95%CI: 1.0–1.7). The risk of death for low SES patients diagnosed since 1997 was twice as high as that for the highest SES group (HR: 2.01, 95%CI: 1.3–3.0). The overall effect of socio-economic status was significant in the last period (P = 0.02). Additional adjustment for treatment did not change risk estimates more than 5% (data not shown). Table 2Multivariate regression analysis of survival of breast cancer patients age 50-69 years according to period of diagnosis, Southeastern Netherlands1983–19901991–19961997–2002HR*a95% CIHR*a95% CIHR*a95% CIAge (continuous)1.031.0–1.01.041.0–1.11.011.0–1.0Socio-economic status1 (low)1.010.8–1.21.291.0–1.72.011.3–3.021.030.8–1.31.281.0–1.71.541.0–2.330.950.8–1.21.180.9–1.61.531.0–2.340.990.8–1.31.391.0–1.81.330.9–2.05 (high)b1.001.001.00Χ2 trend0.58 (P = 0.97)5.9 (P = 0.21)11.4 (P = 0.02)TNM stageIb1.001.001.00II1.751.5–2.12.331.9–2.92.001.4–2.8III3.112.5–3.84.673.5–6.25.393.6–8.1IV9.857.6–1316.012–2216.511–24unknown2.251.4–3.52.561.4–4.64.812.3–10* HR = Hazard Ratio, CI = Confidence IntervalaAdjusted for all variables listedbReference Age and stage-adjusted survival improved over time for all socioeconomic strata (Table 3), the largest improvements were found for the highest social classes. Table 3Multivariate regression analysis of survival according to socio-economic status (SES) of breast cancer patients age 50–69 years in Southeastern NetherlandsSES1 (low)2345 (high)HR*a95% CIHR*a95% CIHR*a95% CIHR*a95% CIHR*a95% CIPeriod of diagnosis1983–1990b1.001.001.001.001.001991–19960.870.7–1.10.720.6–0.90.800.6–1.00.840.6–1.10.600.5–0.81997–20020.770.6–1.10.490.4–0.70.610.4–0.80.490.3–0.70.360.2–0.5X2 trend3.21 (P = 0.21)20.9 (P < 0.001)9.2 (P = 0.01)14.1 (P < 0.001)35.6 (P < 0.001)HR = Hazard Ratio, CI = Confidence IntervalaAdjusted for age at diagnosis and stage of diseasebReference Discussion We found that the proportion of breast cancer patients with a low SES has decreased since the introduction of a mass biennial mammography screening programme with high response rates. Although stage distribution improved for all socio-economic groups, the proportion with advanced disease decreased the most in the highest socio-economic group. In the 1980s survival was similar for all socio-economic groups, but since the introduction of screening the survival of women with a high SES has improved more than that for low socio-economic classes, also after adjustment for age and stage. We used an indicator of socio-economic status based on the postal code of a residential area. This aggregate covers a relatively small geographical area, and thus represents a reliable approximation of individual socio-economic status. Furthermore, routinely collected income tax data (no questionnaires or interviews) have been found to provide reliable estimates of household income. Previous studies have proven that socio-economic differences based on neighbourhood data tend to reflect socio-economic differences well at the individual level [15–17]. Furthermore, this objective measure of SES is also applicable for older women (born before 1955), whose occupation or education does not always properly reflect their social class [18]. We also repeated the analyses comparing both SES indicators if they were applied for the whole study period (1983–2002) to ensure any that effect of diagnostic period was not attributable to the indicator we used, and it was not. The lower proportion of patients with a low SES since the introduction of screening is not likely to reflect the higher attendance rate of women from a higher social class because of the very high participation rate, although this is not known according to social class. Studies from other countries have shown that SES does play a role in participation in the screening programme, [2] sometimes [3] but not always [4] due to the costs of a screening mammogram. However, the costs for the mass screening programme in the Netherlands are completely covered by public funds. Furthermore, the mean attendance rate in the Netherlands has always been rather high (about 80%), [13] and in our study area even higher than the national mean (85% in 2005) [19]. Foreign-born women are more likely to be non-attenders in the Netherlands, [20] as well as in Sweden, [21] Australia, [3] and the US, [22] for a variety of reasons. However the incidence of breast cancer among these groups of migrants is relatively low in the Netherlands and the stage distribution is comparable to that of women born in the Netherlands [20]. So this is unlikely to have affected survival rates in our study. A lower attendance rate of low social classes will result in more advanced disease stages at presentation [23]. Before the start of the mammography screening programme, we found that the stage distribution for breast cancers diagnosed in 1980–1989 was slightly more favourable for the highest socio-economic group [24]. We have now shown that this was also true after the introduction of screening, although the differences were small. In fact, we found that, although stage distribution became more favourable for all socio-economic groups, the proportion with advanced disease decreased less in the lower socio-economic group. This differential stage distribution was also described in a recent Danish study, [6] although our differences were smaller. The variation in survival according to SES may also be related to differences in treatment, which depends on the disease stage and varies over time. The use of surgery and radiotherapy was similar across SES groups. However, we found that the administration of adjuvant chemotherapy varied across the social strata among stage II patients (8% of the lowest SES group versus 17% of the highest SES group, P < 0.001). Patients with a higher SES seem to have benefited more from the general trend towards more adjuvant chemotherapy independent of the disease stage. This may explain, at least in part, the diverging trend in survival rates. Another explanation for differential survival could be socio-economic variations in lifestyle. Smoking has become relatively more prevalent among low SES groups [25, 26]. This may have had an adverse effect on survival due to a poor general health while undergoing breast cancer treatment or to smoking related diseases (such as chronic obstructive pulmonary diseases (COPD) or cardiovascular disease). Also related to an unhealthy lifestyle is obesity, which has become an increasingly important problem in the last decade, [27, 28] especially among women from the lower social classes [29]. Serious concomitant diseases besides breast cancer also affect survival rates, [30] which may explain differences in survival if comorbidity occurs more frequently in low SES groups. Since the Eindhoven Cancer Registry has recorded comorbidity for all newly diagnosed patients since 1993, we checked whether the prevalence varied across socioeconomic strata. Indeed, the proportion of patients with comorbidity was higher among those with a lower SES (70% of patients in the lowest SES group had one or more concomitant conditions compared to 60% of the high SES group). In particular, the prevalence of diabetes and cardiovascular disease was highest in the low SES groups (diabetes in 10% with low SES and 4% with high SES, cardiovascular disease 7% and 4%, respectively). Several studies have reported increased survival rates after the introduction of breast cancer screening 31-35. As far as we know, no studies describe a differential effect of the introduction of screening on survival rates for socio-economic strata. However, socioeconomic inequalities in mortality have been widening in recent decades in western European countries [36]. In fact, socio-economic differences in breast cancer mortality increased between 1983 and 1993 among women in Finland and Italy (Turin), but remained stable in Denmark and decreased somewhat in Norway where a mass screening programme was only introduced later [36, 37]. In conclusion, despite a very high participation rate women from lower socio-economic strata clearly benefited less from the introduction of the breast cancer screening programme than those with a lower SES, probably due to a higher prevalence of comorbidity and suboptimal treatment (for both the cancer and the concomitant disease).

Eur_J_Clin_Pharmacol-3-1-1914266

Coumarin anticoagulants and co-trimoxazole: avoid the combination rather than manage the interaction

Objective The objective of our study was to examine the management of the interaction between acenocoumarol or phenprocoumon and several antibiotics by anticoagulation clinics and to compare the consequences of this interaction on users of co-trimoxazole with those for users of other antibiotics. Introduction Coumarin-type anticoagulants have a narrow therapeutic range. One important aspect of their safety is their sensitivity to drug interactions, many of which have been described [9, 11]. There are several reasons why antibiotic use can be considered to be indicative of a change in anticoagulation status in users of coumarin-type anticoagulants. When the antibiotic is used for febrile illness, it may be associated with overanticoagulation [5, 12]. In two studies on the interaction between coumarin anticoagulants and antibiotics, the risk of severe overanticoagulation, defined as an International Normalised Ratio (INR) ≥6.0, was increased more in users of sulfamethoxazole-trimethoprim (co-trimoxazole) than in users of other antibiotics [12, 21]. Sulfamethoxazole is a strong inhibitor of CYP2C9 [24], the main liver enzyme involved in the metabolism of warfarin [13], acenocoumarol [19] and probably phenprocoumon [20], which could explain this stronger association with overanticoagulation. Current clinical guidelines in The Netherlands for the management of coumarin drug interactions advise healthcare givers to avoid prescribing the concurrent use of co-trimoxazole and coumarins [1]. Nevertheless, in daily practice co-trimoxazole is frequently prescribed to users of coumarins, since physicians in anticoagulation clinics assume that an interaction with co-trimoxazole can be managed in a manner similar to those used to manage interactions that arise with the concurrent use of coumarins with other antibiotics. An anticoagulation clinic will initiate one of the following procedures once it has been notified of the initiation of the use of an antibiotic: (1) measurement of the INR during the antibiotic course and adjustment of the coumarin dose depending on the INR value (a reactive dose-adjustment); (2) a preventive (coumarin) dose reduction (PDR) preceding an INR measurement during or after the antibiotic course, assuming that use of an antibiotic or the intercurrent infection itself increases the risk of overanticoagulation. The PDR approach seems even more relevant to co-trimoxazole than to other antibiotics because the CYP2C9-inhibiting effect of the former might increase the risk of overanticoagulation more than the infectious state alone. However, PDR could also lead to temporary undertreatment, and evidence for the effectiveness of this approach is currently lacking. There are no official guidelines for such dose adjustments, and the application of PDR strongly depends on the personal view of the responsible physician. The aim of the present study was to examine the management of the interaction between coumarin anticoagulants and antibiotics by anticoagulation clinics and its consequences for users of co-trimoxazole and other antibiotics. To this end, we conducted a prospective follow-up study at four anticoagulation clinics in The Netherlands. Materials and methods Study design This was a follow-up study conducted at four anticoagulation clinics in The Netherlands. We included patients who were stabilised on one of the coumarin anticoagulants acenocoumarol or phenprocoumon and who had started using one of the following antibiotics between January 2001 and October 2003: co-trimoxazole, amoxicillin, amoxicillin-clavulanic acid, clarithromycin, doxycyclin, nitrofurantoin, norfloxacin or trimethoprim. In addition to co-trimoxazole, we chose the other antibiotics based on their use for the same kind of infections, mainly those of the urinary and respiratory tract. The subjects included in our study were prospectively followed during the antibiotic course until the last INR measurement, which occurred within 6 weeks following the starting date of the antibiotic (follow-up time). We did not intervene in the daily routine of the participating anticoagulation clinics and, in particular, we made no agreements on checking the INR of patients during the antibiotic course, on making additional INR measurements, on the time intervals between INR measurements after the antibiotic courses or on dose adjustments when antibiotics were prescribed. To assess the consequences of interaction management reliably and to avoid confounding by an unstable anticoagulation status preceding the antibiotic course, we only included stabilised patients in our study. Criteria for the assessment of stability were: (1) use of the coumarin anticoagulant for at least 50 days before initiation of the antibiotic; (2) availability of at least four INR measurements before the initiation of the antibiotic; (3) the last two INR measurements before initiation of the antibiotic were within the therapeutic range; (4) a maxim of one out of the last four INR measurements or a maxim of 30% of the INR measurements during the 50 days immediately preceding initiation of the antibiotic were outside of the therapeutic range, with no INR being above 5.5. Similar criteria for stability have been used in other studies [12, 21]. We excluded subjects from our analyses in whom the INR was not measured during the course of the antibiotic and who used the antibiotic for a period shorter than 3 days and longer than 14 days. If the INR was not measured during the course, an interaction effect of the antibiotic could be missed. Antibiotics used for less than 3 days or more than 14 days are usually prescribed for prophylaxis not for acute infections. All patients were informed of the aims of the study and were asked for their written consent to participate in the study. Setting and attitudes of anticoagulation clinics on antibiotic use All anticoagulation clinics in The Netherlands monitor the INR in outpatients at a frequency varying from a few days to maximally 6 weeks. The two target therapeutic ranges are the normal therapeutic range (INR: 2.0–3.5) and the high therapeutic range (INR: 2.5–4.0). The initiation of the use of an antibiotic is usually reported to the anticoagulation clinics by the patients, their pharmacists and/or the prescribing physicians. The four anticoagulation clinics participating in this study had different attitudes on the management of the interaction between coumarins and antibiotics. The approach of three of the anticoagulation clinics was to decrease the coumarin dose preventively if co-trimoxazole was prescribed; one of the anticoagulation clinics applied a PDR of 20–25% in the case of co-trimoxazole use. If one of the other antibiotics examined in this study was prescribed, the application of a PDR would depend on the seriousness of the disease and on the occurrence of fever. The fourth anticoagulation clinic had no established protocols for dose reduction but indicated that it would monitor the INR of every user of co-trimoxazole within 3–5 days after initiation of the course. Data collection We collected relevant data on the participating patients and recorded these in a database: sex and age of patient; dosage and indication of the coumarin; prescribed antibiotics (indication, dosage and duration of use); results of INR measurements before, during and after the antibiotic course; co-medication; relevant co-morbidities (malignancies, thyroid diseases, heart failure). These data were retrieved from the medical files of the anticoagulation clinics. Patients were asked to indicate on a questionnaire for which infection the antibiotic was prescribed and whether they had suffered from fever during the antibiotic course. We recorded this as fever yes/no in our database. If the coumarin-dose was reduced as soon as the antibiotic was started in the absence of an actual INR, we recorded this as a preventive dose-reduction and calculated the percentage of the dose reduction from the data on dosage in the file of the anticoagulation clinic. In order to assess the anticoagulation status shortly after the antibiotic course, we recorded the time spent within, above and under the therapeutic range from the starting date of the antibiotic until the last INR measurement within 6 weeks following the starting date of the antibiotic. Six weeks is the maximal period between two INR measurements if a patient is well stabilised. Furthermore, after a longer follow-up period, differences between patients could be more attributable to other factors than to the infection or antibiotic use. If after the first INR during the antibiotic course no second INR measurement was available within the 6-week period after the starting date of the antibiotic, we recorded no follow-up time and no time spent within, above or under the therapeutic range. Outcomes The end points of our study were chosen to assess the effectiveness of the management of the interaction between coumarin anticoagulants and co-trimoxazole and other antibiotics. We examined the following parameters in users of co-trimoxazole with and without PDR as well as in users of other antibiotics with and without PDR: occurrence of moderate overanticoagulation (INR >4.5) and severe overanticoagulation (INR >6.0);time spent within, above and under the therapeutic INR range from the starting date of the antibiotic until the last INR measurement within 6 weeks following the starting date of the antibiotic. Calculations and statistical analysis We assessed the effects of the PDR within the group of users of co-trimoxazole and within the group of users of other antibiotics by comparing the occurrence of overanticoagulation in patients for whom a PDR had been applied with the occurrence of overanticoagulation in patients for whom PDR had not been applied (logistic regression models). We also compared the occurrence of overanticoagulation and time spent within, under and above the therapeutic range of co-trimoxazole users with users of other antibiotics (reference group). These comparisons were made for patients with PDR and for patients without PDR. Finally, we compared the time spent within, under and above the therapeutic range in patients for whom a PDR had been applied with those for whom a PDR had not been applied (reference) within the groups of co-trimoxazole users and users of other antibiotics (linear regression models). In all models we adjusted for the potential confounding covariates sex, age, target therapeutic range and fever, as indicated by the patient. Covariates were added to the statistical models one at a time. We adjusted for a covariate if it changed the point estimation of the outcome of interest by 5% or more upon inclusion in the model. Time spent within, above and under the therapeutic INR range was calculated by the step-up method described by Rosendaal et al. [15]. Although all patients were stable when they were included in our study, we re-analysed our statistically significant outcomes after excluding patients in whom destabilisation could be due to factors other than those of infection and/or fever (presence of thyroid disease, malignancy or use of other enzyme-inhibiting or-inducing drugs). All statistical analyses were performed with the statistical software package SPSS ver. 12 (SPSS, Chicago, Ill.). Results A total of 424 patients who met the inclusion criteria gave their informed consent to participate in our study. Of these patients, 81 did not have assesment of the INR during the antibiotic course, 14 used the antibiotic for less than 3 days, and 3 used the antibiotic for more than 14 days. A PDR was applied more frequently for users of co-trimoxazole (28/43; 65%) than for users of other antibiotics (60/283; 21.2%) (Table 1). Table 1Characteristics of patients (n = 326) using antibiotics, treated by four anticoagulation clinicsCharacteristicCo-trimoxazole (n = 43)Other antibioticsa (n = 283)PDR+b (n = 28)PDR-c (n = 15)PDR+b (n = 60)PDR-c (n = 223)Men, no. (%)22 (78.6)10 (66.7)30 (50.0)114 (51.1)Age in years, mean (SD)75.4 (10.9)75.1 (8.2)72.6 (10.9)71.4 (11.2)Users of acenocoumarol, no. (%)24 (85.7)10 (66.7)52 (86.7)169 (75.8)Follow-up time, mean (SD) 33.2 (5.6)28.9 (8.0)30.4 (7.2)30.2 (7.2)Fever, no. (%)18 (64.3)6 (40.0)27 (45.0)120 (53.8)Normal target therapeutic range, no. (%)d19 (67.9)5 (33.3)30 (50.0)112 (50.2)Respiratory infections, no. (%)8 (28.6)3 (20.0)33 (55.0)116 (52.0)Urinary tract infections, no. (%)13 (46.4)8 (53.3)11 (18.3)53 (23.8)Malignancies, no. (%)1 (3.6)3 (20.0)3 (5.0)9 (4.0)Thyroid diseases, no. (%)001 (1.7)11 (4.9)Users of inhibiting drugs, no. (%)01 (6.7)5 (8.3)17 (7.6)Users of inducing drugs, no. (%)1 (3.6)01 (1.7)4 (1.8)INR measurements, mean no. (SD)3.5 (0.9)3.9 (1.5)3.5 (1.1)3.1 (1.1)Acenocoumarol, mean dose, mg/day (SD)2.42 (1.26)2.41 (1.41)2.61 (1.06)2.60 (1.12)Percentage PDR applied, mean (SD) in acenocoumarol users15.0 (7.6)10.3 (11.1)Phenprocoumon, mean dose, mg/day (SD)2.81 (0.86)2.53 (1.02)2.99 (1.31)2.36 (1.00)Percentage PDR applied, mean (SD) in phenprocoumon users17.9 (15.8)11.4 (7.0)Percentage PDR applied, all coumarins, mean (SD)15.4 (8.8)10.5 (10.6)aOther antibiotics: Trimethoprim (n = 3), doxycyclin (n = 104), amoxicillin (n = 77), amoxicillin-clavulanic acid (n = 36), clarithromycin (n = 14), norfloxacin (n = 33), nitrofurantoin (n  =  16)bPDR+, Preventive dose reduction appliedcPDR-, Preventive dose reduction not applieddNormal target therapeutic range: INR 2.0–3.5 The PDR applied was significantly greater in users of co-trimoxazole than in users of other antibiotics (15.0 and 10.3%, respectively; P value for difference: 0.036; two-sided t-test). The number of INR measurements during follow-up was significantly higher in both users of co-trimoxazole (PDR applied and PDR not applied) and users of other antibiotics (PDR applied) than in users of other antibiotics in whom a PDR was not applied. (P values of 0.028, 0.006 and 0.007, respectively; two-sided t-test). Mean daily dosages for acenocoumarol were lower in users of co-trimoxazole than in users of other antibiotics, but this difference was not statistically significant and even smaller (0.14 mg) after adjustment for differences in age (Table 1). In co-trimoxazole users, the PDR protected strongly against both moderate and severe overanticoagulation [adjusted odds ratio (OR): 0.06, 95% confidence interval (CI): 0.01–0.51 for INR >4.5; adjusted OR: 0.09, 95% CI: 0.01–0.92 for INR >6]. For other antibiotics, the effect of the PDR on overanticoagulation was not as strong and not statistically significant (Tables 2 and 3). Table 2Occurrence of overanticoagulation and time spent within, above and under the therapeutic range by patients using co-trimoxazole and other antibioticsaOutcomeCo-trimoxazoleOther antibioticsPDR+b (n = 28)PDR-b (n = 15)PDR+b (n = 60)PDR-b (n = 223)INR >4.5, no. (%)3 (10.7)25 (89.3)9 (15.0)45 (20.2)dINR > 6.0, no. (%)1 (3.6)4 (26.7)5 (8.3)14 (6.3)eTime within therapeutic range, mean % (95%CI)c71.1 (60.4–81.8)51.8 (34.6–69.0)76.2 (69.5–82.9)75.7 (72.2–79.3)fTime above therapeutic range, mean % (95%CI)15.0 (5.7–24.3)20.3 (10.7–29.8)12.3 (6.8–17.7)18.9 (15.5–22.2)gTime under therapeutic range, mean % (95%CI)14.0 (5.6–22.2)27.9 (7.7–48.1)11.5 (6.9–16.1)5.4 (3.7–7.2)ha Calculated for the time from the date the antibiotic was first taken until the last INR measurement within 6 weeks following the starting date of the antibiotic. Time within, above and under therapeutic range was calculated for antibiotic users in whom at least one INR measurement had been performed within 6 weeks after the INR measurement during the antibiotic course. This resulted in the exclusion from the analysis of the following number of subjects: co-trimoxazole (1 1); other antibiotics 14 (1 13). The numbers in parenthesis indicate the number of subjects with a PDR and those without a PDR, respectively.bPDR+, Preventive dose reduction applied; PDR-, preventive dose reduction not applied.c95% CI, 95% Confidence interval for the reported mean valuedRange: 7.1% for nitrofurantoin to 27.6% for norfloxacineRange: 0% for nitrofurantoin and trimethoprim to 8.6% for doxycyclinfRange: 71.8% for amoxicillin to 84.8% for norfloxacingRange: 8.6% for norfloxacin to 22.0% for amoxicillinhRange: 3.3% for norfloxacin to 11.1% for clarithromycinTable 3Odds ratios for effect of preventive dose reduction (PDR) and for (severe) overanticoagulation in users of co-trimoxazole compared with users of other antibiotics Odds ratios (95%CI)PAdjusted Odds ratios (95% CI)aPEffect of PDR on overanticoagulationb Co-trimoxazole   PDR applied, INR >4.50.10 (0.02–0.50)0.005*0.06 (0.01–0.51)0.010*   PDR applied, INR >6.00.10 (0.01–1.02)0.0510.09 (0.01–0.92)c0.042*   PDR not appliedReferenceReference Other antibiotics   PDR applied, INR >4.50.70 (0.32–1.52)0.37N.A.d   PDR applied, INR >6.01.36 (0.47–3.93)0.57N.A   PDR not appliedReferenceRisk of overanticoagulation PDR not applied   Co-trimoxazole, INR >4.54.52 (1.56–13.1)0.006*3.96 (1.33–11.8)0.013*   Co-trimoxazole, INR >6.05.43 (1.53–19.2)0.009*3.86 (1.03–14.6)0.046*   Other antibioticsReferenceReference PDR applied   Co-trimoxazole, INR >4.50.68 (0.17–2.73)0.59N.A.   Co-trimoxazole, INR >6.00.41 (0.04–3.66)0.420.30 (0.03–3.05)e0.30   Other antibioticsReferenceReference*Statistically significant difference at P ≤ 0.05aAdjusted for differences in fever as indicated by patient, age, sex, target therapeutic range, unless otherwise indicatedbPDR, Preventive dose reductioncAdjusted for differences in age and sexdN.A., Adjustment not applied because the inclusion of covariates in our model did not result in a change of at least 5% in the odds ratios (see text)eAdjusted for differences in age, sex and fever as indicated by patient If PDR was applied, the risk of overanticoagulation was not increased in users of co-trimoxazole compared with users of other antibiotics. However, if PDR was not applied, there was a strongly increased risk of moderate as well as severe overanticoagulation in co-trimoxazole users compared with users of other antibiotics (adjusted OR: 3.96, 95% CI: 1.33–11.8 for INR >4.5; adjusted OR: 3.86, 95%CI: 1.03–14.6 for INR >6.0) (Tables 2 and 3). During the 6-week follow-up, co-trimoxazole users with a PDR spent more time within and less time under the therapeutic range than co-trimoxazole users without a PDR, but these differences were not statistically significant. Users of co-trimoxazole without a PDR spent significantly less time within the therapeutic range than users of other antibiotics with a PDR, whereas significantly more time was spent under the therapeutic range. Moreover, co-trimixazole users with a PDR also spent significantly more time under the therapeutic range than did all users of other antibiotics (adjusted mean difference: 6.9%; 95%CI: 1.0–12.9) (Tables 2 and 4). Table 4Comparisons of time spent within, under and above the therapeutic range by users of co-trimoxazole and other antibioticsaMean differencePAdjusted mean differencebPPDR applied  Co-trimoxazole, % time within TRc−5.1 (−17.2 to 6.9)0.40−4.2 (−17.1 to 8.6)0.51  Co-trimoxazole, % time above TR2.7 (−7.3 to 12.7)0.601.9 (−8.9 to 12.7)0.73  Co-trimoxazole, % time under TR2.4 (−6.3 to 11.1)0.58N.A.d  Other antibioticsReferenceReferencePDR not applied  Co-trimoxazole, % time within TR−23.8 (−38.2 to –9.6)< 0.001*−22.3 (−36.6 to –8.0)e0.002*  Co-trimoxazole, % time above TR1.4 (−11.7 to 14.5)0.83N.A.  Co-trimoxazole, % time under TR22.5 (14.4 to 30.6)< 0.001*20.4 (12.4 to 28.5)f< 0.001*  Other antibioticsReferenceReferenceCo-trimoxazole  PDR applied, % time within TR19.3 (0.7 to 37.9)0.042*14.6 (−5.8 to 35.1)f0.16  PDR applied, % time above TR−5.3 (−19.6 to 9.0)0.46−4.4 (−19.1 to 10.3)g0.55  PDR applied, % time under TR−14.0 (−31.6 to 3.6)0.16−10.7 (−29.0 to 7.5)g0.24  PDR not appliedReferenceReferenceOther antibiotics  PDR applied, % time within TR0.5 (−7.0 to 8.1)0.890.6 (−6.8 to 8.2)0.87  PDR applied, % time above TR−6.6 (−13.5 to 0.3)0.061N.A.  PDR applied, % time under TR6.1 (2.0 to 10.1)0.003*N.A.  PDR not appliedReferenceReferenceCo-trimoxazole, PDR applied  % time within TR−4.7 (−15.1 to 5.6)0.37−3.0 (−13.5 to 7.5)f0.58  % time above TR−2.4 (−11.9 to 7.1)0.61−3.7 (−13.3 to 6.0)h0.46  % time under TR7.2 (1.2 to 13.1)0.018*6.9 (1.0 to 12.9)h0.022*  Other antibiotics     PDR applied + PDR not appliedReferenceReference*Statistically significant difference at P ≤ 0.05aCalculated for the time from the starting date of the antibiotic until the last INR measurement within 6 weeks following the starting date of the antibiotic.bAdjusted for differences in fever as indicated by patient, age, sex, target therapeutic range, unless otherwise indicated.cTR, Therapeutic range;dN.A., Adjustment not applied because including covariates in our model did not result in a change of at least 5 % of mean difference (see text).eAdjusted for differences in sex.fAdjusted for differences in sex and target therapeutic range.gAdjusted for differences in age, sex and target therapeutic range.hAdjusted for differences in sex, target therapeutic range and fever as indicated by patients. Co-trimoxazole users with more than a 20% PDR spent significantly more time under the therapeutic range than users of other antibiotics (adjusted mean difference: 7.4 mg; 95%CI: 0.9–14.0; P = 0.027). If less than a 20% PDR was applied, the difference between the users of co-trimoxazole and those of other antibiotics shrunk and was no longer significant. The application of a PDR differed between anticoagulation clinics. Three of the four anticoagulation clinics participating in this study applied PDR as a rule in co-trimoxazole users (83.3–85.7%). In terms of users of other antibiotics, the application of a PDR was more varied: in three of the anticoagulation clinics PDR was sometimes applied (in 17.6–50.8% of all cases), whereas one anticoagulation clinic did not apply the PDR approach at all. The overall percentage of time spent within the therapeutic range during the first 6 weeks after initiation of an antibiotic ranged from 73.7 to 78.0% at all four anticoagulation clinics. In the anticoagulation clinic that did not apply a PDR, overanticoagulation (INR>4.5) occurred most frequently for the all antibiotics class (26.9 vs.10.8–22.7% in the other clinics), with the difference being most marked for co-trimoxazole (54.4 vs. 14.3–16.7% in the other clinics). We also analysed our data separately for users of acenocoumarol and phenprocoumon. There were no differences in the point estimates of most of our main outcomes between users of either of these coumarins, with the exception of percentage of time spent under the therapeutic range in phenprocoumon users for whom PDR was applied. However, most of the results that were statistically significant for all coumarin users were also significant for users of acenocoumarol (n = 252; 78.2 %), whereas they were in most cases not significant for the smaller group of users of phenprocoumon (n = 71; 21.8%) (Table 5). Table 5Main outcomes stratified for users of acenocoumarol and phenprocoumonOutcomeAcenocoumarol (n = 254)Phenprocoumon (n = 71)Protective effect of PDRAdjusted OR (95%CI)aPAdjusted OR (95%CI)aPCo-trimoxazole   PDR applied, INR >4.50.08 (0.01–0.70)0.022*0.16 (0.01–4.48)>0.3b   PDR not appliedReferenceReferenceRisk of overanticoagulation   PDR not applied   Co-trimoxazole, INR >4.54.40 (1.15–16.8)0.030*3.83 (0.55–26.7)0.18   Other antibioticsReferenceReference% Time within or under TRMean difference (95%CI)Mean difference (95%CI)   Co-trimoxazole, % time within TR−22.1 (−39.1 to –5.0)0.011*−21.4 (−49.4 to 6.6)0.13   Co-trimoxazole, % time under TR20.3 (10.9–29.7)<0.001*22.6 (5.7–39.5)0.010*   Other antibioticsReferenceReference   Co-trimoxazole, PDR applied   % time under TR9.1 (3.0–15.1)0.004*−6.7 (−23.0 to 9.6)0.42   Other antibiotics      PDR applied + PDR not appliedReferenceReference*Statistically significant difference at P ≤ 0.05aAdjusted for differences in fever as indicated by patient, age, sex and target therapeutic rangebAdjustment not applied because the number of patients with INR >4.5 was zero; OR was calculated by increasing the values of each of the cells of the crosstable by 0.5 Re-analysis of our results after excluding patients with thyroid diseases and malignancy or those using enzyme-inhibiting or -inducing drugs gave similar point estimates or trends, significance for severe overanticoagulation in users of co-trimoxazole compared to other antibiotics and for time spent within the therapeutic range for users of co-trimoxazole in whom PDR was not applied (data not shown). Discussion The results of the present study, in which we evaluated the management of the interaction between antibiotics and coumarin anticoagulants by anticoagulation clinics, demonstrated that a PDR reduces the risk of overanticoagulation in co-trimoxazole users to the level of other antibiotic users, but also that management of the interaction between coumarins and co-trimoxazole results in a significantly longer period of undertreatment during the first 6 weeks after initiation of the antibiotic. In three of the four anticoagulation clinics PDR was applied more frequently and was significantly higher in users of co-trimoxazole than in users of other antibiotics, indicating that anticoagulation clinics are aware of the seriousness of the interaction between coumarins and co-trimoxazole. In the cases and case series that have reported on overanticoagulation and bleeding with the concurrent use of antibiotics and co-trimoxazole [2, 3, 6–8, 10] an effect of the intercurrent infection on the anticoagulation status could not be ruled out. However, Penning-van Beest et al. (case control study) and Visser et al. (follow-up study) both demonstrated that an increased risk of severe overanticoagulation (INR >6.0) was particularly associated with co-trimoxazole [12, 21]. A plausible explanation is the strong inhibition of the main metabolising enzyme, CYP2C9, of the coumarins by sulfamethoxazole, the sulphonamide component of co-trimoxazole [24]. Although PDRs as applied in clinical practice are effective in reducing the overanticoagulation risk in co-trimoxazole users, the price that has to be paid for the concurrent use of co-trimoxazole is a significantly prolonged period of underanticoagulation compared with the use of other antibiotics during the first 6 weeks after the antibiotic course. This difference was more marked in the subgroup of subjects in whom PDR was not applied. Possible explanations for this result are (1) the usually shorter time span between PDR and the first INR measurement (always within the course) compared to the time span between a reactive dose reduction following supratherapeutic INR and subsequent INR measurement (usually after the course) and (2) the higher reactive dose reduction which is applied in the case of severe overanticoagulation (INR >6.0). However, even co-trimoxazole users for whom the PDR had been applied had a significantly prolonged period of underanticoagulation compared with all of the users of other antibiotics (PDR applied and PDR not applied taken together). This last comparison is totally logical because our results strongly suggest that a PDR should always be applied in co-trimoxazole users, whereas this is as a rule not required in users of other antibiotics. The adjusted difference in time spent under the therapeutic range – ranging from 6.9 (PDR applied) to 22.5% (PDR not applied) – corresponds to about 2–7 days of the mean follow-up time of 30 days in otherwise stabilised patients; this time interval is clinically relevant and can be avoided by substituting co-trimoxazole. It is not difficult to explain the prolonged period of underanticoagulation in co-trimoxazole users. The application of a PDR, which was in this study higher in co-trimoxazole users, might overcompensate for overanticoagulation, whereas the reactive dose reduction following overanticoagulation carries the same risk of overcompensation and undertreatment as PDR. Consequently, the inhibition of CYP2C9 by co-trimoxazole superimposes an additional problem upon the already potentially destabilising effects of the infection and fever. Because our results for acenocoumarol in the separate analyses were predominantly in agreement with the overall results, our findings primarily apply to acenocoumarol users. It is possible that users of phenprocoumon are less sensitive to interactions with CYP2C9 inhibitors such as co-trimoxazole [20, 22]. We do expect that our results also apply to users of warfarin, which seems to be even more CYP2C9 sensitive than acenocoumarol [18]. Our study has several limitations. Because we retrieved medical data from anticoagulation clinics, it is possible that not all of the relevant data on potentially destabilising factors, such as malignancies, thyroid diseases and the use of other inhibitors of coumarin metabolism, were available. However, by only including patients who were obviously stable at the moment of initiation of the antibiotic, we decreased the chance that such factors changed the anticoagulant status during the antibiotic course. A second limitation is the absence of data on the presence of polymorphisms of the genes encoding the coumarin-metabolising enzyme CYP2C9 or the pharmacodynamic target of coumarins, VKORC1. The genotypes of both CYP2C9 [16, 22] and VKORC1 are strongly associated with interindividual variability in coumarin dose requirements [4, 14, 17, 23]. Further studies would be needed to assess whether the risk of overanticoagulation in co-trimoxazole users differs between carriers of a CYP2C9 or VKORC1 polymorphism and wild-type patients. It should be clear that our results only apply to patients with a stabilised anticoagulation state at the initiation of the antibiotic course. In conclusion, if co-trimoxazole is prescribed to users of coumarin anticoagulants, the interaction can be managed by applying PDR, which adequately decreases the risk of overanticoagulation, but this successful management comes at the cost of a prolonged period of underanticoagulation after the course. Consequently, rather than managing the interaction it is better to avoid prescribing co-trimoxazole as a therapeutically equivalent alternative is always available.

Appl_Microbiol_Biotechnol-3-1-2039825

Properties, production, and applications of camelid single-domain antibody fragments

Camelids produce functional antibodies devoid of light chains of which the single N-terminal domain is fully capable of antigen binding. These single-domain antibody fragments (VHHs or Nanobodies®) have several advantages for biotechnological applications. They are well expressed in microorganisms and have a high stability and solubility. Furthermore, they are well suited for construction of larger molecules and selection systems such as phage, yeast, or ribosome display. This minireview offers an overview of (1) their properties as compared to conventional antibodies, (2) their production in microorganisms, with a focus on yeasts, and (3) their therapeutic applications. Introduction The field of recombinant antibody technology has rapidly progressed during the last two decades, mainly because of the interest in their human therapeutic use. The ability to select specific human antibodies by display technologies and to improve their affinity, stability, and expression level by molecular evolution has further boosted the field. Whole antibodies are complex molecules that consist of heavy and light chains (Fig. 1a). They contain an N-linked oligosaccharide attached to the second heavy-chain constant domain (CH2) that is essential for antibody effector functions such as antibody-dependent cellular cytotoxicity (ADCC), complement-dependent cytolysis (CDC), and for retaining a long serum half-life. Fig. 1Schematic diagram of conventional (a) and heavy-chain (b) antibodies and fragments thereof. Variable domains derived from the antibody heavy (VH) and light (VL) chains are shaded dark gray and light gray, respectively, whereas constant domains (CH and CL) are not shaded. Note the absence of the light chain and CH1 domain in heavy-chain antibodies. Antibody domains that pair by noncovalent interactions are indicated by overlaying them. The B-subunits of naturally pentamerizing toxins that are used to generate pentabodies are indicated as hatched spheres Although isolated antibody heavy (Utsumi and Karush 1964) and light chains (Yoo et al. 1967) can retain antigen-binding specificity, their affinity and solubility is often reduced (Ward et al. 1989). However, the paired N-terminal variable domains of heavy (VH) and light (VL) chains are sufficient for antigen binding (Sundberg and Mariuzza 2002). Such antibody fragments can be produced as monovalent antibody fragment (Fab) or as single-chain Fv (scFv) where the VH and VL domains are joined by a polypeptide linker (Fig. 1a). Their production in microbial cells is often cumbersome, especially when producing multivalent formats, because of the requirement for domain association. The discovery that camelids (bactrian camels, dromedaries, and llamas) produce functional antibodies devoid of light chains (Hamers-Casterman et al. 1993) formed a further breakthrough because their single N-terminal domain (VHH, also referred to as Nanobody®) binds antigen without requiring domain pairing. These heavy-chain antibodies also lack the CH1 domain, which in a conventional antibody associates with the light chain and to a lesser degree interacts with the VH domain (Fig. 1b). Although single-domain antibodies later were also identified in particular cartilaginous fish (Greenberg et al. 1995), most research on the biotechnological application of single-domain antibodies was done using camelids because of their ease of handling, including immunization. Methods to isolate antigen-specific VHHs from immune (Arbabi-Ghahroudi et al. 1997; Van der Linden et al. 2000a), nonimmune (Tanha et al. 2002; Yau et al. 2003; Verheesen et al. 2006), or semisynthetic (Goldman et al. 2006) libraries using phage, yeast, or ribosome display are now well established. For further reading on these topics, we refer to recently published reviews (Muyldermans 2001; Dufner et al. 2006). Properties Sequence analysis (Hamers-Casterman et al. 1993; Muyldermans et al. 1994; Vu et al. 1997; Harmsen et al. 2000) and elucidation of the crystal structure (Desmyter et al. 1996, 2001; Spinelli et al. 1996) has revealed several structural features of VHH domains. Similar to conventional VH domains, VHHs contain four framework regions (FRs) that form the core structure of the immunoglobulin domain and three complementarity-determining regions (CDRs) that are involved in antigen binding. This contrasts with shark single-domain antibodies that have a vestigial CDR2 that does not contribute to antigen binding (Streltsov et al. 2004). As compared to human VH domains, the VHH FRs show a high sequence homology of more than 80%, and their 3D structures can be superimposed (Muyldermans et al. 2001; Holliger and Hudson 2005). The most characteristic feature of VHHs is the presence of amino acid substitutions at four FR2 positions (positions 37, 44, 45, and 47; Kabat numbering) that are conserved in conventional VH domains and that are involved in forming the hydrophobic interface with VL domains. Occasionally, antigen-binding single-domain antibody fragments that lack these characteristic FR2 substitutions are isolated from camelids. These fall into two groups. The low-affinity binders isolated from a nonimmune library originated from conventional antibodies, presumably because of the polymerase chain reaction crossover cloning artifact, as they were linked to the CH1 domain (Tanha et al. 2002). We refer to these as VHH-like conventional VHs. However, such single-domain antibody fragments with conventional-like FR2 sequences that bind antigen with high affinity are isolated from immune libraries with the high efficiency of about 10% (Conrath et al. 2001a; Saerens et al. 2004; Harmsen et al. 2005a, 2007), which equals their presence in unselected libraries (Harmsen et al. 2000). This is not expected when such clones originate from a cloning artifact. Unlike the clones isolated by Tanha et al. (2002), these clones often contain a hydrophilic residue (mostly arginine) at position 103. This substitution is probably important for their single-domain nature (Desmyter et al. 2001) because conventional antibodies contain a highly conserved hydrophobic residue (tryptophan) at this position that contacts VL. This suggests that these represent functional VHH domains derived from recombination of conventional VH gene segments with heavy-chain constant gene segments during B cell maturation. This was confirmed by the absence of the CH1 domain when such VHH domains were reisolated from the original immune repertoire using a CDR3-specific primer (De Haard, unpublished observation). Therefore, we refer to these as conventional-like VHH domains. Although the increased hydrophilicity of VHHs predominantly relies on the aforementioned changes in the former VL interface, some amino acids at positions that form a slightly hydrophobic patch on conventional VH domains that contacts CH1 (Lesk and Chothia 1988) are also changed into hydrophilic residues in VHHs (Muyldermans et al. 1994; Harmsen et al. 2000). Furthermore, the CDRs of VHHs contain some characteristic features. Firstly, the N-terminal part of CDR1 is more variable (Vu et al. 1997; Harmsen et al. 2000; Nguyen et al. 2000). Secondly, many dromedary VHHs have an extended CDR3 that is often stabilized by an additional disulfide bond with a cysteine in CDR1 or FR2 (Muyldermans et al. 1994) resulting in the folding of the CDR3 loop across the former VL interface (Desmyter et al. 1996). A particular subfamily of llama VHHs (VHH3) also contains an extended CDR3 that is stabilized by an additional disulfide bond with a cysteine at position 50 in FR2. However, VHHs of this subfamily are rarely isolated, and most llama VHHs have CDR3 loops similar in length to those found in human VHs. VHHs have many advantages for biotechnological applications, which are summarized in Table 1. An important advantage is their high microbial production level (see next section). Table 1Advantages of camelid single-domain antibody fragments as compared to conventional antibody fragmentsAdvantageMolecular basisFacile genetic manipulationSingle-domain natureIncreased functional size of immune librariesNo decrease in library size because of reshuffling of VL and VH domainsFacile production of multivalent formatsMore flexible linker design and no mispairing of VL and VH domainsFacile production of oligoclonal preparations from single cellsNo mispairing of VL and VH domainsHigh physicochemical stabilityEfficient refolding due to increased hydrophilicity and single-domain natureHigh solubilityIncreased hydrophilicityRecognition of hidden antigenic sitesSmall size and extended flexible CDR3Rapid tissue penetration, fast clearanceSmall sizeWell expressedEfficient folding due to increased hydrophilicity and single-domain natureSee text for references Several advantages result from their single domain nature. Thus, VHH libraries generated from immunized camelids retain full functional diversity. This contrasts with the diminished diversity of conventional antibody libraries because of reshuffling of VL and VH domains during library construction. As a result, high-affinity antigen-binding VHHs can be isolated by directly screening a limited number of clones from immune libraries without prior selection using display technologies (Frenken et al. 2000; Harmsen et al. 2005b). Furthermore, the single-domain nature facilitates subsequent molecular manipulation. For example, for many applications, it is advantageous to engineer monovalent antibody fragments into multivalent formats to increase functional affinity (termed avidity) or to produce bispecific antibody fragments that can simultaneously bind to different antigens. Such molecules (diabodies, Fig. 1a) can be produced using conventional recombinant antibodies using linkers between the VH and VL domains of a specific length, although this often results in aggregation and reduced affinity because of mispairing of VH and VL domains (Glockshuber et al. 1990; Whitlow et al. 1993). VHHs are more suitable for production of such formats because they allow more flexible linker design, which is important for simultaneous binding of multivalent antigens, without the problems posed by domain mispairing. Thus, several functional trivalent-bispecific VHHs have been successfully produced (Coppieters et al. 2006; Roovers et al. 2007). The use of mixtures of a limited number of monoclonal antibodies (oligoclonal antibodies) is advantageous over single monoclonal antibodies for particular applications, such as toxin neutralization (Nowakowski et al. 2002). Because of regulatory requirements, such oligoclonals are preferentially produced from single cells. Again, VHHs are predicted to be more suitable for single-cell production of oligoclonals because of the absence of domain mispairing, although this is yet to be demonstrated experimentally. Contrary to conventional antibodies, VHHs have been shown to remain functional at 90°C (Van der Linden et al. 1999) or after incubation at high temperatures (Van der Linden et al. 1999; Perez et al. 2001). This high apparent stability is mainly attributed to their efficient refolding after chemical or thermal denaturation and to a lesser extent because of an increased resistance against denaturation (Perez et al. 2001; Dumoulin et al. 2002; Ewert et al. 2002). The increased apparent stability is probably due to an increased hydrophilicity of the former VL interface region because a “camelized” human VH fragment that contains several of the hallmark hydrophilic amino acid residues of VHHs was more stable than the original VH fragment (Davies and Riechmann 1995, 1996), whereas “decamelization” of a VHH to mimic a VH domain reduces its thermodynamic stability (Conrath et al. 2005). In addition to these specific mutations, the packing of extended CDR3 loops against this former VL interface contributes to domain stability (Bond et al. 2003). Furthermore, refolding of VHHs only requires domain refolding, whereas conventional antibodies also require association of VH and VL domains. VHHs can also recognize antigenic sites that are normally not recognized by conventional antibodies such as enzyme active sites (Lauwereys et al. 1998; De Genst et al. 2006) and conserved cryptic epitopes (Stijlemans et al. 2004). This facilitates their use as enzyme inhibitors or in diagnosis of trypanosome infections. The ability to recognize these recessed antigenic sites has been attributed to their smaller size and the ability of the extended CDR3 loop to penetrate into such sites (Desmyter et al. 1996; De Genst et al. 2006). It is interesting to note that this structure–function relation is also observed in a rare example of a broadly reactive human mAb that recognizes the recessed and conserved CD4-binding cavity of human immunodeficiency virus type 1 gp120 by virtue of an extended CDR3 (Zwick et al. 2003). With respect to antigen binding, the single-domain nature could be a disadvantage for binding to small antigens such as haptens and peptides because these normally bind in a groove or cavity at the VH–VL interface (Sundberg and Mariuzza 2002). Indeed, llamas immunized with clenbuterol developed conventional but not heavy-chain antibodies against this hapten (Lange et al. 2001). However, hapten- and peptide-binding VHHs have been successfully isolated using strong selection systems (Spinelli et al. 2000; Yau et al. 2003; Alvarez-Rueda et al. 2007; Harmsen et al. 2007). The affinities of VHHs are generally comparable to those of conventional antibody fragments (Muyldermans et al. 2001). Occasionally, VHHs with affinity constants (KD) as low as 100 pM are isolated (Saerens et al. 2004; De Genst et al. 2006; Harmsen et al. 2006), which equals the affinity ceiling proposed for natural antibodies (Sundberg and Mariuzza 2002). Because of their small size of about 15 kDa, VHHs rapidly pass the renal filter, which has a cutoff of about 60 kDa, resulting in their rapid blood clearance. In addition, the small size results in a fast tissue penetration. This is advantageous for targeting of VHHs coupled to toxic substances to tumors (Cortez-Retamozo et al. 2004), in vivo diagnosis using imaging, and treatment of snake bites (Harrison et al. 2006). However, for other therapeutic applications, such as treatment of infectious or inflammatory diseases, the short serum half-life of about 2 h (Cortez-Retamozo et al. 2002; Harmsen et al. 2005a) is a disadvantage. Production in microorganisms Although a fully active nonglycosylated IgG was recently produced at high level in Escherichia coli, most functional complete antibodies can only be efficiently produced using mammalian cells, especially when their appropriate glycosylation is required for therapeutic applications. However, antibody fragments that lack the Fc with its N-linked oligosaccharide are preferably produced in microbial systems (Arbabi-Ghahroudi et al. 2005). These have a shorter development time from gene to product and require simple well-established fermentation conditions that can be performed on large-scale resulting in costs of goods that can be as low as $1 per gram heterologous protein (Estell 2006). Most large-scale microbial production systems are based on E. coli, yeasts, or filamentous fungi. Production in E. coli can be done by secretion into the oxidizing periplasmic space or expression in the reducing cytosol. The latter requires the often cumbersome refolding of antibody fragments (Arbabi-Ghahroudi et al. 2005). Using eukaryotic microbial hosts, antibody fragments are generally produced by targeting to the secretory pathway. This enables efficient disulfide bond formation, addition of N-linked oligosaccharide, and secretion of soluble, correctly folded product to the culture medium. VHHs have often been produced in E. coli (Arbabi-Ghahroudi et al. 1997; Rahbarizadeh et al. 2005). There is only one example of VHH production in filamentous fungi, which resulted in limited proteolytic degradation of the secreted product (Joosten et al. 2005) because of the high levels of proteases secreted by filamentous fungi (Gerngross 2004). VHHs have also often been produced in the yeast Saccharomyces cerevisiae (Frenken et al. 2000; Thomassen et al. 2002; Van der Vaart 2002). VHH production by the favored yeast expression host Pichia pastoris was only recently described (Rahbarizadeh et al. 2006). Occasionally, yeast-produced VHHs are N-glycosylated (Frenken et al. 2000; Harmsen et al. 2005a). This can affect antigen binding (Van der Vaart et al. 2006). Furthermore, it could complicate their therapeutic use because the addition of yeast-specific high-mannose oligosaccharides results in a high immunogenicity and decreased serum half-life because of binding to specific mannose receptors on cells of the reticulo-endothelial system (Sethuraman and Stadheim 2006). Although VHHs are generally well produced in microorganisms, the production level of different clones can vary by a factor of 100 (Frenken et al. 2000; Harmsen et al. 2005b; Van de Laar et al. 2007). Several VHH sequence patterns can be associated with their production level. First, the presence of a potential N-linked glycosylation site often increases production levels in yeast (Sagt et al. 2000). Second, in our experience (Harmsen, unpublished observations), conventional-like VHHs are generally produced at reduced levels in yeast. This contrasts with the reported efficient production in E. coli of VHH-like VHs (Tanha et al. 2002) but is consistent with the increased production level of “camelized” conventional VH domains in E. coli (Davies and Riechmann 1995). Third, the presence of unpaired C-terminal cysteines reduces expression levels (Simmons et al. 2006). Fourth, replacement of hydrophobic residues of conventional VH domains normally interacting with CH1 increased scFv production in E. coli (Nieba et al. 1997), suggesting that the hydrophilic mutations that naturally occur at these positions in VHHs also contributes to their high expression level. However, there are many examples of VHHs that differ by only a few amino acids and are produced at highly variable levels where the exact amino acid change responsible for the difference in production level is difficult to predict (Frenken et al. 2000; Harmsen et al. 2005b). Furthermore, without such knowledge, VHH production can be improved by random molecular evolution using deoxyribonucleic acid shuffling (Van der Linden et al. 2000b), as has often been done for conventional antibody fragments (Dufner et al. 2006). The high refolding capability of VHHs, which is a consequence of their sequence, has also been correlated with a high production level in E. coli (Jespers et al. 2004; Olichon et al. 2007). In addition to the nature of the VHH, host factors affecting VHH production have been identified. In baker’s yeast, the specific VHH production rate is correlated with growth rate (Thomassen et al. 2005) and can be up to fivefold increased by growing on ethanol as the carbon source (Van de Laar et al. 2007). Supplementation of the medium with sorbitol, casamino acids, or ethylenediamine tetraacetic acid improves VHH production by P. pastoris (Rahbarizadeh et al. 2006). In addition to monovalent VHHs, several expression formats for the production of VHH multimers have been described (Fig. 1b). These include genetic fusions of two (Conrath et al. 2001b; Harmsen et al. 2005a) or three VHHs (Coppieters et al. 2006; Roovers et al. 2007) that either recognize different antigens or the same repeating antigen to increase functional affinity. Although such VHH fusions are less efficiently produced than their monovalent versions, their production level exceeds that of their conventional-antibody-based fusion counterparts without aggregation or low solubility. However, antigen binding by the C-terminal VHH in such fusions can be compromised (Conrath et al. 2001b) presumably because of steric hindrance by the N-terminal VHH. The avidity of VHHs has also been strongly increased using genetic fusions to the B-subunits of an E. coli toxin that self-assembles into a homopentamer (Zhang et al. 2004), resulting in pentameric recombinant antibodies (“pentabodies,” Fig. 1b). VHHs on their own cannot recruit effector functions such as ADCC and CDC. This limits their therapeutic application. Although such effector functions can be indirectly recruited using bispecific (conventional) antibody fragments binding to host immunoglobulin (Holliger et al. 1997), it may be more efficient to recruit these functions by fusing VHHs to host Fc domains. Production of such functional antibodies requires the correct glycosylation of the CH2 domain, which until recently could only be accomplished using higher eukaryotic cells (Nguyen et al. 2003) but not by microbial production. However, this may now be feasible using P. pastoris strains with an engineered glycosylation machinery that are able to produce proteins with a specific human glycoform (Hamilton et al. 2006). Furthermore, transgenic mice containing hybrid llama/human antibody loci that contain llama V regions and human D, J, and C regions have recently been used to generate human heavy-chain antibodies in mice (Janssens et al. 2006). Therapeutic applications Although VHHs are highly suited for applications that require a high stability, such as use in shampoo for the prevention of dandruff (Dolk et al. 2005), as capturing reagents in immunoaffinity purification (Verheesen et al. 2003), or use in biosensors (Pleschberger et al. 2004), we would like to focus on their therapeutic applications, which are more challenging. Several VHHs are now being studied for use in various disease areas, including oncology (Revets et al. 2005) and in infectious, inflammatory, and neurodegenerative diseases (Table 2). Table 2Examples of therapeutic applications of camelid VHHsDiseasePathogenTarget antigenVHH valency for disease targetAdditional fusion partnerReferenceSleeping sicknessTrypanosomesVSG oligomannoseMonovalentApolipoprotein L-IBaral et al. 2006Infant diarrheaRotavirusUnknownMonovalentNoneVan der Vaart et al. 2006Infant diarrheaRotavirusUnknownMonovalentLactobacillus cell-surface anchorPant et al. 2006Piglet diarrheaE. coliF4 fimbriaeMonovalentNoneHarmsen et al. 2006CariesS. mutansI/II adhesionMonovalentNoneKruger et al. 2006FMDFMD virusVP1MonovalentPEGHarmsen et al. 2007SepsisN. meningitidisLPSMonovalentNoneEl Khattabi et al. 2006Cancer–CEAMonovalentβ-LactamaseCortez-Retamozo et al. 2004Cancer–EGF receptorBivalentAnti-albumin VHHRoovers et al. 2007Rheumatoid arthritis–TNFαBivalentAnti-albumin VHHCoppieters et al. 2006Brain disorders–α (2,3)-SialoglycoproteinMonovalentNoneMuruganandam et al. 2002Neurodegenerative diseases–BaxMonovalentNoneGueorguieva et al. 2006 VHHs are especially suited for oral immunotherapy because of their resistance against extremes of pH and the capacity to bind to the target at high concentrations of chaotropic agents (Dumoulin et al. 2002, 2003). Administration to piglets of a VHH that effectively prevents intestinal attachment of E. coli bacteria that cause diarrhea resulted in poor in vivo protection (Harmsen et al. 2005b) because of degradation by gastrointestinal proteases (Harmsen et al. 2006). However, by selection for proteolytic stability, a VHH could be isolated from the original library that was not degraded in vivo (Harmsen et al. 2006). VHHs that successfully prevented diarrhea caused by rotavirus in a mouse model were similarly selected for resistance against the acidic environment of the stomach (Van der Vaart et al. 2006). Alternatively, VHH proteolysis can be prevented by local VHH production using natural gut commensal bacteria. Thus, diarrhea could also be prevented by lactobacilli that produce rotavirus-neutralizing VHHs fused to a cell surface anchor (Pant et al. 2006). Treatment of caries, caused by Streptococcus mutans, with VHHs conferred only limited protection (Kruger et al. 2006). Because these VHHs should function in the oral cavity, the low level of protection cannot be due to proteolytic VHH degradation within the gastrointestinal tract. The short serum half-life because of a rapid renal clearance limits the efficacy of VHHs in many parenteral applications. Therefore, VHHs have been targeted to normally long-lived serum proteins such as albumin (Coppieters et al. 2006; Roovers et al. 2007) or immunoglobulin (Harmsen et al. 2005a) using bispecific VHHs recognizing these serum proteins in addition to the therapeutic target, resulting in half-lives that equal the half-life of albumin (2 days in mice) and immunoglobulin (9 days). An alternative well-known approach to increase serum half-life of proteins is the chemical addition of polyethylene glycol (PEG). Such PEGylation of foot-and-mouth disease (FMD) virus-neutralizing VHHs not only increased serum half-life but also increased in vitro neutralizing potency to levels above that of the hyperimmune serum (Harmsen et al. 2007). However, in contrast to the full protection afforded by the hyperimmune serum, these VHHs poorly protected guinea pigs from FMD viral challenge infection, suggesting that Fc-mediated effector functions are required for efficient in vivo protection (Harmsen et al. 2007). Nevertheless, many diseases were successfully treated with VHHs in the absence of Fc-mediated effector functions. These VHHs either are used as targeting devices for toxic enzymes or block a specific molecular interaction. For example, sleeping sickness was successfully treated with VHHs that bind to a trypanosome coat protein and were fused to the apolipoprotein L-1 enzyme, resulting in trypanosome lysis (Baral et al. 2006). In oncology, a VHH directed against carcinoembryonic antigen was used for targeting the genetically fused β-lactamase to tumor cells. This enzyme then converts an injected nontoxic prodrug into a toxic drug in the vicinity of the targeted tumor cells, leading to their killing (Cortez-Retamozo et al. 2004). Several VHH therapies are also being developed for treatment of oncology or inflammatory diseases based on blocking molecular interactions. VHHs binding to epidermal growth factor receptor (EGFR) can block epidermal growth factor (EGF) binding to its receptor, which can be used to treat solid tumors (Roovers et al. 2007). Tenfold more potent EGFR-binding VHHs could be obtained by construction of bivalent formats. It is interesting to note that the recently approved conventional antibody Panitumumab directed against EGFR also blocks EGF binding and is expected to give poor ADCC and CDC (Reichert and Valge-Archer 2007). Furthermore, by blocking receptor interaction, VHHs binding to tumor necrosis factor-α can be used for treatment of rheumatoid arthritis (Coppieters et al. 2006). The potency of bivalent formats was 500-fold increased as compared to monovalent VHHs and even exceeded the potency of clinically used conventional antibodies both in vitro and in a murine arthritis model. Similarly, lipopolysaccharide (LPS)-binding VHHs were isolated that block LPS binding and signaling to host cells for treatment of sepsis (El Khattabi et al. 2006). The potential immunogenicity of VHHs could compromise their parenteral therapeutic use, especially in treatments that require repeated injections. Until now, multiple injections of VHHs have not shown any immunogenicity in mice, as assessed by the presence of specific antibodies, T cell proliferation, or cytokine levels (Cortez-Retamozo et al. 2002; Coppieters et al. 2006). This could rely on their high sequence homology to conventional VH domains and on their high stability because aggregation of proteins is known to increase immunogenicity (Hermeling et al. 2004). If necessary, technologies developed to decrease immunogenicity of mouse monoclonal antibodies (Presta 2006) could also be applied to VHHs. Alternatively, immunogenicity could be reduced by the use of conventional-like VHHs, which have an even higher structural homology to conventional VH domains. For their use in targeting drugs across the blood–brain barrier (BBB) into the brain, VHHs were selected that transmigrate the human BBB in an in vitro model and accumulate in the brain after intravenous injection into mice (Muruganandam et al. 2002). These could be used for treatment of neurological disorders. Finally, Bax-specific VHHs have been expressed in the cytoplasm, resulting in so-called intrabodies, to prevent oxidative-stress-induced apoptosis that is implicated in several neurodegenerative diseases (Gueorguieva et al. 2006). Because of their stability, VHHs are especially suited for intrabody production because this requires expression in the reducing environment of the cytoplasm (Gueorguieva et al. 2006; Rothbauer et al. 2006). Conclusions Since the discovery of heavy-chain antibodies in 1993, the field of single-domain antibody fragments has been rapidly growing. VHHs have many advantages for biotechnological applications. They can be economically produced in microorganisms and have a high stability. Furthermore, they are highly suited for expression as multivalent, including bispecific, formats or as enzyme fusions. This permits a plug-and-play approach, where, depending on the target, biology potency can be increased by multivalent constructs or bispecific VHH recognizing two different targets can be made. This also enables the tailor-made design of serum half-life using site-directed PEGylation or by targeting to long-lived serum proteins using bispecific VHHs. Although fusions of targeting VHHs to Ig-binding VHHs or Fc can be used to recruit effector functions most current research on VHHs focuses on therapeutic applications where such effector functions are not required. Finally, conventional whole antibodies occasionally give side effects because of their bivalent nature, which can result in target cross-linking, or the presence of the Fc region. Evidently, such side effects are not expected to occur using monovalent VHHs. This, however, is yet to be confirmed as the first VHH has entered phase I clinical trials in 2007 (http://www.ablynx.com).

J_Headache_Pain-4-1-2245992

Performances in cerebellar and neuromuscular transmission tests are correlated in migraine with aura

In previous studies, we described subclinical abnormalities of neuromuscular transmission and cerebellar functions in migraineurs. The aim of this study was to search if these two functions are correlated in the same patient. Thirteen migraineurs [five without aura (MO) and eight with aura (MA)] underwent both stimulation-SFEMG and 3D-movement analysis. Single fiber EMG (SFEMG) results were expressed as the “mean value of consecutive differences” (mean MCD). Precision of arm-reaching movements (measured with an infrared optoelectronic tracking system) was expressed as the average deviation in the horizontal plane. Median values of mean MCD and mean horizontal deviation were not different between MO and MA. However, in MA, but not in MO, both variables were positively correlated. Thus, we conclude that neuromuscular transmission and cerebellar functions are correlated in the same patient when affected by migraine with aura. We suggest that this correlation might be due to a common molecular abnormality. Introduction Migraine is a paroxysmal neurological disorder with a high prevalence in the general population. There is strong epidemiological and genetic evidence that genetic factors play a major, though variable role, in its pathogenesis. Migraine patients are characterized between attacks on various CNS-evoked responses by a deficit of habituation [1, 2], which may have a familial character [3–5]. The lack of habituation may play a role in migraine pathogenesis [6]. By contrast, other mild abnormalities have been identified that are unlikely to play a pathogenic role: a decreased safety factor at the neuromuscular junction on single fiber EMG (SFEMG) [7–9] and subclinical cerebellar hypermetria in the horizontal plane on opto-electronic analysis of upper arm-reaching movements [10]. It was suggested that these two abnormalities might reflect the same underlying genetic abnormality. If this is correct, one may expect that in the same individual they be of a similar degree. We have therefore compared in the same migraine patient the SFEMG and 3D-movement analysis. Materials and methods Patients Thirteen migraine patients (ICHD-II) [11] were recruited from the Headache Clinic of the Headache Research Unit in Liège, Belgium. Eight patients suffering from typical aura with migraine headache (MTA–ICHD-II code 1.2.1; three women and five men; median age 28.5 years; range 13–67) and five patients with migraine without aura (MO–ICHD-II code 1.1; three women and two men; median age 40.0 years; range 31–51) underwent both stimulation-SFEMG and 3D-movement analysis. All patients were right-handed. None of them had any other medical condition detectable by history and clinical examination; none was taking drugs on a regular basis, nor had taken any drug within 3 days before the recordings. The recordings took place at least 3 days after and before an attack (checked by telephone-interview). The study was conducted after approval of our Institution’s ethics committee and performed in accordance with the ethical standards of the 1964 Declaration of Helsinki, with the understanding and consent of each involved subject. SFEMG recordings A Nicolet Viking IV device (Nicolet® Biomedical, Madison, Wisconsin, USA) was used for stimulation single fiber electromyography [12]. Single muscle fiber activity was recorded with 25-mm-long single fiber needles (Medelec Neurodiagnostic Accessories, ref: 16829, Witney, Oxfordshire, UK), and the motor nerve was stimulated with Nicolet teflon-insulated monopolar needles. We stimulated suprathreshold the motor branch of the radial nerve and assessed the variability in latency, i.e. the jitter (Fig. 1), of single fiber action potentials in m. extensor digitorum communis (EDC) of the right arm. Stimulations–acquisitions were 100 per muscle fiber, and stimulation rate was 10 Hz. We recorded 25 EDC muscle fibers per patient. Off-line analyses of recordings were performed, and on average 18 artifact-free EDC muscle fibers per patient were selected to assess their mean MCD. Results were expressed as the “mean value of consecutive differences” (MCD) of successive interpotential intervals, as usual in SFEMG studies. Fig. 1Figures representing a normally (on the top) and an abnormally (on the bottom) jittered fibers 3D-movement analysis Movements were recorded at 100 Hz in three dimensions (3D) using an infrared optoelectronic-tracking-system (ELITE™, Milan, Italy) with a reflective marker attached to the tip of the index finger and another marker to the movement target. Participants were seated with the target in the medio-sagittal plane on eye level. They were instructed to start with the right arm extended to the right, to touch the target with high precision (but fast) without any trunk movement, to go back to the starting position and to repeat above movements, in a given pace over 15 s (one trial) to result in 8–10 movements (Fig. 2). The used Cartesian coordinate system is head-fixed with a nasooccipital, a horizontal and a vertical axis and the origin in the target. Since in our previous study [10] abnormalities in migraineurs were most pronounced in the horizontal plane, we limited ourselves to an analysis of the mean deviation in the horizontal plane (in millimeters), measured over four trials, each consisting of 8–10 arm movements. Fig. 2Experimental setup for the movement task Quantitative variables in each group of migraine patients (MO and MTA) were expressed as medians. Differences between groups were analysed with the Mann–Whitney U test. The Spearman rank order correlation test was used to compare the values of the mean MCD on SFEMG and the mean horizontal deviation on 3D-movement analysis. Results On SFEMG, the median value for mean MCD was not significantly different between MO (16.05 μs; range 9.50–22.93) and MTA (18.91 μs; range 11.50–24.55). The median value of mean horizontal deviation was 4.01 mm (range −3.25 to 18.12) in MO and 10.74 mm (range 1.60–17.03) in MTA, a nonsignificant difference. In MO, there was no significant correlation between mean MCD and mean horizontal deviation. By contrast, in MTA, both variables were positively correlated (R = 0.71, P = 0.046) (Fig. 3). Fig. 3Scatter plot relating mean MCD on SFEMG (X axis, μs) and mean horizontal deviation on 3D analysis of a reaching arm movement (Y axis, mm), and linear regression lines. Migraine without aura patients (MO): squares, dashed line. Migraine with aura patients (MTA): triangles, continuous line  Discussion Our within-patient analysis shows that, in migraine with aura, the mean MCD on SFEMG increases significantly with the degree of horizontal deviation in a visually guided reaching movement. This suggests that, in subgroups of migraine patients, neuromuscular transmission (NMT) performance and control of ballistic movements by the lateral cerebellum are similarly influenced by a common biochemical and/or neural mechanism. Although no gene mutations have been identified till now, the genetic load is thought to be higher in the pathogenesis of migraine with aura than that without aura [13]. In familial hemiplegic migraine 1 (FHM1) [14], mutations have been found in the CACNA1A gene, which codifies for the main subunit of P/Q-type Ca2+ channels, heavily concentrated at the neuromuscular junction and in the cerebellum [15–18]. Thus, we initially hypothesized that the NMJ and cerebellar abnormalities found in some migraine patients might be due to dysfunctioning Ca++ channels. However, SFEMG studies were normal in FHM1 [19], and mutations in the CACNA1A gene are not found in patients suffering from the common forms of migraine, with or without aura [20]. Some evidence that the CACNA1A gene may be involved in migraine with typical aura [21], and the facts that NMT impairment can be found in episodic ataxia type 2, an allelic disorder of FHM1 [22], and normalizes after treatment with acetazolamide [23] are still in favor of a possible involvement of Ca++ channels, particularly in migraine with aura. Other proteins implicated in migraine pathophysiology may constitute a common link between neuromuscular junctions (NMJ) and the cerebellum. This is unlikely for the alpha-2 subunit of the Na-K ATPase, of which the gene ATP1A2 is mutated in FHM2 [24], because the alpha-2 isoform is found at intracellular membranes [25] and not at the NMJ, where the major role is played by the alpha-3 isoform [26]. Moreover, SFEMG results were found normal in FHM2 patients [19]. By contrast, a different group of calcium channels, the R-type, influence neuromuscular transmission and can compensate for dysfunctioning P/Q channels [27]. Moreover, they seem to play a role in cerebellar functions [28]; thus, their activity may influence in parallel NMJ and cerebellar performances, similarly to what we found in the subgroup of migraineurs with aura. The CACNA1E gene, which codes for the alpha-1E subunit of R-type channels, is precisely located on locus 1q31, for which significant linkage was found in the common forms of migraine [29], and a single nucleotide polymorphism was more frequent in subgroups of patients affected by migraine with aura [30]. In conclusion, the correlation between NMJ and cerebellar performances that we have found in migraine with aura patients might be due to a common genetically determined molecular mechanism, possibly influencing ion channels functions, but this has still to be proven by appropriate genetic studies.

J_Chem_Ecol-4-1-2266969

Prey and Non-prey Arthropods Sharing a Host Plant: Effects on Induced Volatile Emission and Predator Attraction

It is well established that plants infested with a single herbivore species can attract specific natural enemies through the emission of herbivore-induced volatiles. However, it is less clear what happens when plants are simultaneously attacked by more than one species. We analyzed volatile emissions of lima bean and cucumber plants upon multi-species herbivory by spider mites (Tetranychus urticae) and caterpillars (Spodoptera exigua) in comparison to single-species herbivory. Upon herbivory by single or multiple species, lima bean and cucumber plants emitted volatile blends that comprised mostly the same compounds. To detect additive, synergistic, or antagonistic effects, we compared the multi-species herbivory volatile blend with the sum of the volatile blends induced by each of the herbivore species feeding alone. In lima bean, the majority of compounds were more strongly induced by multi-species herbivory than expected based on the sum of volatile emissions by each of the herbivores separately, potentially caused by synergistic effects. In contrast, in cucumber, two compounds were suppressed by multi-species herbivory, suggesting the potential for antagonistic effects. We also studied the behavioral responses of the predatory mite Phytoseiulus persimilis, a specialized natural enemy of spider mites. Olfactometer experiments showed that P. persimilis preferred volatiles induced by multi-species herbivory to volatiles induced by S. exigua alone or by prey mites alone. We conclude that both lima bean and cucumber plants effectively attract predatory mites upon multi-species herbivory, but the underlying mechanisms appear different between these species. Introduction Plants can employ arthropod natural enemies and defend themselves against herbivorous insects and mites. Such arthropod natural enemies can have an impact on local herbivore populations and thereby also on plant fitness (Sabelis and van der Meer 1986; Dicke and Sabelis 1989; Pels and Sabelis 1999; van Loon et al. 2000; Fritzsche Hoballah and Turlings 2001; Kessler and Baldwin 2001). Plants can promote the effectiveness of natural enemies by providing alternative food, shelter, or volatiles (Price et al. 1980; Dicke and Sabelis 1988). The emission of herbivore-induced plant volatiles (HIPV) has been demonstrated in several families (e.g., Dicke et al. 1990a; Turlings et al. 1990; van den Boom et al. 2004; van Poecke and Dicke 2004). Many parasitoid wasps, predatory mites, and predatory insects use HIPV to locate their prey or hosts (reviewed in Dicke 1999; Dicke and Vet 1999; Sabelis et al. 1999). HIPV are thus thought to have an important influence on the interactions between plants, herbivores, and natural enemies in food webs (Turlings et al. 1995; Dicke and Vet 1999). To date, studies on the production of HIPV by plants and the responses of natural enemies to these have been carried out mostly for systems with one species of plant, herbivore, and natural enemy. Yet, in the field, most plants are likely to be attacked by several herbivore species, or by herbivores and pathogens, at the same time. It is difficult to predict whether or not changes in HIPV blends upon multi-species herbivory affect attraction of natural enemies, and in what direction. Several studies have attempted to fill this gap by analyzing the chemical composition of volatile blends upon multi-species infestation of plants (Shiojiri et al. 2001; Cardoza et al. 2002; Rodriguez-Saona et al. 2003; Rostás et al. 2006; Moayeri et al. 2007; Rasmann and Turlings 2007; Soler et al. 2007). For instance, Shiojiri et al. (2000) showed that the parasitoid wasp Cotesia plutellae prefers the HIPV blend induced by its host Plutella xylostella to the blend induced by multi-species herbivory by its host and the non-host caterpillar Pieris rapae although volatile blends are similar (Shiojiri et al. 2001). Cotesia glomerata, on the other hand, prefers HIPV blends induced by its host P. rapae and non-host caterpillars to blends induced by one caterpillar species feeding alone (Shiojiri et al. 2000; Vos et al. 2001). Cotesia marginiventris and Microplitis rufiventris do not discriminate between the multi-species-induced HIPV blend of plants infested with their host (beet armyworm caterpillars) and a plant pathogen or plants infested with their host alone, although the pathogen significantly suppresses volatile induction (Rostás et al. 2006). Similarly, C. glomerata does not discriminate between mustard plants infested with a root herbivore and its host, or plants infested with its host alone (Soler et al. 2007). The parasitoid C. marginiventris and the entomopathogenic nematode Heterorhabditis megidis exhibited a reduced attraction to volatiles from maize plants that were infested by a non-host in addition to their host, although the non-host was feeding on different tissues from the host, i.e., leaves vs. roots (Rasmann and Turlings 2007). In this study, we investigated the effects of multi-species herbivory by spider mites (Tetranychus urticae) and beet armyworm caterpillars (Spodoptera exigua) on volatile emissions by two different plant species: lima bean (Phaseolus lunatus) and cucumber (Cucumis sativus), and evaluated the responses of Phytoseiulus persimilis, a specialist natural enemy of spider mites. Both herbivore species are known to induce volatile emission in several plant species, and remarkable differences in HIPV blend composition have been found in lima bean infested by either of these herbivores (e.g., Ozawa et al. 2000; Schmelz et al. 2003; de Boer et al. 2004). We statistically compared volatile emission by multi-species-infested plants with the sum of volatile emissions of plants infested with either herbivore species alone to evaluate the presence of potentially antagonistic or synergistic effects among signal transduction pathways (e.g., Engelberth et al. 2001). Plant responses to multi-species herbivory may depend on feeding modes of the herbivore species involved (Walling 2000), but whether or not plant species differ in these responses is unknown. Therefore, we compared two plant species in their relative responses to multi-species vs. single-species herbivory by using the same two herbivore species to attack both plant species. We also studied the olfactory responses of predatory mites (P. persimilis). In nature, plants are commonly attacked by more than one herbivore. Knowledge on natural enemy responses to volatile blends induced by multi-species vs. single-species herbivory is essential to understand whether differences in volatile composition are relevant in an ecological context. Methods and Materials Plants, Herbivores, and Predators Lima bean plants (Phaseolus lunatus L. cv Sieva) and cucumber plants (C. sativus L. cv Lange Groene Giganten) were grown in a greenhouse at 20–30°C, 50–70% R.H. and 16 hr of light. Bean plants were used when their primary leaves had unfolded (and were about 8–12 cm wide), which was 10–15 d after sowing. Cucumber plants were used when their first two leaves were about 8–12 cm wide, which was 15–35 d after sowing.Two-spotted spider mites, Tetranychus urticae Koch (Acari: Tetranychidae), were reared on lima bean plants in a greenhouse under the same conditions as uninfested plants. Eggs of the beet armyworm Spodoptera exigua Hübner (Lepidoptera: Noctuidae) were obtained from a colony reared on artificial diet in the laboratory of Virology, Wageningen University, The Netherlands (Smits et al. 1986). Two to three batches of eggs were placed in a Petri dish together with a lima bean or cucumber leaf in a growth chamber at 23 ± 1°C. Larvae were used in experiments within 24 hr of hatching.A colony of predatory mites, Phytoseiulus persimilis Athias-Henriot (Acari: Phytoseiidae), was kept on spider mite-infested lima bean leaves in a growth chamber at 23 ± 1°C, 50–70% R.H. and continuous light. To obtain females of the same age for olfactometer experiments, gravid females were allowed to oviposit on a spider mite-infested bean leaf for 2 d. The female offspring were used in experiments 9 d after initiation of the colonies, i.e., 1–2 d after their final molt. Before the experiments, females were kept individually in 1.5-ml microcentrifuge tubes with a small droplet of water, for 24 ± 2 hr at 23 ± 1°C. Plant Treatments Treatments were applied to the two oldest leaves of lima bean or cucumber plants 3 d before an experiment. Plants were incubated in plastic cages in a climate-controlled chamber at 24 ± 1°C, 50–70% R.H. and 16L:8D, each treatment in a separate cage. We used the following four treatments: T. urticae: Adult female spider mites were transferred to experimental plants. Twenty mites per leaf were used on lima bean, and 100 mites per leaf were used on cucumber because the same number of spider mites leads to lower attraction of P. persimilis to cucumber than to lima bean plants (Dicke et al. 1990b).S. exigua: Newly hatched S. exigua larvae were placed in clip cages to keep caterpillars on the leaves (clip cages were made of two plastic cylinders, 2.5 cm diam., and a hairpin; the upper side was covered with gauze and the underside with Parafilm). Two larvae in one clip cage were used per leaf. Clip cages were moved daily to a new position on the same leaf, and any missing larvae were replaced. The clip cages were supported by sticks to prevent the leaves from bending and incurring damage due to the weight of the cages.Multi-species herbivory: Two newly hatched S. exigua larvae in a clip cage and 20 (lima bean) or 100 (cucumber) T. urticae females were placed on each leaf.Uninfested: Plants without herbivores were treated in the same way as plants of the other treatments. This treatment was only used for volatile collections of cucumber. We did not analyze volatile emission from uninfested lima bean plants because it is documented that they emit low amounts of volatiles (e.g., de Boer et al. 2004).When clip cages were used to keep caterpillars in place, T. urticae-infested or uninfested plants were treated in the same way with empty clip cages to avoid any cage effects. Volatile Collections and Analyses Just before volatile collections, plants were cut above the soil line. Stems of individual plants were wrapped in moist cotton and aluminum foil to prevent them from wilting during volatile collection, which lasted at most 4 hr. All clip cages, caterpillars, and their products were removed. Five plants of one treatment were transferred to a 5-l glass vessel. A viton O-ring and a metal clamp were used to attach the glass lid to the vessel to make it airtight. Purified air (filtered through silica, a molecular sieve, activated charcoal, and 90 mg Tenax) was split into two air streams of about 100 ml/min each and led into the vessels through teflon tubing. Volatiles from two treatments were collected simultaneously in parallel. The system was purged for 30 min before attaching a collection tube with Tenax (90 mg for bean samples, 200 mg for cucumber samples) to the air outlet in the lid of the vessel. Bean volatiles were collected for 25 min, cucumber volatiles were collected for 3 hr. For lima bean, volatile collection was repeated four times for each treatment, except for uninfested from which volatiles were not sampled. For cucumber, volatile collections were replicated seven times for the uninfested, six times for T. urticae, and five times for S. exigua and multi-species herbivory each. One sample of the T. urticae and one of the S. exigua treatments of cucumber were lost during the analysis.Volatiles from lima bean and cucumber were analyzed on two different gas chromatography-mass spectrometry (GC-MS) systems. Lima bean Volatiles were released from Tenax traps with a thermodesorption cold trap setup (Chrompack, Middelburg, The Netherlands) by heating at 250°C for 10 min, with a He-flow of 12 ml/min. Desorbed compounds were collected in the cold trap at −90°C. Volatiles were injected in splitless mode into the DB5 column (60 m × 0.25 mm ID, 0.25 μm film thickness) by heating the cold trap to 220°C. After an initial column temperature of 40°C for 4 min, the temperature was raised to 280°C at 4°C/min. The column was directly coupled to the ion source of a Finnigan MAT 95 mass spectrometer, which was operating in the 70-eV electron impact (EI) ionization mode with scanning from mass 24 to 300 at 0.5 scan/sec. Compounds were identified by comparison of mass spectra with those in the NIST 98 library and in the Wageningen Mass Spectral Database of Natural Products, and by checking the retention indices. Cucumber Volatiles were released from Tenax traps with a thermodesorption cold trap setup (Markes, UK) by heating at 200°C for 10 min, with a He-flow of 30 ml/min. The desorbed volatiles were collected in the cold trap at −10°C. Volatiles were injected in splitless mode into the RTX-5Silms column (Restec, 30 m × 0.32 mm ID, 0.33 μm film thickness) by heating of the coldtrap to 270°C. After an initial column temperature of 40°C for 2 min, the temperature was raised to 95°C at 3°C/min, then to 165°C at 2°C/min, and subsequently to 250°C at 15°C/min. The column was directly coupled to the ion source of a Finnigan quadrupole mass spectrometer, which was operated in the 70-eV EI ionization mode and scanned from mass 33 to 300 at three scans/sec. Compounds were identified by comparison of mass spectra with those in the NIST 98 and Wiley 7th edition spectral libraries, and by checking the retention indices. Analyses Not all compounds were detected in every replicate of every treatment. We therefore added one peak area unit to all measurements for lima bean and 10 units to all measurements for cucumber. These were the lowest recorded non-zero values for bean and cucumber, respectively. This enabled us to plot volatile emission on a log scale (Figs. 1a and 2a), and to determine ratios among treatments (see below). Fig. 1Effect of multi-species herbivory on volatile emission by lima bean plants. (a) Volatile emission (mean + SE peak area units) upon single herbivory by Tetranychus urticae (20 mites per leaf) (open bars), single herbivory by Spodoptera exigua (two larvae per leaf) (filled bars), or multi-species herbivory (hatched bars). Asterisks indicate compounds that were significantly affected by herbivore treatment (P < 0.05, Kruskal–Wallis tests). (b) Ratio of emission rates upon multi-species herbivory to the sum per volatile emitted by the two single herbivore treatments (see “Methods and Materials” for detailed explanation). Symbols indicate mean ratio, and error bars indicate upper and lower 95% confidence limits. Asterisks indicate compounds that were significantly more strongly induced by multi-species herbivory than by the sum of T. urticae and S. exigua (lower 95% confidence limit larger than 0). N = 4 for all treatments. Compound numbers: (1) (Z)-3-hexen-1-ol, (2) methyl salicylate, (3) (E)-β-ocimene, (4) (E)-4,8-dimethyl-1,3,7-nonatriene, (5) (3E,7E)-4,8,12-trimethyl-1,3,7,11-tridecatetraene, (6) p-mentha-1,3,8-triene, (7) (Z)-3-hexen-1-ol acetate, (8) 2-methylbutanal-O-methyl oxime, (9) 3-methylbutanal-O-methyl oxime, (10) linalool, (11) 1-octen-3-ol, (12) hexyl acetate, (13) limonene, (14) β-caryophyllene, (15) nonanal, (16) indole, (17) (E)-2-hexen-1-ol acetate, (18) 3-pentanone, (19) 3-octanone, (20) 2-methylbutanal nitrile, (21) (Z)-β-ocimene, (22) rose furan, (23) unknown 95B, 150, (24) 3-methylbutanal nitrile, (25) unknown 91B, 148, (26) unknown 41, 69B, 164, (27) 2-methylpropanal-O-methyl oximeFig. 2Effect of multi-species herbivory on volatile emission by cucumber plants. (a) Volatile emission (mean ± SE peak area units) upon single herbivory by Tetranychus urticae (100 mites per leaf, N = 5, open bars), single herbivory by Spodoptera exigua (2 larvae per leaf, N = 4, filled bars), or multi-species herbivory (N = 4, hatched bars). Asterisks indicate compounds that are significantly affected by herbivore treatment (P < 0.05, Kruskal–Wallis tests). (b) Ratio of emission rates upon multi-species herbivory to the sum per volatile emitted by the two single herbivore treatments (see “Methods and Materials” and for detailed explanation). Symbols indicate mean ratio, and error bars indicate upper and lower 95% confidence limits. Asterisks indicate compounds that were significantly less induced by multi-species herbivory than by the sum of T. urticae and S. exigua (upper 95% confidence limit smaller than 0) (N = 4) Compound numbers: (1) (E)-2-hexenal + (Z)-3-hexanal, (2) (Z)-3-hexen-1-ol acetate, (3) indole, (4) (E,E)-α-farnesene, (5) butyl aldoxime, (6) 3-methylbutanal-O-methyl oxime, (7) (E)-4,8-dimethyl-1,3,7-nonatriene, (8) (3E,7E)-4,8,12-trimethyl-1,3,7,11-tridecatetraene, (9) (E)-β-ocimene. Note that peak area units in Fig. 2a cannot be compared with Fig. 1a because measurements were done on a different GC-MS systemFor cucumber, we first determined which volatile compounds were induced by herbivory by using Kruskal–Wallis tests per compound. Variation in volatile emission levels of uninfested plants was low, indicating that other types of inducing factors (e.g., general stress) were absent or had only minor effects. Only herbivore-induced volatiles were used in subsequent analyses. We assumed that all compounds emitted by infested lima bean plants were induced by herbivores based on previous measurements of volatile emissions by uninfested lima bean plants (e.g., de Boer et al. 2004). Per plant species, we then used Kruskal–Wallis tests to analyze which compounds were differentially induced by the three herbivore treatments (T. urticae, S. exigua, or multi-species herbivory) (α = 0.05). Note that we did not statistically compare absolute levels of volatile emission across plant species because volatiles were collected for different periods of time, and analyses were run on two different GC-MS systems.We also tested whether the effect on HIPV of multi-species herbivory was the same as the combined effect of the two herbivores feeding alone, i.e., T. urticae + S. exigua. Per replicate, the amount of each volatile compound of the multi-species herbivory treatment was divided by the sum of the amounts emitted by T. urticae and S. exigua. The ratios were log-transformed before taking their averages. An average ratio of 0 means the effect of multi-species herbivory and the sum of T. urticae and S. exigua is equal. We concluded that a compound was significantly more strongly induced by multi-species herbivory than by the sum of T. urticae and S. exigua if the lower 95% confidence limit was larger than 0. Similarly, we concluded a significantly weaker effect of multi-species herbivory if the upper 95% confidence limit was smaller than 0 (see Figs. 1 and 2). We also used a χ2 test to test whether the number of compounds with a ratio >0 and a ratio <0 differed from a 50:50 distribution. A 50:50 distribution would be expected if the effect of multi-species herbivory and the sum of T. urticae and S. exigua is equal. One replicate of the T. urticae treatment in cucumber was excluded from this analysis because we had no corresponding replicates of the S. exigua and multi-species herbivory treatments. Note that these tests reflect volatile responses to multi-species herbivory relative to responses to single-species herbivory. This allows us to compare these data between lima bean and cucumber because the data no longer depend on the absolute levels of volatiles detected by the different GC-MS systems. Y-tube Olfactometer Experiments A closed system Y-tube olfactometer setup was used to test the responses of predatory mites to volatiles induced by multi-species herbivory (Takabayashi and Dicke 1992; de Boer et al. 2004). Odor sources consisted of four leaves of the same treatment that were cut from the plant just before an experiment; the petioles were wrapped in wet cotton wool and aluminum foil. Individual female predators were observed for a maximum of 5 min. A choice was recorded when the finish line, halfway one of the olfactometer arms, was reached within this period. Otherwise, it was recorded as “no-choice”. Sixteen to 23 predators were tested per experimental day, and each experiment was repeated on 4 different days. Per replicate, new groups of predatory mites and new sets of leaves were used. We tested the responses of predatory mites to volatiles from bean or cucumber plants of the multi-species herbivory treatment vs. volatiles from plants of the T. urticae or S. exigua treatments.The choices of predatory mites between two odor sources in the Y-tube olfactometer were analyzed with two-sided binomial tests to investigate whether the distribution of the predators differed from 50:50. Predators that did not make a choice were excluded from statistical analyses. Results Volatile Emission by Lima Bean Plants The main compounds emitted by T. urticae-infested bean plants were methyl salicylate (MeSA; compound number 2 in Fig. 1a), (3E)-4,8-dimethyl-1,3,7-nonatriene [(E)-DMNT] (4), and (3E,7E)-4,8,12-trimethyl-1,3,7,11-tridecatetraene [(E,E)-TMTT] (5). Herbivory by S. exigua alone resulted in the emission of similar amounts of (E)-DMNT (4), but amounts of MeSA (2) and (E,E)-TMTT (5) were small. Instead, S. exigua-infested bean leaves emitted large amounts of (E)-β-ocimene (3) and (Z)-3-hexen-1-ol acetate (7). Multi-species herbivory on bean plants induced the same volatile compounds as S. exigua feeding alone. Multi-species-infested plants emitted larger amounts of all volatile compounds than plants infested with a single herbivore species, except for (Z)-3-hexen-1-ol (1) that was emitted in larger amounts by S. exigua-infested plants, and MeSA (2) that was emitted in somewhat larger absolute amounts by T. urticae-infested plants.In lima bean, herbivore treatment had a significant effect on the emission of (Z)-3-hexen-1-ol (1), (E)-β-ocimene (3), (Z)-3-hexen-1-ol acetate (7), 2-methylbutanal-O-methyl oxime (8), linalool (10), 1-octen-3-ol (11), hexyl acetate (12), β-caryophyllene (14), indole (16), (E)-2-hexen-1-ol acetate (17), (Z)-β-ocimene (21), and two unknown compounds (23 and 26) (all P < 0.05, Kruskal–Wallis tests, Fig. 1a). These compounds were all induced in larger amounts by S. exigua and/or multi-species herbivory than by T. urticae. Total volatile emission was also significantly affected by treatment (P < 0.006, Kruskal–Wallis test, Fig. 1a).We compared the ratio of volatiles emitted in the multi-species herbivory treatment to the sum of volatiles emitted by the T. urticae and S. exigua treatments to determine whether infestation by two herbivore species simultaneously had additive, synergistic or antagonistic effects. Figure 1b shows the average ratio of multi-species herbivory to the sum of T. urticae and S. exigua for lima bean. Eight of 27 compounds were significantly more strongly induced by multi-species herbivory than by the sum of the two herbivore species feeding alone, suggesting synergistic effects (Fig. 1b). These compounds included members of various groups of chemicals, for example, (E)-2-hexen-1-ol acetate (17) and (Z)-β-ocimene (21). In addition, more compounds had a ratio >0 than expected (χ2 = 9.80, P = 0.002). None of the compounds was significantly less induced by multi-species herbivory, suggesting the absence of antagonistic effects of T. urticae and S. exigua feeding simultaneously on lima bean. Volatile Emission by Cucumber Plants Nine compounds were induced by herbivory on cucumber (all P < 0.10, Kruskal–Wallis tests; data not shown). The main compounds emitted by T. urticae-infested cucumber plants were (Z)-3-hexen-1-ol acetate (compound number 2 in Fig. 2a), (E,E)-α-farnesene (4), 3-methylbutanal-O-methyl oxime (6), (E)-DMNT (7), and (E)-β-ocimene (9) (Fig. 2a). Similar compounds were emitted in the largest amounts by S. exigua-infested cucumber plants. Multi-species herbivory resulted in a volatile blend that consisted of the same compounds as the blends induced upon herbivory by each of the herbivore species alone. The multi-species herbivory treatment resulted in the strongest induction of 3-methylbutanal-O-methyloxime (6) and (E)-β-ocimene (9), while T. urticae-infested plants emitted the largest amounts of (Z)-3-hexen-1-ol acetate (2). Herbivory by T. urticae alone and multi-species herbivory induced larger amounts of (E,E)-TMTT (8) than single herbivory by S. exigua alone. No MeSA was detected in any of the cucumber samples.In cucumber, herbivore treatment had an effect on the emission of 3-methylbutanal-O-methyloxime (6) and (E)-DMNT (9) (P < 0.05, Kruskal–Wallis tests, Fig. 2a). Both compounds were induced in larger amounts by T. urticae or multi-species herbivory than by S. exigua. Total volatile emission was not affected by herbivore treatment in cucumber (P > 0.10, Kruskal–Wallis test, Fig. 2a).Two volatile compounds [(E)-2-hexanal/(Z)-3-hexenal (1) and (Z)-3-hexen-1-ol acetate (2)] were induced significantly less by multi-species herbivory than by the sum of T. urticae and S. exigua (Fig. 2b), suggesting the presence of antagonistic effects of the two herbivore species feeding on cucumber. The seven other compounds were equally induced by multispecies herbivory and the sum of the two herbivores feeding alone, indicating an additive effect of T. urticae and S. exigua. The distribution of compounds with a ratio >0 or <0 did not differ from 50:50 (χ2 = 0.06, P = 0.81). Responses of P. persimilis We tested the preferences of the predatory mite P. persimilis for the volatile blends analyzed above in a Y-tube olfactometer. On bean and cucumber plants, predatory mites preferred HIPV induced by multi-species herbivory to blends induced by either T. urticae or S. exigua alone (Fig. 3, P ≤ 0.02, binomial tests). Fig. 3Responses of P. persimilis to volatile blends in the Y-tube olfactometer. Odor sources consisted of four leaves from lima bean (a and b) or cucumber plant (c and d) simultaneously infested by Tetranychus urticae and Spodoptera exigua or infested by one of the herbivore species alone (S. exigua: a and c, T. urticae: b and d). Bars present the overall percentages of predators choosing for each odor source. Numbers in bars are the total numbers of predators responding to each odor source. Choices between odor sources were analyzed with a two-sided binomial test (*P < 0.05; ***P < 0.001) Discussion We investigated volatile emissions by lima bean and cucumber plants upon single and multi-species herbivory by T. urticae and S. exigua. In lima bean, we found that most compounds were more strongly induced by multi-species herbivory than by the sum of each of the herbivores feeding separately (Fig. 1b). This suggests that the two herbivores feeding together on the same plant have more than an additive effect on volatile emission, potentially indicating that biosynthetic pathways may act synergistically. It is also possible that one or both herbivore species inflicted more damage on multi-species-infested plants than on plants with conspecifics only. However, it is unlikely that spider mite feeding was affected by the presence of S. exigua because in a preliminary experiment oviposition rates of mites, which are correlated with food intake rates, were similar on leaf discs with or without S. exigua damage (de Boer et al., unpublished). Some studies have observed that feeding by S. exigua is enhanced by the presence of aphids or a fungal pathogen (Rodriguez-Saona et al. 2005; Cardoza and Tumlinson 2006), but other studies have found no effect of the presence of whiteflies or a fungus (Rodriguez-Saona et al. 2003; Rostás et al. 2006). In contrast to bean, we found that in cucumber plants, two volatile compounds were induced less by multi-species herbivory than by the sum of the two herbivores feeding separately, suggesting that their induction was suppressed by simultaneous feeding of T. urticae and S. exigua. Whether this reduction is caused by a negative interaction between the signal transduction pathways induced by spider mites and beet armyworms, or to reduced feeding of one or both herbivore species as a result of feeding on the same plant, remains to be investigated. From our data, it appears that plant species is an important determinant in the way in which herbivore-induced defense pathways interact upon multispecies herbivory. The plant hormones salicylic acid (SA) and jasmonic acid (JA) play important roles in plant defense pathways and the production of HIPV (e.g., Kessler and Baldwin 2002; van Poecke and Dicke 2004). Interaction between these pathways may occur, for example, when SA inhibits JA-dependent defense pathways (e.g., Doares et al. 1995; Niki et al. 1998; Engelberth et al. 2001). To determine whether synergistic (lima bean) and antagonistic (cucumber) interactions between biosynthetic defense pathways caused the changes in HIPV blends that we observed, endogenous levels of JA and SA should be determined in future studies. Arimura et al. (2002) showed that both JA and SA levels are significantly elevated in lima bean after 3 d of spider mite feeding, whereas several Spodoptera species are known to increase JA levels in corn and broad bean (Blechert et al. 1995; Schmelz et al. 2003). We are not aware of analyses of SA levels in response to Spodoptera feeding. No studies of JA and SA levels in plants upon multi-species herbivory are known to us either. Another important question that needs to be addressed is the effect of herbivore sequence on volatile emission patterns, since in nature two herbivore species may infest an individual plant sequentially. In fact, some herbivore species are known to be attracted to HIPV themselves (e.g., Bolter et al. 1997; Kalberer et al. 2001), but spider mites are not attracted to S. exigua-infested lima bean plants (Horiuchi et al. 2003). To determine whether the differences in volatile profiles upon multi-species herbivory and single-species herbivory are relevant in an ecological context, we investigated the behavior of the predatory mite P. persimilis, a specialist natural enemy of spider mites. Our results show that P. persimilis prefers volatiles induced by multi-species herbivory by spider mites and the non-prey caterpillar S. exigua to volatiles induced by one of the herbivore species feeding alone (Fig. 3). The strong preference of P. persimilis for the volatiles induced by multi-species herbivory may be explained by the relatively large amount of some specific compounds. Of the three major compounds emitted by spider mite-infested lima bean, two, namely (E,E)-TMTT and especially (E)-DMNT, were induced more strongly by multi-species herbivory, whereas induction of the third, MeSA, was somewhat reduced (Fig. 1a). Of the five major compounds emitted by spider mite-infested cucumber, four were induced more strongly by multi-species herbivory: (E)-DMNT, (E)-β-ocimene, (E,E)-α-farnesene, and 3-methylbutanal-O-methyl oxime, while (Z)-3-hexen-1-ol acetate emission was reduced (Fig. 2a). (E)-DMNT, (E,E)-TMTT, and (E)-β-ocimene are known attractants for P. persimilis (Dicke et al. 1990a, de Boer et al. 2004), and (E,E)-TMTT can be used by P. persimilis to discriminate between complex blends of HIPV (de Boer et al. 2004). Similarly, when S. exigua-induced volatiles are compared to the blend induced by multi-species herbivory, all of these compounds are produced in larger amounts in the latter. This suggests that a greater emission of (the sum of) attractive compounds may be the main determinant in attracting the predatory mite P. persimilis. Previous studies on the responses of parasitoid wasps to HIPV also demonstrated a preference for HIPV induced by multi-species herbivory compared to single-species herbivory (Shiojiri et al. 2000; Vos et al. 2001; Cardoza et al. 2003; Moayeri et al. 2007). Rodriguez-Saona et al. (2005) even showed that naïve Cotesia marginiventris females were only attracted to tomato plants infested with their host S. exigua when the same plant was also infested with aphids. The presence of aphids thus made the host-infested plants more detectable. The opposite, i.e., multi-species herbivory-induced volatiles being less attractive than volatiles induced by only the host has been found in two cases: C. plutellae wasps preferred volatiles induced by their host, P. xylostella, to volatiles induced by their host and the non-host caterpillar P. rapae (Shiojiri et al. 2000), and C. marginiventris as well as the entomopathogenic nematode H. megidis were less attracted to volatiles from maize plants that were infested by a non-host in addition to their host (Rasmann and Turlings 2007). In such cases, the presence of non-host or non-prey herbivores may decrease the detectability or reliability of HIPV cues for natural enemies. For example, feeding by a non-prey or non-host herbivore may suppress the induction of volatile compounds used by natural enemies. Multi-species herbivory thus adds a new dimension to the trade-off between reliability and detectability of searching cues used by natural enemies of herbivorous arthropods (Vet and Dicke 1992).

Exp_Brain_Res-4-1-2248215

Attenuation of N2 amplitude of laser-evoked potentials by theta burst stimulation of primary somatosensory cortex

Theta burst stimulation (TBS) is a special repetitive transcranial magnetic stimulation (rTMS) paradigm, where bursts of low-intensity stimuli are applied in the theta frequency. The aim of this study was to investigate the effect of neuronavigated TBS over primary somatosensory cortex (SI) on laser-evoked potentials (LEPs) and acute pain perception induced with Tm : YAG laser stimulation. The amplitude changes of the N1, N2, and P2 components of LEPs and related subjective pain rating scores of 12 healthy subjects were analyzed prior to and following continuous TBS (cTBS), intermittent TBS (iTBS), intermediate TBS (imTBS), and sham stimulation. Our results demonstrate that all active TBS paradigms significantly diminished the amplitude of the N2 component, when the hand contralateral to the site of TBS was laser-stimulated. Sham stimulation condition had no significant effect. The subjective pain perception also decreased during the experimental sessions, but did not differ significantly from the sham stimulation condition. The main finding of our study is that TBS over SI diminished the amplitude of the N2 component evoked from the contralateral side without any significant analgesic effects. Furthermore, imTBS produced responses similar to those observed by other forms of TBS induced excitability changes in the SI. Introduction Functional neuroimaging studies have provided unequivocal evidence of the participation of the primary somatosensory cortex (SI), secondary somatosensory cortex (SII), and insula in pain processing (Talbot et al. 1991; Casey et al. 1994; Apkarian et al. 1999; Gelnar et al. 1999). Evidence suggests that the nociceptive input into these regions at least partially underlies the perception of sensory features of pain (Bushnell et al. 1999; Coghill et al. 1999; Peyron et al. 1999; Chen et al. 2002; for a reviews see: Peyron et al. 2000; Apkarian et al. 2005). Electrophysiological studies have also confirmed the participation of the SI in pain processing, contralateral to the stimulated side. Tarkka and Treede (1993) first reported pain induced-activity in the SI using electroencephalogram (EEG) and applying brain electric source analysis (BESA). In their model, the peak latency measured at the SI was very similar to that of SII. In a combined magnetoencephalogramm (MEG) and laser-evoked potential (LEP) study Ploner et al. (1999) reported SI activity also contralateral to the side of stimulation and further to this proposed a parallel pain processing in SI and SII. This parallel activation of SI and SII was confirmed by other studies (Ploner et al. 2000, 2002) and other groups (Timmermann et al. 2001; Kanda et al. 2000; Inui et al. 2003; Nakata et al. 2004). However, some previous LEP dipole modelling studies showed that a dipole source in SI area was necessary to explain the scalp LEP topography, none of them reported a clear correspondence between the SI activity and a definite LEP component (Tarkka and Treede 1993; Ploner et al. 2002; Kanda et al. 2003). Kanda et al. (2000) detected SI activity following painful stimulation by recording intracranial EEG, but recently, intracerebral depth recordings in an epileptic patient have shown no reliable LEP response from the area 3b of the SI after painful laser stimulation, although a reliable N2–P2 response could be recorded at Cz (Valeriani et al. 2004). Inui et al. (2003) reported the absence of activation from area 3b of the SI after noxious electrical stimulation as well, however they found activity in the area 1 of SI. Other studies found no pain-related activation of the SI at all (for a review see: Garcia-Larrea et al. 2003; Kakigi et al. 2005). Recent studies showed, that several kinds of external stimulation methods such as single-pulse transcranial magnetic stimulation (TMS) (Kujirai et al. 1993), 1 Hz repetitive TMS (rTMS) (Enomoto et al. 2001), paired associative stimulation (PAS) (Tsuji and Rothwell 2002; Wolters et al. 2005), transcranial direct current stimulation (tDCS) (Matsunaga et al. 2004; Dieckhofer et al. 2006) or theta burst stimulation (TBS) (Ishikawa et al. 2007) modulate the amplitude of cortical components of median nerve somatosensory evoked potentials (SEPs). It was recently reported that, cathodal tDCS over the SI (Dieckhofer et al. 2006) significantly reduced the N20 amplitude of median nerve SEPs. Furthermore, cathodal polarization over the SI induced a prolonged decrease of tactile discrimination (Rogalewski et al. 2004) and diminished acute pain perception and the amplitude of the N2 component of LEPs (Antal et al. 2007). Recently Huang et al. (2005) developed a special “theta burst” paradigm to modulate human motor cortex (M1) excitability using low intensity, repetitive bursts of magnetic stimuli. The authors distinguished three stimulation patterns, which were proved to have different effects over M1 activity, when it was monitored by the amplitude of transcranial motor evoked potentials (MEPs). Continuous TBS (cTBS) caused a significant reduction in MEP amplitudes, which was probably due to the inhibition of specific excitatory circuits (I1-wave inputs to corticospinal neurons), as later confirmed by another study (Di Lazzaro et al. 2005). In contrast, intermittent TBS (iTBS) facilitated M1 activity and produced increase in MEP amplitudes. Interestingly, intermediate TBS (imTBS) had no effect at all. In addition, TBS has also been shown to have an effect on the human premotor (Mochizuki et al. 2005) and visual cortex (Franca et al. 2006). Non-invasive cortical stimulation of M1 for the treatment of certain kinds of chronic and experimentally induced pain has recently attracted much interest. Both low and high frequency rTMS was reported to reduce subjective pain perception and has been used experimentally to reduce chronic pain (for reviews see: Leo and Latif 2007; Fregni et al. 2007). However, although the involvement of SI in pain perception and processing is reported by several imaging studies (for reviews see: Peyron et al. 2000; Apkarian et al. 2005), the magnetic stimulation of this cortical area in order to modify pain perception is neglected. Therefore, the aim of our study was to investigate the effects of cTBS, iTBS, and imTBS, on the early (N1) and late (N2, P2) components of LEPs and related subjective pain perception when applied over the left SI. We hypothesized that the three TBS types, which have short durations (maximum 190 s), would affect LEP components in a specific, paradigm-dependent manner similarly to the effect over M1 as revealed by Huang and et al. (2005). Methods Subjects Nineteen healthy volunteers between 18 and 35 years were informed about all aspects of the experiments and signed an informed consent. Three subjects chose not to continue the experiment after the first or second session and during off-line EEG analysis four further subjects were dropped out because of their hardly detectable LEP components. Hence, twelve of the subjects (six male, six female; mean age = 26.33 ± 3.17 years) were included in the final analysis. We conformed with the Declaration of Helsinki and the experimental protocol was approved by the Ethics Committee of the University of Göttingen. None of the subjects suffered from chronic pain syndromes, nor took any medication regularly. None had a history of neurological or psychiatric illness. All of them participated in all four sessions, the three TBS and sham stimulations. TBS stimulation Theta burst stimulation was applied over the hand area of the left SI using a standard, figure-of-eight-coil (MCF-B65 Butterfly Coil) and MagPro stimulator (Medtronic, Denmark) with an outer half-coil radius of 75 mm, with a posterior–anterior–posterior current flow in the coil. Stimulus intensity was 80% of active motor threshold (AMT) (Huang et al. 2005). For AMT determination, the coil was placed tangentially to the scalp, with the handle pointing backwards and laterally 45° from mid-line. MEPs of the right abductor digiti minimi muscle (ADM) were recorded by Ag-AgCl-electrodes in a belly tendon-montage before each stimulation. The signals were amplified and filtered (1.59 Hz-1 kHz, sampling rate of 5 kHz), digitalized with a micro 1401 AD converter (Cambridge Electronic Design, Cambridge, UK), recorded by a computer using SIGNAL software (Cambridge Electronic Design, Version 2.13). Complete muscle relaxation was controlled though auditory and visual feedback of EMG activity. AMT was defined as the minimum intensity eliciting a MEP of a superior size when compared to spontaneous moderate muscular activity in at least three of six pulses. The pattern of TBS consisted of bursts containing three pulses at 50 Hz which were repeated at 200 ms intervals (i.e., 5 Hz) for up to 600 pulses for 40 s continuously (cTBS), or triads repeated at 200 ms intervals for 2 s intermittently with 8 s breaks for up to 600 pulses (iTBS). In the case of imTBS the triads (three pulses at 50 Hz) were repeated at 200 ms intervals for 5 s, in every 15 s up to 600 pulses (Huang et al. 2005). In separate experimental sessions, sham stimulation was applied with the cTBS protocol using the same coil held over the same position but tilted to a 90° angle (one-wing 90°) (Lisanby et al. 2001) with only the margin of the coil in contact with the scalp. Determination of the primary somatosensory cortex (SI) Anatomical magnetic resonance imaging (MRI) (Siemens 3 T, T1 weighted) dataset was used to determinate the coil localization for the stimulation of SI for all subjects using the Brainsight neuronavigation system (Rogue Research Inc., Montreal Quebec, Canada). The hand area was determined by previous fMRI studies (Bushnell et al. 1999; Blankenburg et al. 2003) and was located and marked in the MRI dataset as the target for TMS application (Fig. 1). Fig. 1Three dimensional anatomical MRI of a single subject. Anatomical magnetic resonance imaging (MRI) (Siemens 3 T, T1 weighted) dataset was used to determinate the coil localization for the stimulation of SI by all subjects using the Brainsight neuronavigation system. The hand area was located and marked in the MRI dataset as target for TMS application. The black point indicates the hot-spot of the stimulation coil Laser stimulation A Tm : YAG laser system (WaveLight Laser Technologie AG, Erlangen, Germany) was used to induce painful stimulation. The thulium laser emits near-infrared radiation (wavelength 2,000 nm, pulse duration 1 ms, laser beam diameter 7 mm) with a penetration depth of 360 μm into the human skin and allows a precise restriction of the emitted heat energy to the termination area of primary nociceptive afferents without affecting the subcutaneous tissue (Treede et al. 2003). The distal handpiece of the laser was positioned 30 cm from the radial part of the dorsal surface of the hand. The pain threshold was determined on both hands at the beginning each session before baseline EEG recording by applying laser stimuli from 200 mJ in 50 mJ steps. During EEG recordings, each laser stimulus was delivered with an intensity of 1.4–1.6 times the threshold intensity to a slightly different spot in a 5 × 5 cm square on the dorsum of the hand in order to reduce receptor fatigue or sensitization by skin overheating (Treede et al. 2003). Skin temperature of the stimulated area was checked prior to every switch between hands, and corrected with a heating lamp if it fell below 35°C. Psychophysical evaluation We used the verbal analogue score (VAS) to assess the subjective intensity of pain. The subjects were instructed to pay attention to the laser stimuli and to rate the perceived pain verbally (1-warm, 1.1 smallest pain, and 1.9 most intense pain) about 2–3 s after each laser-impulse. The values were individually averaged separately for both hands in each session and conditions. The ears of the subjects were plugged during the measurements to avoid auditory artifacts accompanying laser stimulation. Electrophysiological recordings The EEG was recorded using a 64-channel montage applying 64 ring electrodes (inner diameter: 6 mm, outer diameter: 12 mm) (EasyCap; Falk and Minow GmbH, Münich, Germany). The electrodes were placed in accordance with the extended international 10–20 system. The impedance was kept <5 kΩ. The Fz was used as reference, the ground was placed 2 cm anterior to the tragus of the right ear. Data were collected at a sampling rate of 1,000 Hz with the BrainAmp system (Brain Products GmbH, Munich, Germany) and were analyzed offline. The obtained data were re-referenced to the connected mastoids (TP9-TP10). A 0.5 Hz low-cutoff as well as a 30 Hz high-cutoff filter was used. In addition to automatic artifact detection (200 μV amplitude criterion) all epochs were visually inspected, and those containing eye blinks or muscle movement artifacts were excluded. Baseline correction was performed on the basis of the 100 ms prestimulus interval. The amplitudes of N1 (referring to Fz) and N2–P2 (referring to TP9-TP10) components were measured. Although we recorded data on 64 channels, we assessed LEPs according to the scalp distribution on the analyzed peaks. The N2 is a negative component (referring to TP9-TP10) was peaking around 160–240 ms. The amplitude of the early N1 negative peak which came before N2 on T7 and T8 channels (referring to Fz) was analyzed. The P2 positive component after N2 was peaking around 300–360 ms. The N2 component is largest over the lateral temporal and fronto-central areas on electrodes Fz, Cz, CPz, F1–F4, FC1–FC6, C1–C6, T7, T8, and CP1–CP6. In contrast, the P2 peak has its maximum amplitude over the vertex on electrodes FCz, Cz, CPz, Pz, F1–F4, FC1–FC4, C1–C4, CP1–CP4, and P1–P4. For the analysis of LEPs according to regional distribution, we defined three distinct areas with pooling the data: central (with all the mid-line electrodes such as Fz, FCz, Cz, CPz, and Pz), left (F3, FC3, C3, CP3, P3, FC5, C5, CP5, and T7) and right (electrodes according to the left side) instead of separate electrodes. Experimental design The subjects were sitting in a reclining chair. In case of all sessions first the EEG cap was placed on the head. After pain threshold determination the baseline LEP measurements were performed. Every run for the LEP recording consisted of 40 epochs of laser stimulation on each hand. The interstimulus interval of the stimulation ranged from 8 to 15 s (Raij et al. 2003). Thus the LEP recording lasted for 8–10 min for the first hand and also for 8–10 min for the second hand laser-stimulation. In all three TBS and sham TBS conditions, the right hand was stimulated first in half of the cases and the left hand was stimulated first in the other half. This order was kept for the subjects for all conditions. After baseline LEP recording the AMT was measured (∼15–20 min) and the TBS were applied for 40–190 s through the cap. After TBS the impedance of the EEG electrodes were retested and corrected below 5 kΩ if it was necessary. The TBS was followed by a post-stimulation LEP recording in ∼5 min after TBS. Thus, the interval between the two LEP recordings with regard to the same hand was about 30–35 min including all subjects and conditions. The experimental sessions were separated from each other by at least 5 days. The subjects were blinded as to the type of magnetic stimulation. The order of the sessions was randomized across subjects. Data analysis The individually averaged VAS values and LEP amplitudes were entered into a repeated-measures-ANOVA for both hands and LEP components separately [four TBS CONDITION (cTBS, iTBS, imTBS, and sham) × 2 TIME (before, after TBS)]. We considered a psychophysical or an electrophysiological change only if the CONDITION × TIME interaction was significant. In case of the LEP components we investigated if this effect was dependent on the defined areas by calculating the CONDITION × TIME × REGION interaction. Post hoc analysis was done using Student’s t tests (paired samples, two-tailed, level of significance P < 0.05). Results Psychophysics The intensity of the laser stimulation (1.4–1.6 × of the pain threshold) was 19.88 mJ/mm2 for cTBS, 20.53 mJ/mm2 for iTBS, 20.52 mJ/mm2 for imTBS, and 20.33 mJ/mm2 for sham stimulation. None of the subjects reported any side-effect after the stimulation. In case of the contralateral hand (right) stimulation, repeated-measures-ANOVA revealed no main effect of CONDITION [F(3,33) = 0.828, P = 0.488] but the TIME was significant [F(1,11) = 27.270, P < 0.001]. The CONDITION × TIME interaction was also not significant [F(3,33) = 0.080, P = 0.97]. In case of the ipsilateral hand (left) stimulation, there was no significant main effect of CONDITION [F(3,33) = 1.329, P = 0.282] but the TIME was significant [F(1,11) = 15.395, P < 0.005]. The CONDITION × TIME interaction was also not significant [F(3,33) = 0.716, P = 0.55] (Fig. 2). Fig. 2The effects of the TBS on subjective pain perception. The VAS-values were standardized for each subject, for each condition by calculating the after/before ratio. The independent variables were the CONDITION and HAND in order to determine if there is any difference between stimulation of the two hands. There was no main effect of the stimulation CONDITION and the HAND (left or right). The interaction CONDITION × TIME was also not significant Electrophysiology The N1, N2, and P2 components could clearly be identified in all subjects. The LEPs are presented in Figs. 3 and 4. Fig. 3The effects of the TBS on the N1 component. The figure shows N1 amplitude changes during the experimental sessions on the grand averages of LEPs. The N1 was analyzed over the electrodes T7 and T8 referred to Fz (international 10–20 electrode system) in case of both hands. The solid line shows LEPs before and the intermittent line after TBS interventions. There was no significant difference between stimulation conditionsFig. 4Grand averages of LEPs obtained by contralateral (right) hand laser stimulation for three scalp regions (central, left and right) re-referenced to the connected mastoids (TP9-TP10), before and after the three TBS conditions and sham stimulation. The solid line shows LEPs before and the intermittent line after TBS interventions The N1 component We analyzed the amplitudes of the early N1 components on channels T7 and T8 (referring to Fz). There was no significant main effect of CONDITION [contralateral hand: F(3,33) = 2.216, P = 0.105; ipsilateral hand: F(3,33) = 2.865, P = 0.052]. The TIME was significant if the contralateral hand was stimulated [F(1,11) = 6.186, P = 0.030] but not for the ipsilateral hand [F(1,11) = 3.305, P = 0.096]. The interaction of the CONDITION × TIME resulted in no significant interaction neither after the contraleteral [F(3,33) = 0.727, P = 0.543] or ipsilateral [F(3,33) = 1.694, P = 0.187] hand stimulation (Fig. 3). The N2 component The repeated-measures-ANOVA (ipsilateral hand) showed no significant main effect of CONDITION [F(3,33) = 0.555, P = 0.65], but the effect of TIME was significant [F(1,11) = 11.769, P = 0.006]. The CONDITION × TIME interaction resulted in no significance [F(3,33) = 0.149, P = 0.93] (Fig. 5a). In case of the (contralateral hand) the main effect of CONDITION [F(3,33) = 0.250, P = 0.86] was not significant, but the TIME was significant [F(1,11) = 32.034, P < 0.001]. The CONDITION × TIME interaction was also significant [F(3,33) = 4.058, P = 0.015] (Fig. 5b). The interaction with electrode position was not significant [F(6,66) = 1.068, P = 0.39]. The post hoc t test showed that all active TBS stimulation significantly decreased the amplitudes of the N2 component at all defined regions for the contralateral hand stimulation. Table 1 summarizes the results of t tests. Fig. 5The mean N2 amplitude changes during the four TBS conditions at the ipsilateral (left) (Fig. 5a) and contralateral (right) (Fig. 5b) hand laser-stimulation for the three calculated regions (central, left and right). The stars mark significant differences between before-after TBS conditions (post hoc t-tests, paired samples, two-tailed, P < 0.05)Table 1Post hoc analysis of the N2 componentcTBSiTBSimTBSshamBefore versus afterBefore versus afterBefore versus afterBefore versus afterCentralP-levels0.00330.00020.00210.3942t-values−3.730−5.499−4.005−0.887LeftP-levels0.00540.00040.00310.5921t-values−3.451−5.041−3.775−0.552RightP-levels0.03940.00850.00390.8118t-values−2.337−3.196−3.6450.244Shows the results of the Student’s t-tests (paired samples, two-tailed) in case of the right hand stimulation. The level of the significance was P < 0.05 The P2 component The repeated-measures-ANOVA showed no significant main effect of CONDITION [contralateral: F(3,33) = 1.571, P = 0.22; ipsilateral: F(3,33) = 1.054, P = 0.38], but the TIME was significant [contralateral: F(1,11) = 17.038, P = 0.002; ipsilateral: F(1,11) = 15.362, P = 0.002]. The CONDITION × TIME interaction was also not significant [contralateral: F(3,33) = 2.669, P = 0.064; ipsilateral: F(3,33) = 0.418, P = 0.74]. The means of the different LEP components from all 12 subjects are presented in Table 2. Table 2The mean amplitudes of the LEP componentsPeakcTBSiTBSimTBSshamBeforeAfterBeforeAfterBeforeAfterBeforeAfterLeft handN1T7−4.42 ± 2.75−4.60 ± 3.13−4.68 ± 2.52−3.71 ± 2.65−3.33 ± 3.22−3.50 ± 3.15−4.98 ± 2.70−5.20 ± 3.08T8−6.13 ± 2.17−4.78 ± 3.60−6.73 ± 3.44−5.13 ± 2.89 −4.99 ± 4.27−4.64 ± 3.62−6.56 ± 3.47−6.33 ± 3.11N2Central−9.77 ± 5.48−7.92 ± 4.91−10.29 ± 5.01−8.88 ± 3.98−9.04 ± 5.43−7.25 ± 5.52−10.02 ± 4.43−8.11 ± 6.16Left−7.13 ± 3.60−6.05 ± 3.44−7.44 ± 3.49−6.69 ± 2.91−6.81 ± 3.69−5.55 ± 3.88−7.13 ± 3.55−5.98 ± 4.28Right−6.70 ± 3.05−4.59 ± 3.25−6.42 ± 3.54−5.41 ± 2.56−5.88 ± 3.05−4.70 ± 3.19−6.11 ± 3.24−4.82 ± 3.85P2Central14.79 ± 8.0211.74 ± 7.0314.86 ± 7.6912.43 ± 7.0013.19 ± 7.7411.26 ± 8.2313.83 ± 6.7512.03 ± 6.31Left8.40 ± 3.897.18 ± 3.669.19 ± 4.74 7.74 ± 3.597.36 ± 4.806.63 ± 5.167.90 ± 3.646.84 ± 3.67Right8.80 ± 3.806.78 ± 3.038.68 ± 4.657.59 ± 3.687.91 ± 3.706.90 ± 3.988.28 ± 4.277.76 ± 4.17Right handN1T7−8.18 ± 4.79−5.98 ± 3.39−6.38 ± 4.67−5.12 ± 3.62−7.34 ± 4.70−5.88 ± 3.73−7.43 ± 3.30−6.38 ± 3.21T8−4.92 ± 4.43−3.96 ± 3.24−3.25 ± 3.29−2.98 ± 2.35−4.56 ± 4.61−3.03 ± 2.39−4.37 ± 3.53−4.15 ± 2.62N2Central−10.11 ± 3.83−7.44 ± 4.52−11.05 ± 4.87−6.69 ± 4.23−9.99 ± 5.06−6.40 ± 4.27−9.57 3.97−8.94 ± 5.40Left−7.74 ± 2.70−5.67 ± 2.77−8.62 ± 3.65−5.71 ± 3.43−7.77 ± 3.80−5.09 ± 2.64−6.79 ± 3.29−6.42 ± 3.78Right−6.58 ± 3.37− 5.12 ± 3.34−6.76 ± 3.00−4.80 ± 3.26−6.52 ± 3.59−4.24 ± 2.72−5.82 ± 2.74−6.00 ± 3.99 P2Central14.30 ± 6.5611.62 ± 5.9314.52 ± 7.1410.90 ± 5.4914.04 ± 7.7910.74 ± 7.1311.37 ± 6.3810.75 ± 5.73Left8.47 ± 3.876.95 ± 3.038.90 ± 4.147.00 ± 2.918.65 ± 4.796.64 ± 3.376.21 ± 3.456.62 ± 3.22Right8.52 ± 3.836.73 ± 2.758.29 ± 4.176.41 ± 3.328.75 ± 4.296.90 ± 3.766.84 ± 4.446.67 ± 4.04The mean amplitudes of the LEP components before and after stimulation in all four conditions. (mean ± standard deviation) Discussion The main finding of our study is that all theta burst paradigms over the SI were able to diminish the amplitude of the N2 component of LEPs significantly when compared to sham stimulation. Surprisingly, the imTBS condition, that is suggested to be used as a placebo condition, when it applied over the M1 (Huang et al. 2005), also caused a strong amplitude decrease. The N1 and P2 components and the subjective pain rating scores were not significantly influenced by any type of TBS. Recent studies using the theta burst paradigm have concentrated on the effects of continuous (cTBS) and intermittent theta burst stimulation (iTBS) (Franca et al. 2006; Andoh et al. 2007; Ishikawa et al. 2007; Koch et al. 2007; Mochizuki et al. 2007). None of them investigated the intermediate (imTBS) pattern since Huang and et al. published that it has no effect over the M1 as revealed by MEPs and could thus be used as a sham condition (Huang et al. 2005). According to our knowledge, only five studies applied TBS over non-motor cortical areas. First Franca et al. (2006) used the theta burst pattern of rTMS over the visual cortex. They found that cTBS increased phosphene thresholds whilst iTBS was found to be ineffective. In another study, both cTBS and iTBS over the left dorsal premotor cortex decreased the transcallosal inhibition revealed by pairs of transcranial magnetic stimuli (Koch et al. 2007). In a recent study, Wernicke’s area was stimulated with iTBS while the reaction time of auditory word detection was measured (Andoh et al. 2007). In this work, iTBS facilitated the detection of foreign words when compared with native words. Concerning the SI, a shorter form of the iTBS (300 pulses) over the left SI resulted in a significant oxy-hemoglobin decrease at the contralateral SI and M1, detected by near infrared spectroscopy (Mochizuki et al. 2007). In another study cTBS of the SI resulted in a temporary decrease (13 min), whereas stimulation of the M1 caused a long-lasting increase (up to 53 min) of the amplitudes of cortical components of the median nerve SEPs (Ishikawa et al. 2007). In summary, these results suggest that cTBS has an inhibitory effect on non-motor areas; whereas the effect of iTBS is more facilitatory, but clearer results have still to emerge. Our results are the first demonstrating that all three TBS paradigms, but not by sham stimulation over the SI resulted in similar after-effects regarding the amplitude of the N2 component of LEPs evoked by the laser stimulation of the contralateral hand. The N2 component (peaking around 160–220 ms), is generated bilaterally in the operculoinsular region and in the anterior cingulate cortex (ACC) (Garcia-Larrea et al. 2003) and reflects sensory, discriminatory processes (Garcia-Larrea et al. 1997; Iannetti et al. 2005); whilst the P2 component (peaking around 300–360 ms) arose mostly from the ACC and represents attentional, cognitive and affective factors of pain perception and processing (Treede 2003). However, other studies did not find significantly different brain sources for N2 and P2, revealing both parasylvian and ACC contributions for the N2–P2 components (Raij et al. 2003; Ohara et al. 2004). Thus, LEP changes in N2 or P2 component might result from changes in either sensory-discriminative or cognitive aspects of pain. Studies using subdural recordings (Kanda et al. 2000; Ohara et al. 2004) or MEG (Kanda et al. 2000) demonstrated that the LEP components can be recorded over the SI and SII simultaneously, and the N2 peak may indicate the arrival of input originating from nociceptors. With regard to the N1 component which is an early LEP potential (peaking around 140–170 ms) and reflecting the early sensory-discriminative processing of pain perception (Iannetti et al. 2005), we did not find any significant change in amplitude. According to scalp topography (maximum near T3 and T4), the N1 is generated near to the SII in the fronto-parietal operculum (Treede et al. 2003). The participation of operculoinsular cortex in coding the pain intensity was recently suggested by using LEP measurements (Ianetti et al. 2005). However, another study (Gracia-Larrea et al. 1997) did not find any significant correlation between the amplitude of the N1 component and subjective pain rating. The fact that we found a decrement of both the N2 and P2 amplitudes in the sham condition as well, should be discussed. This phenomenon is known as habituation and has been described in previous LEP studies (Spiegel et al. 2000; Tamura et al. 2004a). Still, the effects of real TBS conditions on N2 amplitudes were greater than that of the sham condition and it was significant above the stimulated left, the neighboring central and the contralateral area as well when the right hand was laser stimulated. Attention can also directly influence the N2–P2 components of LEPs as well as subjective pain rating as it was suggested by previous experimental results (Gracia-Larrea et al. 1997; Ohara et al. 2006). In our experiment the subjects were asked to pay attention to each laser stimuli and since the TBS and sham condition were applied in a randomized order and the subjects were blinded as to the type of magnetic stimulation, the significant difference between sham and the other three TBS on the N2 amplitude is more than simply habituation or the effect of the different attentional states. The possible origin of the N2 component is mainly the bilateral operculoinsular region and the ACC (Garcia-Larrea et al. 2003). Therefore, when we inhibited the left SI, the activity of the pain related cortical network decreased due to the widespread cortical connections between SI and other cortical areas. However, the involvement of the contralateral SI and bilateral SII—parasylvian region in the N2 component generation was also reported (Kakigi et al. 2005; Kanda et al. 2000; Ohara et al. 2004). According to this, it is also possible that the inhibition of the SI itself may cause direct effects on LEP components. In this study we found no significant difference between the stimulation conditions with regard to subjective pain perception (Fig. 2), however, the subjective pain rating decreased during the experimental sessions after every type of TBS including the sham condition. It is important to mention that the placebo effect is high in almost every pain perception study, regardless of the paradigm used (Khedr et al. 2005; Lefaucheur et al. 2004). The explanation of this discrepancy between the electrophysiological and psychophysical changes is rather speculative, because the relationship between the anatomical origin of the N2–P2 components and their psychophysical correlates controversially discussed in the literature. In a PET study different cortical activations for pain threshold, intensity and unpleasantness have been found (Tölle et al. 1999). One possibility is that modulation the excitability of the left SI by TBS activated some elements of the pain related network that caused a decrease in the N2 amplitude. However, this stimulation intensity or duration was not strong or adequate enough to modify the subjective pain rating. In our previous study (Antal et al. 2007), we found that cathodal tDCS over the SI, similarly to the present findings, significantly decreased the N2 component of LEPs. In contrast to the present findings the subjective pain perception in healthy subjects was also diminished. TDCS is a method that modifies the resting membrane potentials of cortical neurons intracortically (Nitsche and Paulus 2000, 2001). Cathodal stimulation decreases, whilst anodal stimulation increases cortical excitability (Nitsche and Paulus 2000, 2001). However, in that study we have used a large electrode size (5 × 7 cm) in order to optimize stimulation’s parameters (Nitsche and Paulus 2000) and therefore we might have covered a large part of the SI. It is possible that we have stimulated one part of the somatosensory association cortex (BA 5/7) that is posterior to the SI. Activation of human BA5/7 has also been linked to pain perception (Apkarian et al. 1999; Forss et al. 2005). BA 5/7 is anatomically connected to other nociceptive brain areas such as the ACC, insula, thalamus and primary motor cortices (Friedman et al. 1986). In order to increase the focality of the transcranial stimulation in the present study we used a neuronavigation system to determine the hand area over the SI. Many of the previous pain-related studies stimulated the M1. The effects of low and high frequency rTMS of M1 on experimentally induced acute pain perception seem to depend on the type of noxious stimulation. C-fiber mediated acute pain as induced by intradermal capsaicin administration could be attenuated by 1 Hz rTMS over the M1 (Tamura et al. 2004b), whereas it increased Aδ-fiber mediated laser-induced pain in another study (Tamura et al. 2004a). Similarly, controversial effects were observed after 20 Hz rTMS (Johnson et al. 2006; Summers et al. 2004). In contrast 10 Hz rTMS over M1 increased electrically induced Aδ-fiber mediated pain threshold (Yoo et al. 2006), but others found that 10 Hz rTMS has only an effect on the unpleasantness of the pain without any effect on pain threshold (Mylius et al. 2007). In case of chronic pain the high frequency rTMS seems to more effective than the application of low frequencies (Leo and Latif 2007). The neuronal mechanism of the theta burst paradigm is highly speculative. The results of the experiments with single trains of TBS suggest that in the human M1 TBS produces a mixture of facilitatory and inhibitory effects on synaptic transmission (Huang and Rothwell 2004). Huang and Rothwell proposed, that facilitation develops faster than inhibition, thus in case of the inhibitory cTBS, several seconds after an initial facilitation the inhibition overrides this effect. ITBS uses only the early excitatory effect in the initial 2 s and after this the stimulation is interrupted for 8 s. Most likely, the underlying mechanisms will involve many of the basic elementary mechanisms described previously in the LTP/LTD literature (Paulus 2005). Recently Huang et al. (2007) have demonstrated that the after-effects produced by both iTBS or cTBS are NMDA-receptor dependent and hence they are more likely to involve plasticity-like changes at the synapse in the M1. More recently, it was found that the excitatory effects of iTBS were reversed after NMDA receptor activation by D-cycloserin (Teo et al. 2007). This is in contrast with the findings of tDCS induced LTP where the excitatory effects are prolonged by D-cycloserin (Nitsche et al. 2004). However, it is important to note that these observations were done on the M1. It is possible that the different effectiveness observed between TBS protocols on motor and sensory cortices could be due to differences in the physiological and functional states of the stimulated cortex. In summary, in our study we found a significant decrease of the N2 component of the LEPs after cTBS, iTBS, and imTBS when compared to sham stimulation over the SI. In addition we found, that the subjective pain perception did not show significant differences among the stimulation conditions suggesting that this method is probably not the most effective in decreasing subjective pain perception. Further findings show that imTBS resulted in more impressive modification of the LEPs, than were found in previous TMS-studies over the M1 using MEP measures (Huang et al. 2005), implying that it should not be used in further studies as ‘sham’ condition, at least with regard to LEPs and when it is applied over the SI. However these results are not directly comparable with the effects of TBS on M1, thus further studies are needed to clarify the effects of imTBS on different cortical areas. Future studies should also clarify the effectiveness of the different TBS paradigms applied over the M1 and non-motor cortical areas, such as the SII on acute pain perception and in chronic pain.