Datasets:

tejasexpress

/

ocr-annotation

like 0

9 5 14 1 Breast Cancer Research 24 5 15 1 et al. Vol 6 No 6 32 5 5 1 Kitayama 9 10 39 1 Previous studies also found that malignant transformation 51 10 39 1 total RNA was reverse transcribed using a SuperScript 9 11 39 1 resulted in aberrant expression of LPA2 or LPA3 in ovarian 51 11 39 1 First-Strand Synthesis System (Invitrogen Co., Carlsbad, 9 13 39 1 or thyroid cancers, suggesting that LPA may play a role in 51 13 39 1 CA, USA). The reverse transcription reaction was carried 9 14 62 1 µl, tumour biology and that shifts in LPA receptor expression 51 14 17 1 out in a total volume of 20 71 14 19 1 in accordance with the man- 9 16 39 1 are related to carcinogenesis [17-20]. However, patterns 51 16 39 1 ufacturer's instructions. The cDNA was stored at -20°C 9 17 39 1 of expression of these LPA receptors in other tumours have 51 17 6 1 until use. 9 19 39 1 not been adequately examined. In the present study we 9 20 39 1 characterized the expression patterns of LPA receptors in 51 20 22 1 Preparation of cDNA calibrators 9 22 81 1 We used pcDNA3 vector (Invitrogen) containing LPA1, human breast cancer tissue (another common form of can- 9 23 81 1 LPA2, or LPA3 as the standard cDNA for each LPA recep- cer in females) and analyzed correlations with other clinical 51 25 24 1 β-actin tor [14]. cDNA for human 9 25 81 1 was prepared using and pathological findings to determine the possible role 9 26 81 1 human colon cDNA as the template DNA and the following played by each LPA receptor in the development of breast 9 28 81 1 CCATCGAATTCACCACCATGGAT- oligonucleotides: cancer. 51 29 20 1 GATGATATCGCCGCGCTC 75 29 2 1 and 81 29 9 1 AAGGTGCG- 9 31 7 1 Methods 51 31 33 1 GCCGCCTAGAAGCATTTGCGGTGGACGAT. 88 31 2 1 The 51 32 36 1 EcoRI/NotI resulting DNA fragments were digested by 9 32 81 1 and Patients and materials 9 34 39 1 In the first part of the study, mRNA expression for each LPA 51 34 39 1 ligated into pBlueScript (Strategene, La Jolla, CA, USA). 9 35 39 1 receptor was evaluated in 25 cases of invasive ductal car- 51 35 39 1 The DNA sequence of cDNA prepared by RT-PCR was 9 37 81 1 β- cinoma, which were surgically resected in the Department 51 37 37 1 confirmed by DNA sequencing and used to calibrate for 9 38 39 1 of Surgery, University of Tokyo, from 1998 to 2003, and in 51 38 3 1 actin. 9 40 39 1 six samples of adjacent normal gland tissue. In the next part 9 41 39 1 of the study, protein expression of LPA2 was evaluated by 51 41 28 1 Real-time fluorescence quantitative PCR 9 43 81 1 The primers used in the analysis of LPA1, LPA2, LPA3 and immunohistochemical staining in the 25 cases and an addi- 9 44 81 1 β-actin gene expression are given in Table 1. All the primers tional 59 cases of invasive ductal carcinoma, which were 9 46 81 1 were designed using Primer Express software (Applied resected from 1992 to 1997, also in the Department of 9 47 81 1 Biosystems, Foster, CA, USA). Real-time PCR reactions Surgery, University of Tokyo. 51 49 39 1 were conducted in an ABI PRISM 7000 (Perkin-Elmer/ 9 50 81 1 Applied Biosystems, Foster, CA, USA) using SYBR Green All of the resected primary tumours and regional lymph 9 51 81 1 I (Perkin-Elmer) with the following profile: one step at 50°C nodes were histologically examined by haematoxylin–eosin 9 53 81 1 for 2 min, one step at 95°C for 10 min, and 40 cycles at staining, in accordance with the International Union Against 9 54 81 1 95°C for 30 s and 60°C for 1 min. Thermocycling was done Cancer TNM classification. Several discrete histological 67 56 1 1 µl 51 56 27 1 µl in a final volume of 20 68 56 8 1 containing 1 9 56 81 1 of cDNA sample. parameters and lymph node metastasis were also exam- 9 57 81 1 The ABI PRISM software constructed the calibration curve ined. Oestrogen receptor levels were evaluated using 9 59 81 1 by plotting the crossing point against the logarithm of the enzyme immunoassay of frozen tumour specimens with cut- 9 60 81 1 number of copies for each calibrator. The number of copies off levels for positivity of 5 fmol/mg protein. 51 62 39 1 in unknown samples was calculated by comparing their 9 63 81 1 crossing points with the calibration curve. To correct for dif- Isolation of total RNA and reverse transcription 9 65 39 1 The tumour tissue resected from the primary lesion and 51 65 39 1 ferences in both RNA quality and quantity between sam- 9 66 39 1 paired nontumour tissue (taken 10 cm away from the neo- 51 66 39 1 ples, data were normalized using the ratio of the target 9 68 68 1 β-actin. plasm) were immediately frozen in liquid nitrogen and kept 51 68 20 1 cDNA concentration to that of 78 68 12 1 Both RT and PCR 9 69 39 1 at -80°C until extraction of RNA. Total RNA was extracted 51 69 39 1 reactions were performed in triplicate, and the mean values 9 71 69 1 β-actin. from each sample using the acid guanidine isothiocyanate/ 51 71 21 1 were calculated against against 9 72 39 1 phenol/chloroform extraction method. One microgram of 9 75 4 1 Table 1 9 78 63 1 Lysophosphatidic acid receptor 2 (LPA2) expression evaluated by RT-PCR and immunohistochemistry 9 80 62 1 P Immunohistochemistry findings 42 80 14 1 Relative mRNA expression 30 82 3 1 <3000 50 82 3 1 >3000 9 85 8 1 Low expression 30 85 1 1 10 50 85 0 1 1 9 87 8 1 High expression 30 87 0 1 3 50 87 1 1 11 71 87 4 1 <0.001 2 92 3 1 R641

50 5 40 1 http://breast-cancer-research.com/content/6/6/R640 Available online 9 10 4 1 Table 2 9 12 68 1 The relationship between lysophosphatidic acid receptor 2 (LPA2) expression and clinical/pathological factor 9 15 62 1 P LPA2 expression 30 15 5 1 Low (36) 50 15 5 1 High (48) 9 17 6 1 Age (years) 11 19 2 1 <50 30 19 5 1 20 (56%) 50 19 5 1 16 (33%) 71 19 3 1 0.042 11 21 2 1 ≥50 30 21 5 1 16 (44%) 50 21 5 1 32 (67%) 9 23 6 1 Tumour size 11 25 4 1 (≤2.0 T1 16 25 2 1 cm) 30 25 5 1 16 (44%) 50 25 5 1 17 (35%) 71 25 1 1 NS 11 27 11 1 ≤5.0 T2 (>2.0 cm to 23 27 2 1 cm) 30 27 5 1 18 (50%) 50 27 5 1 25 (52%) 11 29 7 1 T3 (>5.0 cm) 30 29 3 1 2 (6%) 50 29 4 1 6 (13%) 9 31 7 1 Nuclear grade 11 32 0 1 I 30 32 4 1 5 (14%) 50 32 4 1 5 (10%) 71 32 1 1 NS 11 34 0 1 II 30 34 5 1 24 (67%) 50 34 5 1 31 (65%) 11 36 0 1 III 30 36 4 1 7 (19%) 50 36 5 1 12 (25%) 9 38 10 1 Menopausal status 11 40 8 1 Premenopausal 30 40 5 1 21 (58%) 50 40 5 1 17 (35%) 71 40 3 1 0.037 11 42 9 1 Postmenopausal 30 42 5 1 15 (42%) 50 42 5 1 31 (65%) 9 44 14 1 Oestrogen receptor status 11 46 4 1 Negative 30 46 5 1 10 (55%) 50 46 5 1 27 (56%) 71 46 1 1 NS 11 48 4 1 Positive 30 48 4 1 8 (45%) 50 48 5 1 21 (44%) 9 50 9 1 Nodal metastasis 11 52 4 1 Negative 30 52 5 1 12 (33%) 50 52 5 1 21 (44%) 71 52 1 1 NS 11 54 4 1 Positive 30 54 5 1 24 (67%) 50 54 5 1 27 (56%) 9 56 10 1 Distant metastasis 11 58 4 1 Negative 30 58 5 1 27 (75%) 50 58 5 1 41 (85%) 71 58 1 1 NS 11 60 4 1 Positive 30 60 4 1 9 (25%) 50 60 4 1 7 (15%) 9 62 54 1 Numbers in parentheses indicate percentages in the same expression pattern. NS, not significant. 9 65 39 1 pathological findings in these 84 cases are presented in 51 65 39 1 LPA2 was confirmed at the protein level in some of these 9 66 39 1 Table 2. High expression of LPA2 was frequently detected 51 66 39 1 cases by immunohistochemical staining. Previous studies 9 68 39 1 in relatively old postmenopausal patients. High expression 51 68 39 1 showed that LPA1 is widely expressed in various tissues, 9 69 39 1 of LPA2 was detected in 17 out of 38 (45%) premenopau- 51 69 39 1 whereas LPA2 and LPA3 are known to be highly expressed 9 71 39 1 sal patients and in 31 out of 46 (67%) postmenopausal 51 71 39 1 in malignant cells, suggesting the potential role of these 9 72 37 1 (P women; this difference was statistically significant 47 72 1 1 < 51 72 6 1 receptors 60 72 1 1 in 64 72 2 1 the 68 72 11 1 pathophysiology 82 72 1 1 of 86 72 4 1 cancer 9 74 39 1 0.05). LPA2 expression did not correlate with oestrogen 51 74 39 1 [5,18,19,21,22]. Our findings regarding LPA2 are basically 9 75 39 1 receptor expression. The frequency of nodal or distant 51 75 39 1 consistent with those results and suggest that upregulation 9 77 39 1 metastasis tended to be higher in tumours with low expres- 51 77 39 1 of the LPA2 gene is a common feature of carcinogenesis in 9 78 39 1 sion of LPA2, although the difference was not statistically 51 78 10 1 various organs. 9 80 7 1 significant. 51 81 39 1 Protein expression of LPA2 in various organs has not been 9 83 9 1 Discussion 51 83 39 1 investigated thus far because of the lack of an appropriate 9 84 39 1 In the present study we found that breast cancer frequently 51 84 39 1 antibody against LPA2. Our antibody clearly detected sig- 9 86 39 1 exhibited enhanced expression of LPA2 mRNA as com- 51 86 39 1 nificantly enhanced expression of LPA2 in more than half of 9 87 39 1 pared with normal breast gland tissue, although the expres- 51 87 39 1 the ductal carcinoma cells, and the staining pattern exhib- 9 89 39 1 sion level of LPA1 and that of LPA3 were not significantly 51 89 39 1 ited a strong correlation with the mRNA data obtained with 9 90 39 1 different from those in normal tissue. Over-expression of 51 90 39 1 quantitative RT-PCR. Normal epithelium of mammary 93 92 3 1 R644

51 5 39 1 http://breast-cancer-research.com/content/7/1/R46 Available online 9 10 4 1 Table 1 9 12 79 1 5-Bromo-2'-deoxyuridine (BrdU) labeling* of MCF10A cells grown in monoculture (control group) and in co-cultures with normal 9 14 50 1 breast-associated fibroblasts (NAF) and carcinoma-associated fibroblasts (CAF) 9 16 3 1 Culture 25 16 11 1 BrdU-labeling indices of 42 16 11 1 BrdU-labeling indices of 56 16 31 1 Comparison of BrdU-labeling indices of MCF10A cells between 26 17 11 1 MCF10A cells (mean ± 40 17 14 1 MCF10A cells (group mean ± 57 17 12 1 groups (linear regression) 25 18 13 1 standard error of the mean) 41 18 13 1 standard error of the mean) 14 20 0 1 (n 9 20 62 1 (P MCF10A 15 20 2 1 = 6) 29 20 5 1 30.3 ± 3.0 45 20 5 1 30.3 ± 3.0 56 20 14 1 MCF10A vs MCF10A + NAF 74 20 15 1 MCF10A + NAF vs MCF10A + 57 21 21 1 (P = 0.009) 75 21 2 1 CAF 79 21 4 1 = 0.501) 19 23 0 1 (n 9 23 39 1 (n NAF-1 + MCF10A 20 23 2 1 = 7) 29 23 5 1 21.9 ± 4.2 43 23 5 1 16.1 ± 2.6 49 23 2 1 = 19) 9 25 10 1 (n NAF-2 + MCF10A 20 25 2 1 = 6) 29 25 5 1 10.7 ± 3.6 9 26 10 1 (n NAF-3 + MCF10A 20 26 2 1 = 6) 29 26 5 1 14.7 ± 5.0 19 28 0 1 (n 48 28 0 1 (n 9 28 62 1 (P CAF-1 + MCF10A 20 28 2 1 = 8) 29 28 5 1 15.2 ± 2.0 43 28 5 1 18.5 ± 2.5 49 28 2 1 = 19) 56 28 14 1 MCF10A vs MCF10A + CAF 57 29 4 1 = 0.024) 9 31 10 1 (n CAF-2 + MCF10A 20 31 2 1 = 6) 29 31 5 1 15.5 ± 4.2 9 33 10 1 (n CAF-3 + MCF10A 20 33 2 1 = 5) 29 33 5 1 27.6 ± 6.7 9 35 20 1 *Assessed by immunocytochemistry 9 36 81 1 There was variability among NAF cultures and among CAF Figure 4 51 38 39 1 cultures in their ability to suppress proliferation of MCF10A 51 39 39 1 cells and MCF10AT cells (Tables 1 and 2) in this 3D cul- 51 41 39 1 ture system, potentially reflecting heterogeneity among the 51 42 39 1 individuals from which the fibroblasts were derived. 51 44 39 1 Because of this variability, detection of a significant differ- 51 45 39 1 ence in the function of NAF and CAF required many repli- 51 47 39 1 cates and multiple fibroblast cultures derived from different 51 48 7 1 individuals. 51 51 39 1 In a prior report, CAF was found to promote, rather than 51 53 39 1 inhibit, the growth of MCF10A cells in a similar 3D co-cul- 51 54 39 1 ture system [22]. One of several possible explanations for 51 56 39 1 this discrepancy between the prior result and the present 51 57 39 1 result is a difference in E:F. Shekhar and colleagues used 51 59 39 1 an E:F of 1:1 rather than the E:F of 2:1 we initially used 51 60 39 1 [22]. The number of fibroblasts has been shown to have an 51 62 39 1 effect on the response of epithelial cells [7,9,14]. We 9 63 38 1 Proliferation ture Proliferation and co-culture of of MCF10A MCF10A with fibroblasts cells cells and and MCF10AT MCF10AT cells cells grown grown in in monocul- monocul- 51 63 39 1 therefore repeated the 3D co-cultures of MCF10A cells 9 64 39 1 ture and co-culture with fibroblasts. The rate of proliferation of MCF10A 9 65 81 1 using NAF-2 and CAF-1 with increasing numbers of fibrob- cells and MCF10AT cells, as measured by the 5-bromo-2'-deoxyuridine 9 66 39 1 (BrdU) labeling index (assessed by immunocytochemistry), was signifi- 51 66 39 1 lasts (i.e. a decreasing E:F) (Fig. 6). BrdU labeling was 9 67 39 1 cantly reduced in co-cultures of MCF10A cells with both normal breast- 51 68 39 1 assessed by immunocytochemistry of histologic sections 9 69 17 1 (P associated fibroblasts (NAF) 26 69 19 1 = 0.009) and carcinoma-associated 51 69 9 1 of 3D cultures. 9 70 11 1 (P fibroblasts (CAF) 20 70 26 1 = 0.024) compared with the MCF10A monocul- 9 71 37 1 ture (control). The rate of proliferation of MCF10AT cells was signifi- 24 72 66 1 As previously, NAF-2 at an E:F of 2:1 suppressed prolifer- (P 9 72 32 1 (P cantly suppressed by NAF 25 72 13 1 = 0.013) but not by CAF 41 72 6 1 = 0. 935) in 9 73 30 1 comparison with the MCF10AT monoculture (control). 51 74 39 1 ation of MCF10A cells. However, with increasing numbers 51 75 39 1 of NAF-2, this suppression effect was gradually weakened 51 77 39 1 (P = 0.043). Although we found no significant difference in 51 78 39 1 the suppressive effect of NAF-2 in an E:F of 2:1 versus an 13 79 1 1 (n 9 79 11 1 P index 15 79 3 1 = 22, 20 79 27 1 = 0.935) (Table 2 and Fig. 4). The effect 51 79 39 1 E:F of 1:1 or of 1:2, there was a significantly greater rate of 9 81 39 1 of NAF versus CAF on the rate of proliferation of MCF10AT 51 81 39 1 proliferation of MCF10A cells with NAF-2 in an E:F of 1:3 29 82 1 1 (P 9 82 65 1 (P cells was significantly different 31 82 16 1 < 0.001). The effect was 51 82 21 1 compared with in an E:F of 2:1 75 82 15 1 = 0.028). More impor- 9 84 39 1 further confirmed by repeating the co-cultures to measure 51 84 39 1 tantly, CAF-1 at an E:F of 1:1 did not significantly suppress 9 85 79 1 (P the BrdU-labeling index by flow cytometry, rather than by 51 85 35 1 proliferation, whereas our original ratio of 2:1 did 89 85 1 1 = 9 87 20 1 immunocytochemistry (Fig. 5). 51 87 39 1 0.025). CAF-1 at an E:F of 1:2 also conferred a higher rate 51 88 39 1 of proliferation of MCF10A cells than the E:F of 2:1, but this 51 90 26 1 (P did not reach statistical significance 77 90 12 1 = 0.054). At an E:F 93 92 2 1 R52

9 5 14 1 Breast Cancer Research 24 5 16 1 et al. Vol 7 No 1 32 5 5 1 Sadlonova 9 10 4 1 Table 2 9 12 81 1 5-Bromo-2'-deoxyuridine (BrdU) labeling* of MCF10AT cells grown in monoculture and in co-cultures with normal breast-associated 9 14 38 1 fibroblasts (NAF) and carcinoma-associated fibroblasts (CAF) 9 16 3 1 Culture 26 16 11 1 BrdU-labeling indices of 42 16 11 1 BrdU-labeling indices of 57 16 31 1 Comparison of BrdU-labeling indices of MCF10AT cells between 26 17 12 1 MCF10AT cells (mean ± 41 17 14 1 MCF10AT cells (group mean 58 17 12 1 groups (linear regression) 25 18 13 1 standard error of the mean) 41 18 14 1 ± standard error of the mean) 9 20 6 1 (n MCF10AT 16 20 2 1 = 6) 29 20 5 1 27.7 ± 7.2 46 20 5 1 27.7 ± 7.2 57 20 12 1 MCF10AT vs MCF10AT + 73 20 16 1 MCF10AT + NAF vs MCF10AT + 61 21 1 1 (P 58 21 19 1 (P NAF 62 21 4 1 = 0.013) 74 21 2 1 CAF 78 21 4 1 < 0.001) 19 23 0 1 (n 9 23 40 1 (n NAF-1+ MCF10AT 20 23 2 1 = 7) 29 23 5 1 17.1 ± 2.8 43 23 5 1 14.1 ± 1.4 50 23 2 1 = 20) 9 25 10 1 (n NAF-2 + MCF10AT 21 25 2 1 = 6) 29 25 5 1 13.8 ± 1.9 9 26 10 1 (n NAF-3 + MCF10AT 21 26 2 1 = 7) 29 26 5 1 11.4 ± 1.9 19 28 0 1 (n 9 28 40 1 (n CAF-1 + MCF10AT 21 28 2 1 = 8) 29 28 5 1 25.9 ± 4.9 43 28 5 1 25.5 ± 2.8 50 28 2 1 = 22) 57 28 12 1 MCF10AT vs MCF10AT + 58 29 3 1 (P CAF 62 29 4 1 = 0.935) 9 31 11 1 (n CAF-2 + MCF10AT 21 31 2 1 = 8) 29 31 5 1 26.6 ± 3.2 9 33 11 1 (n CAF-3 + MCF10AT 21 33 2 1 = 6) 29 33 5 1 23.5 ± 7.4 9 35 20 1 *Assessed by immunocytochemistry 9 37 5 1 Figure 5 51 37 5 1 Figure 6 51 55 39 1 Relative co-culture NAF-2 Relative and 5-bromo-2'-deoxyuridine 5-bromo-2'-deoxyuridine with carcinoma-associated varying quantities of (BrdU) (BrdU) fibroblast normal indices indices breast-associated CAF-1 of of MCF10A MCF10A cells cells fibroblast in in 51 56 39 1 co-culture with varying quantities of normal breast-associated fibroblast 51 57 37 1 NAF-2 and carcinoma-associated fibroblast CAF-1. With increasing 9 59 39 1 of ated 5-Bromo-2'-deoxyuridine 5-Bromo-2'-deoxyuridine MCF10AT fibroblasts monocultures (NAF) and (BrdU) (BrdU) carcinoma-associated and co-cultures labeling, labeling, assessed assessed with normal fibroblasts by by flow flow breast-associ- cytometry, cytometry, (CAF) 51 59 34 1 numbers of NAF-2, the mean rate of proliferation of co-cultured 9 60 39 1 of MCF10AT monocultures and co-cultures with normal breast-associ- 51 60 38 1 MCF10A cells increased, with a significant difference in BrdU-labeling 9 61 37 1 ated fibroblasts (NAF) and carcinoma-associated fibroblasts (CAF). 51 61 39 1 index observed between a ratio of epithelial cells to fibroblasts (E:F) of 9 62 38 1 These data are representative of replicate experiments indicating that 51 62 15 1 (P 2:1 versus an E:F of 1:3 66 62 23 1 < 0.05). With increasing numbers of CAF- 9 63 38 1 NAF suppress proliferation of MCF10AT cells to a greater extent than 51 63 39 1 1, the mean rate of proliferation was highest at an E:F of 1:1. The rate of 9 65 36 1 do CAF. Again some variability in extent of suppression is present 51 65 39 1 proliferation at an E:F of 1:1 was significantly higher than that at an E:F 9 66 32 1 among individual NAF cultures and individual CAF cultures. 51 66 4 1 (P of 2:1 56 66 34 1 < 0.05). At an E:F of 1:3, CAF-1 caused a decreased prolifer- 51 67 28 1 ation of and enhanced cell death of MCF10A cells. 51 71 38 1 Quantities of IGF II are no different in NAF versus CAF 9 72 39 1 of 1:3, however, CAF-1 caused a decrease in proliferation 51 72 39 1 As an initial attempt to identify differences between NAF 9 74 39 1 of MCF10A cells and enhanced cell death, as assessed by 51 74 39 1 and CAF that explain our observed results, expression of 9 75 39 1 microscopic morphology. At an E:F of 1:3, the total number 51 75 39 1 IGF II in NAF and in CAF was assessed. A higher level of 9 77 39 1 of viable MCF10A cells was reduced in co-culture with 51 77 39 1 expression of IGF II in CAF than in NAF may provide an 9 78 39 1 both NAF-2 and CAF-1; however, this reduction was more 51 78 39 1 explanation for the higher rate of proliferation of MCF10AT 9 80 13 1 marked with CAF-1. 51 80 31 1 cells allowed by CAF in comparison with NAF. 9 83 39 1 Neither NAF nor CAF had a significant effect on the rate of 51 83 39 1 ELISA performed on cell lysates of NAF and CAF cultures 9 84 39 1 apoptosis of MCF10A cells or MCF10AT cells when grown 51 84 39 1 demonstrated variability in expression of IGF II among cul- 9 86 39 1 at an E:F of 2:1 after 2 weeks of co-culture, as assessed by 51 86 39 1 tures, but no significant difference was observed in the 9 87 14 1 TUNEL assay (Fig. 7). 51 87 39 1 mean IGF II quantity between NAF and CAF in monolayer 51 89 39 1 cultures (Table 3) or in 3D monocultures (Table 4). 51 90 39 1 Although in monolayer cultures more CAF than NAF had 2 92 2 1 R53

51 5 39 1 http://breast-cancer-research.com/content/7/1/R46 Available online 9 10 4 1 Table 3 9 12 80 1 Insulin-like growth factor (IGF) II ELISA for normal breast-associated fibroblasts (NAF), carcinoma-associated fibroblasts (CAF), 9 14 34 1 MCF10A cells and MCF10AT cells in monolayer cultures 9 16 4 1 Cell type 52 16 21 1 IGF II level (normalized to total protein) 49 18 4 1 (ng/µg) 46 18 34 1 (ng/µg) Level 73 18 3 1 Mean 9 21 2 1 NAF 11 22 3 1 NAF-1 48 22 3 1 11.63 11 24 3 1 NAF-2 48 24 2 1 6.87 11 26 3 1 NAF-3 48 26 3 1 10.31 11 28 3 1 NAF-4 48 28 2 1 7.21 11 30 3 1 NAF-5 48 30 2 1 2.92 76 30 2 1 7.79 9 32 2 1 CAF 11 34 3 1 CAF-1 48 34 3 1 11.80 11 36 3 1 CAF-2 48 36 3 1 10.58 11 38 3 1 CAF-3 48 38 3 1 10.59 11 40 3 1 CAF-4 48 40 3 1 11.27 11 42 3 1 CAF-5 48 42 3 1 10.21 11 44 3 1 CAF-6 48 44 3 1 14.07 75 44 3 1 11.42 9 46 11 1 Breast epithelial cells 11 48 5 1 MCF10A 48 48 2 1 5.56 11 50 5 1 MCF10AT 48 50 2 1 4.17 76 50 2 1 4.87 9 57 4 1 Table 4 9 60 81 1 Insulin-like growth factor (IGF) II ELISA for normal breast-associated fibroblasts (NAF), carcinoma-associated fibroblasts (CAF) and 9 61 46 1 in vitro MCF10AT cell monocultures and co-cultures in a three-dimensional 56 61 3 1 model 9 63 4 1 Cell type 47 63 21 1 IGF II level (normalized to total protein) 38 66 6 1 Monoculture 65 66 17 1 Co-culture with MCF10AT cells 33 68 4 1 (ng/µg) 49 68 4 1 (ng/µg) 65 68 4 1 (ng/µg) 29 68 56 1 (ng/µg) Level 46 68 3 1 Mean 62 68 2 1 Level 78 68 3 1 Mean 9 71 2 1 NAF 11 72 3 1 NAF-1 32 72 3 1 49.25 65 72 2 1 9.68 11 74 3 1 NAF-2 32 74 3 1 45.00 64 74 3 1 10.22 11 76 3 1 NAF-3 32 76 3 1 68.14 48 76 3 1 54.13 64 76 3 1 10.27 81 76 3 1 10.06 9 78 2 1 CAF 11 80 3 1 CAF-1 32 80 3 1 70.64 64 80 3 1 10.01 11 82 3 1 CAF-2 32 82 3 1 52.37 64 82 3 1 10.88 11 84 3 1 CAF-3 32 84 3 1 52.96 48 84 3 1 58.65 65 84 2 1 9.96 81 84 3 1 10.28 9 86 11 1 Breast epithelial cells 9 88 5 1 MCF10AT 32 88 2 1 9.98 93 92 2 1 R54

9 5 14 1 Breast Cancer Research 24 5 14 1 et al. Vol 7 No 1 32 5 3 1 Davies 9 10 4 1 Table 1 9 12 9 1 Case selection 9 15 5 1 Parameter 37 15 3 1 Detail 75 15 3 1 Value 9 17 6 1 Age (years) 37 17 3 1 Range 75 17 3 1 48–88 9 19 5 1 Histology 37 19 8 1 Invasive ductal 76 19 2 1 115 37 21 8 1 Invasive lobular 76 21 1 1 11 37 23 3 1 Other 76 23 1 1 10 9 25 2 1 ER-α 37 25 4 1 Positive 76 25 1 1 88 37 27 4 1 Negative 76 27 1 1 42 9 29 3 1 Grade 37 29 0 1 1 76 29 1 1 21 37 31 0 1 2 76 31 1 1 53 37 32 0 1 3 76 32 1 1 62 9 34 7 1 Nodal status 37 34 4 1 Positive 76 34 1 1 51 37 36 4 1 Negative 76 36 1 1 62 9 38 2 1 Size 37 38 3 1 <2 cm 76 38 1 1 55 37 40 3 1 ≥ 2 cm 76 40 1 1 79 9 42 3 1 Stage 37 42 0 1 1 76 42 1 1 33 37 44 0 1 2 76 44 1 1 76 37 46 0 1 3 76 46 0 1 5 9 48 27 1 ≥ Tumour (%) 38 48 1 1 50 76 48 2 1 100 37 50 2 1 ≥ 75 76 50 1 1 91 37 52 2 1 ≥ 90 76 52 1 1 63 9 54 3 1 ER-α, 13 54 12 1 receptor-α. oestrogen 9 56 39 1 mutation was closer to the reverse PCR primer (ERmut2). 51 56 39 1 100% efficient and that by PCR we are able to reamplify 9 57 39 1 The new primers designed for use on cDNA overcame this 51 57 39 1 products originating from 1% or less of the starting DNA. It 9 59 39 1 limitation, while ensuring that only cDNA was amplified. 51 59 39 1 is therefore possible to apply this reamplification and 9 60 39 1 PCR primers for genomic DNA were similarly specific for 51 60 39 1 sequencing technique to all samples, because the PCR 9 62 39 1 genomic DNA and, when used in conjunction with a 51 62 39 1 used routinely amplifies non-digested DNA enriched for 9 63 39 1 sequencing primer (ERADNA3), again gave sequence in 51 63 14 1 mutant PCR product. 9 65 10 1 both directions. 51 66 39 1 The enhanced sensitivity after enrichment using digestion 19 68 3 1 MboII 9 68 49 1 MboII Digestion with 23 68 25 1 restriction enzyme allowed the detec- 51 68 2 1 with 58 68 31 1 and reamplification is not without drawbacks, in 9 69 39 1 tion of as little as 1% mutant DNA by agarose gel electro- 51 69 39 1 that we detected a greater number of PCR and sequencing 9 71 39 1 phoresis in control reactions (Fig. 2). However, no evidence 51 71 39 1 anomalies with this technique. For non-enriched sequence 9 72 39 1 of undigested PCR product was visible for any breast 51 72 39 1 analysis, evidence of a very minor G in genomic DNA from 9 74 39 1 tumour assayed in this way. When sequencing control 51 74 39 1 two breast cancers was subsequently shown to be due to 9 75 39 1 DNA reamplified with ERmut1 and ERmut2 after enrich- 51 75 39 1 sequencing anomalies because they were not present 9 77 22 1 MboII ment for mutant DNA by 32 77 15 1 digestion (Fig. 1), the 51 77 39 1 either in repeat PCR products from the same cases or in 9 78 39 1 mutant G base was clearly detected even when only 51 78 39 1 sequence generated in the reverse direction. After enrich- 9 80 39 1 present in 1% of the original DNA. The wild-type A was also 51 80 39 1 ment, similar errors were noted in eight genomic DNA PCR 9 81 39 1 detected, either because of inefficient digestion or because 51 81 39 1 products and six cDNA PCR products. Ten of these anom- 9 83 64 1 Taq of heterodimer formation in the PCR reaction. Notably in 51 83 18 1 alies were apparently due to 73 83 17 1 polymerase infidelity (that 9 84 32 1 ER-α, control reactions containing only wild-type 42 84 6 1 any post- 51 84 39 1 is, they were identified in reverse sequencing but not 9 86 39 1 digest reamplified DNA was clearly wild type. Therefore, 51 86 39 1 repeatable in replicate PCR) and four were sequencing 9 87 39 1 despite a failure to detect non-digested PCR product on 51 87 39 1 anomalies (that is, they were not identifiable in reverse 9 89 68 1 Taq gels stained with ethidium bromide, reamplification and 51 89 23 1 sequencing). Notably, errors due to 78 89 12 1 polymerase infidel- 9 90 39 1 sequencing confirms that restriction digestion is seldom 51 90 39 1 ity were seen only with the more sensitive assay based on 2 92 3 1 R116

50 5 40 1 http://breast-cancer-research.com/content/7/2/R204 Available online 9 10 4 1 Table 1 9 12 40 1 Single-nucleotide polymorphisms (SNPs) selected for genotyping 9 15 6 1 SNP name 30 15 5 1 dbSNP id 50 15 4 1 Location 77 15 7 1 Base change 9 17 3 1 SNP 1 30 17 6 1 rs4252596 50 17 7 1 START -657a 79 17 3 1 C>Ab 9 19 3 1 SNP 2 30 19 6 1 rs2952155 50 19 4 1 intron 1 79 19 2 1 C>T 9 21 3 1 SNP 3 30 21 6 1 rs1810132 50 21 4 1 intron 4 79 21 2 1 T>C 9 23 3 1 SNP 4 30 23 6 1 rs1801200 50 23 9 1 exon 17 (I655V) 79 23 2 1 A>G 9 25 3 1 SNP 5 30 25 6 1 rs1058808 50 25 10 1 exon 27 (A1170P) 79 25 2 1 G>C 9 27 41 1 a657 base pairs upstream first translated base. bCommon allele given first. 9 29 39 1 determined using standard statistical methods [21]. We 51 29 5 1 Figure 1 9 31 39 1 have over 90% power at the 1% significance level to detect 9 32 39 1 a dominant allele with a frequency of 0.05, which confers a 9 34 39 1 relative risk of 1.5, or a dominant allele with a frequency of 9 35 39 1 0.2 that confers a relative risk of 1.3. Power to detect 9 37 39 1 recessive alleles at the 1% significance level is more lim- 9 38 39 1 ited: 59% for an allele with a frequency of 0.2 that confers 9 40 39 1 a relative risk of 1.5 or 77% for an allele with a frequency of 9 41 39 1 0.3 that confers a relative risk of 1.4. The LDA program [22] 9 43 39 1 was used to calculate pairwise linkage disequilibrium (LD) 9 44 39 1 for each SNP pair in the whole case–control set. The 9 46 39 1 haplo.score program [23] was used to test for association 9 47 39 1 between haplotypes and breast cancer risk. Haplo.score 9 49 39 1 uses a likelihood that depends on estimated haplotype fre- 9 50 39 1 quencies to test the statistical association between haplo- 9 51 39 1 types and phenotype. It is based on score statistics, which 9 53 39 1 provide both global tests and haplotype-specific tests [23]. 9 56 6 1 Results 9 57 39 1 The median age was 48 years (range 25–54) for prevalent 9 59 39 1 cases, 52 years (26–55) for incident cases, and 56 years 9 60 39 1 (25–81) for controls. Incident and prevalent cases were 9 62 26 1 (P similar regarding breast cancer stage 36 62 12 1 = 0.12) and histo- 9 63 10 1 (P logical grade 20 63 28 1 = 0.41). Table 2 shows the genotype fre- 9 65 39 1 quencies in cases and controls as well as genotype- 9 66 39 1 specific risks for the five SNPs assayed. The genotype fre- 9 68 39 1 quencies were similar in the prevalent and incident cases 9 69 39 1 for all polymorphisms (data not shown). None of the geno- 51 69 37 1 r2 Pairwise (right linkage top half) disequilibrium for the five (LD) SNPs measures of D' (left bottom half) Pairwise and 56 69 32 1 linkage disequilibrium (LD) measures of D' (left bottom half) 9 70 63 1 r2 (right top half) for the five SNPs. and type distributions for the controls differed significantly from 9 72 39 1 those expected under Hardy–Weinberg equilibrium. There 9 74 81 1 case–control set, common haplotypes constituted 98% of was no evidence that any of the SNPs is associated with 9 75 81 1 all the observed haplotypes. Two haplotypes (haplotypes 3 breast cancer; genotype-specific ORs were all close to 9 77 81 1 and 5) contained the SNP 4 (I655V) minor allele. The glo- unity with narrow confidence intervals. We also compared 51 78 20 1 (P bal test was not significant 9 78 81 1 = 0.44), nor were there any genotype frequencies within cases stratified by disease 9 80 81 1 differences between cases and controls for individual hap- stage and age group for SNP 4 (I655V). No differences 16 81 1 1 (P 30 81 1 1 P 9 81 81 1 lotypes. Similarly, no differences in haplotype frequencies were seen 19 81 10 1 [stage] = 0.61, 31 81 16 1 [age group] = 0.33). LD 51 83 37 1 (P were seen within cases stratified by disease stage 9 83 81 1 = was strong (D' > 0.7) across pairs involving SNPs 1, 2, 3, 51 84 14 1 (P 0.37) or age group 9 84 63 1 = 0.48). and 5, whereas SNP 4 was in weak LD (D' < 0.3) with all 9 86 39 1 other polymorphisms except SNP 1 (D' [SNP 1-SNP 4] = 9 87 51 1 Discussion (r2 0.98) (Fig. 1). SNPs 3 and 5 were in nearly perfect LD 9 89 39 1 = 0.92). Of 32 possible haplotypes, only 6 were observed 51 89 39 1 Our study is the largest case–control study reported on 9 90 47 1 ERBB2 with a frequency greater than 5% (Table 3). For the whole 57 90 33 1 genetic variation. To our knowledge, this is also the 93 92 3 1 R206

9 5 14 1 Breast Cancer Research 24 5 16 1 et al. Vol 7 No 2 32 5 5 1 Benusiglio 9 10 4 1 Table 2 9 12 64 1 Genotype frequencies and genotype-specific risks in 2192 women with breast cancer and 2257 controls 9 15 75 1 P SNP 17 15 3 1 Series 29 15 5 1 Rare-allele 37 15 10 1 Common homozygote 49 15 8 1 Heterozygote No. 60 15 8 1 Rare homozygote 70 15 9 1 Number genotyped 85 15 2 1 (2df) 30 16 4 1 frequency 41 16 3 1 No. (%) 53 16 1 1 (%) 62 16 3 1 No. (%) 9 18 3 1 SNP 1 17 18 3 1 Cases 31 18 2 1 0.13 40 18 5 1 1618 (77) 51 18 4 1 447 (21) 63 18 3 1 42 (2) 73 18 2 1 2107 17 20 4 1 Controls 31 20 2 1 0.13 40 20 5 1 1645 (75) 51 20 4 1 511 (23) 63 20 3 1 33 (2) 73 20 2 1 2189 84 20 2 1 0.13 17 22 6 1 OR (95% CI) 41 22 2 1 1 (ref) 49 22 8 1 0.89 (0.77–1.03) 60 22 8 1 1.29 (0.82–2.05) 9 23 3 1 SNP 2 17 23 3 1 Cases 31 23 2 1 0.25 40 23 5 1 1151 (57) 51 23 4 1 734 (36) 62 23 3 1 140 (7) 73 23 2 1 2025 17 25 4 1 Controls 31 25 2 1 0.26 40 25 5 1 1219 (55) 51 25 4 1 839 (38) 62 25 3 1 147 (7) 73 25 2 1 2205 84 25 2 1 0.48 17 27 6 1 OR (95% CI) 41 27 2 1 1 (ref) 49 27 8 1 0.93 (0.82–1.05) 60 27 8 1 1.01 (0.79–1.29) 9 29 3 1 SNP 3 17 29 3 1 Cases 31 29 2 1 0.32 40 29 4 1 962 (47) 51 29 4 1 855 (42) 62 29 4 1 219 (11) 73 29 2 1 2036 17 31 4 1 Controls 31 31 2 1 0.32 40 31 5 1 1022 (46) 51 31 4 1 956 (43) 62 31 4 1 230 (11) 73 31 2 1 2208 84 31 2 1 0.69 17 33 6 1 OR (95% CI) 41 33 2 1 1 (ref) 49 33 8 1 0.95 (0.84–1.08) 60 33 8 1 1.01 (0.82–1.24) 9 34 3 1 SNP 4 17 34 3 1 Cases 31 34 2 1 0.25 40 34 5 1 1128 (57) 51 34 4 1 748 (37) 62 34 3 1 113 (6) 73 34 2 1 1989 17 36 4 1 Controls 31 36 2 1 0.25 40 36 5 1 1230 (57) 51 36 4 1 791 (37) 62 36 3 1 134 (6) 73 36 2 1 2155 84 36 2 1 0.69 17 38 6 1 OR (95% CI) 41 38 2 1 1 (ref) 49 38 8 1 1.03 (0.91–1.17) 60 38 8 1 0.92 (0.71–1.20) 9 40 3 1 SNP 5 17 40 3 1 Cases 31 40 2 1 0.33 40 40 4 1 911 (45) 51 40 4 1 870 (43) 62 40 4 1 233 (12) 73 40 2 1 2014 17 42 4 1 Controls 31 42 2 1 0.33 40 42 4 1 960 (44) 51 42 4 1 983 (45) 62 42 4 1 238 (11) 73 42 2 1 2181 84 42 2 1 0.45 17 44 6 1 OR (95% CI) 41 44 2 1 1 (ref) 49 44 8 1 0.93 (0.82–1.06) 60 44 8 1 1.03 (0.84–1.26) 9 45 30 1 OR, odds ratio; SNP, single-nucleotide polymorphism. 9 48 4 1 Table 3 9 51 34 1 Estimated haplotype frequencies in cases and controls 9 53 73 1 Pb Haplotype 26 53 3 1 Allelea 44 53 10 1 Frequency (cases) 60 53 11 1 Frequency (controls) 9 55 1 1 (1) 26 55 2 1 cctIA 48 55 2 1 0.35 65 55 2 1 0.36 81 55 2 1 0.80 9 57 1 1 (2) 26 57 2 1 ctcIP 48 57 2 1 0.19 65 57 2 1 0.18 81 57 2 1 0.88 9 59 1 1 (3) 26 59 3 1 cctVA 48 59 2 1 0.18 65 59 2 1 0.16 81 59 2 1 0.15 9 61 1 1 (4) 26 61 2 1 actIA 48 61 2 1 0.13 65 61 2 1 0.13 81 61 2 1 0.49 9 63 1 1 (5) 26 63 3 1 ctcVP 48 63 2 1 0.06 65 63 2 1 0.08 81 63 2 1 0.07 9 65 1 1 (6) 26 65 3 1 cccIP 48 65 2 1 0.07 65 65 2 1 0.07 81 65 2 1 0.50 9 67 1 1 (7) 26 67 2 1 cctIP 48 67 2 1 0.02 65 67 2 1 0.01 81 67 2 1 0.93 9 69 53 1 P aPolymorphisms in 5' to 3' order as indicated in Table 1. bGlobal score statistic = 6.91, df = 7, 63 69 3 1 = 0.44 9 72 14 1 ERBB2 first study on 24 72 24 1 reporting results for more than two 51 72 39 1 over 90% power to detect a risk of this magnitude at the 9 73 81 1 level of significance. This suggests that previous posi- polymorphisms and looking for involvement of haplotypes 51 73 2 1 10-4 9 75 81 1 tive findings may have been due to type I statistical errors. in breast cancer predisposition. We performed a study of 9 76 81 1 Neither could we replicate findings associating I655V with five common SNPs and found no evidence for association 9 78 81 1 low-stage breast cancer or with breast cancer in younger with breast cancer risk. Four of the polymorphisms may be 9 79 81 1 women [12,13]. Positive results from stratified analyses functional: SNP 1 near the promoter region and SNP 2 in 9 81 81 1 should be treated with caution; very large sample sizes are intron 1 could be involved in regulatory processes whereas 9 82 81 1 required to obtain reliable results, the number of possible SNP 4 and SNP 5 are nonsynonymous coding SNPs that 9 84 81 1 analyses that can be undertaken is large, and there is a could affect tyrosine kinase activity or protein structure [7]. 9 85 81 1 strong possibility that one or more tests will be statistically Two association studies have previously reported a positive 9 87 81 1 significant simply by chance [24]. We could not carry out association between SNP 4 (I655V) and breast cancer risk 9 88 81 1 analyses within cases stratified by family history, because [8,14]. Both genotyped about 700 individuals and showed 9 90 81 1 we only had incomplete family history data [15]. To investi- a similarly increased risk for carriers of the Val allele (OR = 9 91 81 1 gate the possibility that a common polymorphism not 1.4). We were not able to replicate these findings. We have 2 92 3 1 R207

9 5 17 1 Arthritis Research & Therapy 27 5 13 1 et al. Vol 7 No 1 35 5 2 1 Giles 9 10 4 1 Table 1 9 12 70 1 The range of IgG concentrations (ng/ml) produced by expression of the 32 heavy chain/light chain combinations 9 15 6 1 Heavy chain 19 15 16 1 Light chain contributing CDR1 37 15 23 1 Light chain contributing CDR2 and CDR3 63 15 9 1 Light chain name 75 15 14 1 IgG concentration (ng/ml) 9 17 1 1 IS4 26 17 1 1 IS4 48 17 1 1 IS4 67 17 1 1 IS4 80 17 4 1 24–368 9 19 1 1 IS4 26 19 1 1 IS4 48 19 1 1 B3 67 19 1 1 IB 80 19 4 1 22–140 9 21 1 1 IS4 26 21 1 1 IS4 48 21 2 1 UK4 67 21 1 1 IU 80 21 4 1 70–194 9 23 1 1 IS4 26 23 1 1 B3 48 23 1 1 B3 67 23 1 1 B3 81 23 3 1 5–14 9 25 1 1 IS4 26 25 1 1 B3 48 25 1 1 IS4 67 25 1 1 BI 80 25 3 1 50–60 9 27 1 1 IS4 26 27 1 1 B3 48 27 2 1 UK4 67 27 1 1 BU 81 27 3 1 5–60 9 29 1 1 IS4 26 29 2 1 UK4 48 29 2 1 UK4 67 29 2 1 UK4 80 29 3 1 11–22 9 31 1 1 IS4 26 31 2 1 UK4 48 31 1 1 IS4 67 31 1 1 UI 80 31 4 1 50–480 9 32 1 1 IS4 26 32 2 1 UK4 48 32 1 1 B3 67 32 1 1 UB 81 32 3 1 9–50 9 34 1 1 B3 26 34 1 1 IS4 48 34 1 1 IS4 67 34 1 1 IS4 80 34 4 1 71–192 9 36 1 1 B3 26 36 1 1 IS4 48 36 1 1 B3 67 36 1 1 IB 80 36 3 1 41–96 9 38 1 1 B3 26 38 1 1 IS4 48 38 2 1 UK4 67 38 1 1 IU 80 38 4 1 89–376 9 40 1 1 B3 26 40 1 1 B3 48 40 1 1 B3 67 40 1 1 B3 81 40 3 1 3.5–6 9 42 1 1 B3 26 42 1 1 B3 48 42 1 1 IS4 67 42 1 1 BI 80 42 5 1 120–608 9 44 1 1 B3 26 44 1 1 B3 48 44 2 1 UK4 67 44 1 1 BU 80 44 3 1 40–68 9 46 1 1 B3 26 46 2 1 UK4 48 46 2 1 UK4 67 46 2 1 UK4 81 46 3 1 8–28 9 48 1 1 B3 26 48 2 1 UK4 48 48 1 1 IS4 67 48 1 1 UI 80 48 4 1 60–480 9 50 1 1 B3 26 50 2 1 UK4 48 50 1 1 B3 67 50 1 1 UB 81 50 3 1 2–20 9 52 1 1 IS4 23 52 8 1 B3(Arg27aSer) 48 52 1 1 B3 67 52 2 1 B3a 80 52 3 1 48–60 9 54 1 1 B3 23 54 8 1 B3(Arg27aSer) 48 54 1 1 B3 67 54 2 1 B3a 81 54 3 1 2.5–4 9 56 3 1 IS4VHi 26 56 1 1 IS4 48 56 1 1 IS4 67 56 1 1 IS4 80 56 3 1 50–56 9 58 3 1 IS4VHii 26 58 1 1 IS4 48 58 1 1 IS4 67 58 1 1 IS4 80 58 3 1 65–70 9 60 4 1 IS4VHiii 26 60 1 1 IS4 48 60 1 1 IS4 67 60 1 1 IS4 80 60 3 1 48–90 9 62 4 1 IS4VHiv 26 62 1 1 IS4 48 62 1 1 IS4 67 62 1 1 IS4 80 62 3 1 48–90 9 64 3 1 IS4VHx 26 64 1 1 IS4 48 64 1 1 IS4 67 64 1 1 IS4 80 64 3 1 78–94 9 66 5 1 IS4VHi&ii 26 66 1 1 IS4 48 66 1 1 IS4 67 66 1 1 IS4 80 66 3 1 74–80 9 68 3 1 IS4VHi 26 68 1 1 B3 48 68 1 1 B3 67 68 1 1 B3 80 68 3 1 24–54 9 70 3 1 IS4VHii 26 70 1 1 B3 48 70 1 1 B3 67 70 1 1 B3 82 70 1 1 30 9 72 4 1 IS4VHiii 26 72 1 1 B3 48 72 1 1 B3 67 72 1 1 B3 80 72 3 1 30–34 9 74 4 1 IS4VHiv 26 74 1 1 B3 48 74 1 1 B3 67 74 1 1 B3 80 74 3 1 28–30 9 76 3 1 IS4VHx 26 76 1 1 B3 48 76 1 1 B3 67 76 1 1 B3 80 76 3 1 32–34 9 77 5 1 IS4VHi&ii 26 77 1 1 B3 48 77 1 1 B3 67 77 1 1 B3 80 77 3 1 32–47 9 80 79 1 IgG concentrations in COS-7 cell supernatants were determined by ELISA. The hybrid light chains were named by combining the names of the 9 81 80 1 two parent antibodies such that the first letter represented the antibody from which the complementarity determining region (CDR) 1 was derived 9 82 80 1 and the last letter represented the antibody from which both the CDR2 and CDR3 were derived. At least two expression experiments were carried 9 83 69 1 out for each combination; identical concentrations were obtained for IS4VHii/B3VL from two different expression experiments. 9 85 39 1 The importance of arginine residues at specific positions in 51 85 39 1 [27,37,41-43]. In general, these studies have shown that 9 86 39 1 and VL sequences of anti-DNA antibodies has been the VH 51 86 39 1 altering the numbers of arginine residues in the CDRs of 9 88 39 1 in examined by many groups, by expressing the antibodies 51 88 39 1 these antibodies can lead to significant alterations in bind- 9 89 2 1 vitro 12 89 77 1 CDR3 often play a particularly and then altering the sequence of the expressed 51 89 18 1 ing to DNA. Arginines in VH 9 91 81 1 important role in binding to this antigen [27,37,41-43]. immunoglobulins by chain swapping or mutagenesis 2 92 2 1 R52

55 5 35 1 Available online http://ccforum.com/content/8/6/R385 9 10 4 1 Table 1 9 12 74 1 Demographic data, duration of use of endotracheal tube and pediatric airway exchange catheter, and reintubation ratio 9 15 45 1 (n Parameter 50 15 3 1 Value 55 15 3 1 = 36) 9 17 5 1 Sex (F/M) 50 17 11 1 28/8 (77.8%/22.2%) 9 19 6 1 Age (years) 50 19 15 1 52.6 ± 10.8 (range 19–76) 9 21 5 1 Pathology 11 23 10 1 Maxillofacial trauma 50 23 6 1 13 (36.1%) 11 25 7 1 Neck surgery 50 25 6 1 14 (38.8%) 11 27 15 1 Maxillofacial cancer surgery 50 27 5 1 9 (25.0%) 9 29 16 1 Endotracheal tube, oral/nasal 50 29 10 1 18/18 (50%/50%) 9 31 23 1 Duration of endotracheal intubation (days) 50 31 14 1 2.8 ± 1.6 (range 0.1–10) 9 32 9 1 Reintubation ratio 50 32 7 1 4/36 (11.1%) 9 34 12 1 Duration of PAECa (h) 50 34 13 1 10.4 ± 4.2 (range 4–24) 9 36 23 1 PAEC, pediatric airway exchange catheter. 9 37 26 1 aIn 32 patients who did not require reintubation. 9 40 4 1 Case 1 51 40 5 1 Figure 2 9 41 39 1 The reason for reintubation of this male patient, who had 9 43 39 1 undergone radical neck surgery for cancer and had been intu- 9 44 39 1 bated easily by direct laryngoscopy before the operation, was 9 46 39 1 excessive surgical bleeding and haematoma, which developed 9 47 39 1 2 hours after extubation. The patient was immediately taken to 9 49 39 1 the operating room. He could not be ventilated effectively by 9 50 39 1 bag-valve-mask during the induction of anaesthesia (fentanyl 2 9 51 39 1 µg/kg, propofol 2 mg/kg, vecronium 0.1 mg/kg) and his oxy- 9 53 39 1 gen saturation decreased to 85%. He was reintubated orally 9 54 39 1 over the PAEC with the assistance of a laryngoscope within a 9 56 39 1 few seconds by using an 8 mm ETT. During observation with 9 57 39 1 a laryngoscope, reintubation of this patient by direct laryngos- 9 59 39 1 copy was thought to be nearly impossible because the glottis 9 60 39 1 could not be seen as a result of the anatomic abnormality 9 62 39 1 caused by haematoma. He was extubated again using the 9 63 39 1 PAEC 24 hours after his second operation; the PAEC was 9 65 24 1 removed again 6 hours after insertion. 51 65 13 1 A as A patient patient case 3 who who in the underwent underwent text 65 65 25 1 maxillofacial maxillofacial surgery surgery due due to to trauma, trauma, presented presented 51 66 38 1 as case 3 in the text. She was extubated with the use of the pediatric 51 67 37 1 airway exchange catheter (PAEC), and required reintubation after 6 9 68 4 1 Case 2 51 69 37 1 hours of extubation. This was easily achieved over the PAEC without 9 69 39 1 The second patient (a male), who had also undergone neck 51 70 10 1 cutting the archbar. 9 71 39 1 surgery (unilateral dissection), was intubated with difficulty 9 72 39 1 using a Fasttrach (intubating laryngeal mask airway) because 9 74 81 1 romuscular relaxation. During laryngoscopic observation we of anatomical abnormalities, which developed as a result of 9 75 81 1 could not see the glottis. In this patient, a surgical tracheotomy previous operations and radiotherapy. He was extubated 4 9 77 81 1 was performed later because of recurrent aspiration and the hours after the operation in accordance with the criteria men- 9 78 58 1 need for tracheal suction. tioned above. However, he required emergency reintubation 9 80 39 1 18 hours after extubation because he suffered acute respira- 9 81 39 1 tory distress following aspiration and bronchospasm. We 51 81 4 1 Case 3 9 83 81 1 This patient (a female) was admitted to the ICU after she oper- found out from the history obtained from his relatives that the 9 84 81 1 ation for maxillofacial trauma. She had been intubated nasally patient had already had a swallowing disorder before the oper- 9 86 81 1 by direct laryngoscopy using a Magill forceps; she could not ation and suffered from aspiration. Thus, we prolonged the 9 87 81 1 open her mouth after the operation because of inter-maxillary presence of the PAEC. Reintubation of this hypoxic patient 9 89 81 1 fixation (Fig. 2). Six hours after extubation her arterial CO2 was urgently achieved over the PAEC with the assistance of a 9 90 39 1 laryngoscope using a 7.5 mm ETT under sedation and neu- 51 90 39 1 pressure increased, and she became confused as a results of 93 92 3 1 R387

55 5 35 1 Available online http://ccforum.com/content/8/6/R422 9 10 4 1 Table 1 9 12 29 1 Demographic data of the patients investigated 9 15 5 1 Parameter 68 15 3 1 Value 9 17 10 1 Number of patients 69 17 2 1 106 9 19 6 1 Age (years) 67 19 6 1 66.6 ± 18.3 9 21 15 1 (n) Ratio of males to females 68 21 3 1 88/18 9 23 9 1 APACHE II score 67 23 6 1 20.1 ± 6.5 9 25 10 1 (n Clinical category 20 25 2 1 [%]) 11 27 4 1 Medical 67 27 6 1 64 (60.3%) 13 29 4 1 Cancer 67 29 6 1 21 (19.8%) 13 31 7 1 Neurological 67 31 6 1 30 (28.2%) 11 32 4 1 Surgical 67 32 6 1 42 (39.7%) 9 35 33 1 APACHE, Acute Physiology and Chronic Health Evaluation. 9 37 81 1 tures of tracheal aspirates was 81% and the specificity was cultures from tracheal aspirates were calculated according to 9 38 81 1 23%. The likelihood ratio for a positive test was 1.05 and the standard formulae. All samples were collected on the day of 9 40 81 1 likelihood ratio for a negative one was 0.83. The positive pre- clinical and radiological evaluation. 51 41 39 1 dictive value was 18% and the negative predictive value was 9 43 6 1 Results 51 43 3 1 86%. 9 44 39 1 A total of 106 patients were prospectively evaluated during 9 46 81 1 For quantitative analysis, among the 219 evaluations, 117 had the study period. The mean age (± standard error) was 66.6 ± 51 47 0 1 ≥ 9 47 79 1 ≥ 105 18.3 years. A total of 88 patients (83.0%) were male and 18 55 47 31 1 cfu/ml in tracheal secretions (53.4%) and 49 had 88 47 2 1 106 9 49 39 1 (17.0%) were female. The mean Acute Physiology and 51 49 39 1 cfu/ml (22.4%). In the VAP group, 25 of the 38 evaluations 54 50 0 1 ≥ 9 50 74 1 ≥ Chronic Health Evaluation II score was 20.1 ± 6.5. Medical 51 50 2 1 had 55 50 26 1 105 cfu/ml (65.8%) and 10 of them had 84 50 6 1 106 cfu/ml 9 52 39 1 patients constituted the majority (60.38%) compared with sur- 51 52 39 1 (26.3%). Thus, for 105 cfu/ml the sensitivity was 65% and the 9 53 39 1 gical patients (39.62%; Table 1). Among medical patients, 30 51 53 39 1 specificity was 48%. The likelihood ratio of a positive test was 9 54 39 1 (28.2%) were neurological and 21 (19.8%) were cancer 51 54 39 1 1.25 and the likelihood ratio of a negative test was 0.73. The 9 56 5 1 patients. 51 56 39 1 positive predictive value was 21% and the negative predictive 51 57 39 1 value was 87%. For 106 cfu/ml the sensitivity was 26% and 9 59 39 1 In these 106 patients, a total of 314 clinical evaluations were 51 59 39 1 the specificity was 78%. The likelihood ratio of a positive test 9 60 39 1 conducted and endothracheal aspirates collected, corre- 51 60 39 1 was 1.18 and the likelihood ratio of a negative test was 0.95. 9 62 39 1 sponding to 42.3 ± 36.5 days (mean ± standard error) of 51 62 39 1 The positive predictive value was 20% and the negative pre- 9 63 39 1 mechanical ventilation. In 95 of these evaluations the radiolog- 51 63 20 1 dictive value was 83% (Table 3). 9 65 39 1 ical or laboratory investigations for VAP were incomplete at the 9 66 39 1 time of clinical evaluation, and so these evaluations were 51 66 39 1 In the VAP group leucocytosis was present in 26 evaluations 9 68 39 1 excluded. Therefore, a total of 219 evaluations in 106 patients 51 68 39 1 (68.4%) and fever in 24 (63.1%), and purulent endotracheal 9 69 18 1 were included in the analysis. 51 69 39 1 secretions were observed by the therapist in 22 (57.8%) eval- 51 71 39 1 uations. In four evaluations only (10.5%) was blood culture 9 72 39 1 Thirty-eight (17.4%) evaluations were classified as 'with VAP' 51 72 39 1 positive for the same agent as was isolated in endotracheal 9 74 39 1 in 33 patients and 181 (82.6%) were classified as 'without 51 74 13 1 secretions (Table 4). 9 75 39 1 VAP' in 73 patients (Table 2). The overall concordance 9 77 39 1 between the first two observers for a diagnosis of VAP in the 51 77 39 1 Overall, in 96.8% of evaluations patients were receiving at 9 78 39 1 total population was high (94%). Within the VAP group, the 51 78 39 1 least one antibiotic. Prescription of antibiotics for three or 9 80 39 1 overall concordance between the first two observers was 51 80 39 1 more days before data collection was high (86.7%). The most 9 81 4 1 86.9%. 51 81 6 1 frequently 59 81 8 1 administered 69 81 6 1 antibiotics 77 81 3 1 were 81 81 9 1 glycopeptides 51 83 5 1 (49.7%), 58 83 6 1 antifungals 67 83 5 1 (42.4%), 74 83 10 1 third-generation 86 83 4 1 cepha- 9 84 76 1 losporins (39.2%), or carbapenem (34.2%; Table 5). Qualitative and quantitative analyses 9 86 39 1 For qualitative analysis, among all 219 evaluations, 168 9 87 39 1 (76.7%) yielded cultures that were positive for at least one 51 87 39 1 Considering all VAP episodes, the most frequently isolated 60 89 15 1 Staphylococcus aureus 9 89 81 1 P. aeruginosa agent. In the VAP group, 31 of the 38 evaluations yielded pos- 51 89 8 1 agents were 75 89 5 1 (15.7%), 62 90 9 1 Acinetobacter 9 90 70 1 baumanii itive cultures (81.6%). Thus, the sensitivity of qualitative cul- 51 90 5 1 (15.7%) 58 90 2 1 and 80 90 4 1 (7.3%). 87 90 3 1 Fungi 93 92 3 1 R424

9 5 8 1 Critical Care 19 5 10 1 December 2004 31 5 6 1 Vol 8 No 6 40 5 9 1 Camargo et al. 9 10 4 1 Table 2 9 12 40 1 Clinical characteristics of the patients in the events investigated. 9 15 5 1 Parameter 44 15 11 1 Total evaluations (%) 69 15 16 1 Evaluations in VAP group (%) 9 17 12 1 Number of evaluations 46 17 6 1 219 (100%) 74 17 6 1 38 (100%) 9 19 9 1 (n Tracheostomy 19 19 2 1 [%]) 46 19 6 1 59 (26.9%) 74 19 6 1 10 (26.3%) 9 21 7 1 (n Atelectasisa 18 21 2 1 [%]) 47 21 5 1 13 (6.0%) 74 21 5 1 6 (15.7%) 17 23 3 1 (n [%]) 46 23 6 1 35 (16.0%) 9 23 70 1 9 (23.6%) Lung edemaa 9 25 10 1 (n Lung contusiona 20 25 2 1 [%]) 47 25 4 1 6 (2.7%) 74 25 4 1 1 (2.6%) 19 27 3 1 (n [%]) 47 27 5 1 21 (9,5%) 9 27 70 1 8 (21.0%) Pleural effusiona 9 29 13 1 (n Previous lung disease 23 29 2 1 [%]) 46 29 6 1 30 (13.6%) 74 29 5 1 6 (15.7%) 13 31 3 1 COPD 47 31 5 1 16 (7.3%) 74 31 4 1 3 (7.8%) 13 32 4 1 Cancer 47 32 5 1 11 (5.0%) 74 32 4 1 2 (5.2%) 13 34 4 1 Asthma 47 34 4 1 2 (0.9%) 74 34 4 1 1 (2.6%) 13 36 10 1 Pulmonary fibrosis 47 36 4 1 1 (0.4%) 75 36 3 1 None 9 38 10 1 (n Broncoaspiration 21 38 2 1 [%]) 47 38 5 1 12 (5.4%) 74 38 4 1 3 (7.8%) 9 40 5 1 (n Sepsis 15 40 2 1 [%]) 46 40 6 1 46 (21.0%) 74 40 6 1 14 (36.8%) 9 42 4 1 (n ARDS 15 42 2 1 [%]) 47 42 5 1 12 (5,4%) 74 42 4 1 3 (7.8%) 9 44 8 1 (n Renal failure 18 44 15 1 [%]; creatinine >2.0 mg/dl) 46 44 6 1 91 (41.5%) 74 44 6 1 20 (52.6%) 9 46 6 1 (n Diabetes 16 46 2 1 [%]) 46 46 6 1 35 (16.0%) 74 46 5 1 5 (13.1%) 9 48 9 1 (n Chemotherapy 19 48 2 1 [%]) 47 48 5 1 13 (6.0%) 74 48 4 1 2 (5.2%) 9 50 8 1 (n Radiotherapy 19 50 2 1 [%]) 47 50 4 1 2 (0.9%) 75 50 3 1 None 9 52 16 1 (n Immunossupressants drugs 26 52 2 1 [%]) 47 52 4 1 5 (2.2%) 74 52 4 1 1 (2.6%) 9 54 4 1 (n AIDS 14 54 2 1 [%]) 47 54 4 1 3 (1.3%) 75 54 3 1 None 9 56 13 1 (n Renal transplantation 23 56 2 1 [%]) 47 56 4 1 5 (2.2%) 74 56 4 1 1 (2.6%) 9 58 11 1 (n Abdominal surgery 22 58 2 1 [%]) 46 58 6 1 32 (14.6%) 74 58 5 1 7 (18.4%) 9 60 9 1 (n Multiple trauma 20 60 2 1 [%]) 47 60 5 1 21 (9.5%) 74 60 4 1 3 (7.9%) 9 62 19 1 (n Neuromuscular blocking agents 29 62 2 1 [%]) 47 62 4 1 7 (3.1%) 74 62 5 1 7 (18.4%) 9 64 12 1 (n Central venous line 22 64 2 1 [%]) 46 64 7 1 215 (98.0%) 74 64 6 1 38 (100%) 9 66 18 1 (n Intracranial pressure monitoring 29 66 2 1 [%]) 47 66 5 1 20 (9.1%) 74 66 4 1 1 (2.6%) 9 68 78 1 aAccording to clinical judgement. ARDS, acute respiratory distress syndrome; COPD, chronic obstructive pulmonary disease; VAP, ventilator- 9 69 12 1 associated pneumonia. 9 72 81 1 Among the 38 evaluations classified as positive for VAP, tra- accounted for 13.3% of all agents isolated. In 18.4% of eval- 9 74 81 1 cheostomy was present in ten (26.3%). Previous lung disease uations in the VAP group, no agent was recovered from the 9 75 81 1 was observed in six (15.7%) events. Ulcer prophylaxis was endotracheal aspirates (Table 6). 51 77 39 1 present in 100% of evaluations, with H2-receptor blockers in 9 78 81 1 22 (57.8%) and proton pump inhibitors in 16 (42.2%). Sepsis Clinical observations 9 80 69 1 was diagnosed in 14 (36.8%) evaluations. Considering the population as a whole, in 59 evaluations 9 81 39 1 (26.9%) patients had a tracheostomy. Stress ulcer prophylaxis 9 83 39 1 was present at 210 of the 219 evaluations (96%), with H2- 51 83 39 1 Other clinical characteristics are listed in Table 2. A total of 31 9 84 81 1 (29.2%) patients died during their hospitalization: 11 (33.3%) receptor blockers in 58.4%, proton pump inhibitors in 36.5% 9 86 81 1 of the 33 patients in the VAP group and 20 (27.3%) of the 73 and sucralfate in 0.9%. Sepsis was diagnosed in 46 (21%) 9 87 7 1 evaluations. 51 87 24 1 patients without VAP (not significant). 2 92 3 1 R425

55 5 35 1 Available online http://ccforum.com/content/8/6/R422 9 10 4 1 Table 3 9 12 22 1 Qualitative and quantitative analysis 9 15 5 1 Parameter 36 15 5 1 Qualitative 67 15 6 1 Quantitative 57 17 5 1 105 cfu/ml 77 17 5 1 106 cfu/ml 9 19 5 1 Sensitivity 38 19 2 1 81% 58 19 2 1 65% 79 19 2 1 26% 9 21 5 1 Specificity 38 21 2 1 23% 58 21 2 1 48% 79 21 2 1 78% 9 23 13 1 Positive predictive value 38 23 2 1 18% 58 23 2 1 21% 79 23 2 1 20% 9 25 13 1 Negative predictive value 38 25 2 1 86% 58 25 2 1 87% 79 25 2 1 83% 9 27 16 1 Likelihood ratio of positive test 38 27 2 1 1.05 58 27 2 1 1.25 79 27 2 1 1.18 9 29 17 1 Likelihood ratio of negative test 38 29 2 1 0.83 58 29 2 1 0.73 79 29 2 1 0.95 9 31 13 1 cfu, colony-forming units. 9 34 4 1 Table 4 9 36 48 1 Diagnostic criteria for ventilator-associated pneumonia in order of occurrence 9 38 59 1 n Diagnostic criteria 70 39 2 1 (%) 9 41 7 1 Leukocytosis 67 41 6 1 26 (68.4%) 9 43 3 1 Fever 67 43 6 1 24 (63.1%) 9 44 14 1 Purulent tracheal secretion 67 44 6 1 22 (57.8%) 9 46 24 1 Decrease of at least 10% in PaO2/FiO2 ratio 67 46 6 1 16 (42.1%) 9 48 29 1 Rales or dullness to percussion on chest examination 67 48 5 1 9 (23.6%) 9 50 6 1 Leucopenia 67 50 5 1 4 (10.5%) 9 52 12 1 Blood positive cultures 67 52 5 1 4 (10.5%) 9 54 7 1 Hypothermia 68 54 4 1 2 (5.2%) 9 56 34 1 FiO2, fractional inspired oxygen; PaO2, arterial oxygen tension. 9 59 9 1 Discussion 51 59 39 1 ate, and shortening the duration of treatment are presently 9 60 58 1 standard of care for VAP. VAP is the most frequent type of infection in ICU patients in 9 62 39 1 Europe and Latin America (almost half of all nosocomial infec- 9 63 81 1 In order to avoid any delay in instituting antibiotic treatment, tions) [3] and ranks second in US ICUs [11]. The attributable 9 65 81 1 reliable diagnostic methods should be employed. Despite their mortality is higher in medical than in surgical patients, and 9 66 81 1 variable sensitivity and specificity [16], clinical/radiological rates vary according to the case mix and aetiological agent 9 68 81 1 findings may currently be considered the best option, although [12]. 51 69 39 1 rapid tests, such as the percentage of infected leucocytes on 9 71 81 1 bronchial specimens, are promising in that they can provide Inadequate or delayed antimicrobial treatment in VAP is an 9 72 81 1 rapid confirmation [17]. Culture results for bronchial or tra- established independent predictor of death [13]. According to 9 74 81 1 cheal samples may be available late in the course of an epi- published data, changing an initial empirical treatment based 9 75 81 1 sode of VAP and should not be used to decide whether to on subsequent culture results may have either a beneficial 9 77 81 1 treat, especially in patients who are severely ill. On the other effect (in terms of mortality, less antibiotic use, less days on 9 78 81 1 hand, culture results should be used to adjust (narrow or antibiotics) [14] or no effect in more severely ill patients [15]. 9 80 81 1 extend antibiotic spectrum) or withdraw empirical treatment – For this reason, efforts must be directed at choosing adequate 9 81 81 1 a practice that has been shown to be beneficial, with no empirical treatment as early as possible, which may be accom- 9 83 81 1 increase in mortality, and that directs medical staff to seek plished with a high degree of suspicion and adequate guide- 9 84 68 1 other unsuspected foci of infection [18]. lines based on local antibacterial susceptibilities. In addition, 9 86 39 1 adhering to ideal pharmacological principles (choosing contin- 9 87 81 1 Although bronchoscopic samples increase the degree of con- uous as opposed to intermittent administration, adjustment for 9 89 81 1 fidence that a diagnosis of VAP is correct [14], endotracheal renal and hepatic failures), reducing dosages when appropri- 51 90 39 1 aspirates, despite their lack of consistency as a diagnostic tool 93 92 3 1 R426

9 5 8 1 Critical Care 19 5 10 1 December 2004 31 5 6 1 Vol 8 No 6 40 5 9 1 Camargo et al. 9 10 4 1 Table 5 9 12 33 1 Prescription of antimicrobials in all the events studied 9 15 63 1 (n Class of antimicrobial 65 15 6 1 Evaluations 73 15 2 1 [%]) 9 17 7 1 Glycopeptide 66 17 7 1 109 (49.7%) 9 19 6 1 Antifungical 67 19 6 1 93 (42.4%) 9 21 17 1 Third generation cephalosporin 67 21 6 1 86 (39.2%) 9 23 7 1 Carbapenem 67 23 6 1 75 (34.2%) 9 25 6 1 Clindamicin 67 25 6 1 34 (15.5%) 9 27 5 1 Quinolone 67 27 6 1 31 (14.1%) 9 29 7 1 Metronidazol 67 29 6 1 26 (11.8%) 9 31 18 1 Fourth generation cephalosporin 67 31 6 1 23 (10.5%) 9 32 6 1 Macrolides 67 32 6 1 23 (10.5%) 9 34 2 1 Total 66 34 7 1 212 (96.8%) 9 38 4 1 Table 6 9 41 58 1 Infectious agents isolated in the evaluations of patients with ventilator-associated pneumonia 9 43 63 1 (n Aetiological agents 65 43 6 1 Evaluations 73 43 2 1 [%]) 9 45 13 1 Gram-negative bacteria 67 45 6 1 15 (39.4%) 9 47 12 1 Gram-positive bacteria 67 47 6 1 11 (28.9%) 9 49 9 1 Negative cultures 67 49 5 1 7 (18.4%) 9 51 11 1 One or more agents 67 51 5 1 6 (15.7%) 9 53 4 1 Fungus 67 53 5 1 5 (13.3%) 9 55 8 1 Isolated agents 11 57 13 1 Staphylococcus aureus 67 57 5 1 6 (15.7%) 11 59 14 1 Pseudomonas aeruginosa 67 59 5 1 6 (15.7%) 11 61 13 1 Acinetobacter baumanii 68 61 4 1 3 (7.8%) 9 65 49 1 values (104 [19], are widely employed in the management of VAP. Recent 59 65 31 1 cfu/ml for bronchoalveolar lavage [BAL] fluid and 9 66 39 1 small trials have consistently shown that there is no advantage 51 66 39 1 103 cfu/ml for protected brush specimen [PBS]) for improving 9 68 39 1 of using bronchoscopic methods over relying on tracheal aspi- 51 68 39 1 diagnostic performance. On the other hand, use of these cut- 9 69 39 1 rate cultures when mortality is an end-point [6,20,21]. 51 69 39 1 off values has yielded conflicting results, and previous antibi- 9 71 39 1 Reduced costs and similar outcomes were reported using 51 71 39 1 otic treatment has great impact on these values. Souweine 9 72 39 1 either quantitative or qualitative tracheal aspirates for guiding 51 72 39 1 and coworkers [23] showed that the standard cutoff values of 9 74 39 1 or deciding to interrupt antibiotic treatment for VAP [6]. This 51 74 39 1 BAL and PSB would have to be lowered to 103 cfu/ml and 102 9 75 39 1 may be due to the high correlation between tracheal aspirates 51 75 39 1 cfu/ml to retain diagnostic accuracy where antibiotics were 9 77 39 1 (both quantitative and qualitative) and bronchoscopic cultures 51 77 39 1 previously administered, mainly when they are given in the pre- 9 78 39 1 when presence of VAP is highly probable [21,22]. However, 51 78 10 1 ceding 24 hours. 9 80 39 1 the above-mentioned studies did not determine the value of 9 81 39 1 quantification of micro-organisms in tracheal aspirate samples 51 81 39 1 Only a small number of studies have evaluated the role of 9 83 26 1 as compared with qualitative assessment. 51 83 39 1 quantitative endotracheal cultures in the diagnosis of VAP. 51 84 39 1 Albert and coworkers [24], studying 20 ventilated patients and 9 86 39 1 Quantification of micro-organisms in biological samples for the 51 86 39 1 using clinical/radiological parameters, found the threshold of 9 87 39 1 purpose of diagnosing infectious conditions is widely used, 51 87 39 1 105 cfu/ml to have a sensitivity of 81%, specificity of 65%, 9 89 39 1 particularly for nosocomial infections. Regarding respiratory 51 89 39 1 positive predictive value of 55% and negative predictive value 9 90 39 1 infections, bronchoscopic samples have established cutoff 51 90 39 1 of 55%. In that study different cutoff values were not tested to 2 92 3 1 R427

9 5 8 1 Critical Care 19 5 9 1 February 2005 30 5 6 1 Vol 9 No 1 39 5 10 1 Prinianakis et al. 9 10 4 1 Table 1 9 12 44 1 Baseline ventilator settings and patients' respiratory system mechanics 23 15 1 1 VT 50 15 1 1 Crs 64 15 1 1 Rint 9 15 69 1 Rrs No. 37 15 1 1 Fr 9 17 0 1 1 23 17 2 1 0.68 37 17 2 1 12.8 50 17 2 1 47.5 64 17 2 1 12.6 77 17 2 1 17.1 9 19 0 1 2 23 19 2 1 0.64 37 19 2 1 13.0 50 19 2 1 27.2 64 19 1 1 8.4 77 19 2 1 12.4 9 21 0 1 3 23 21 2 1 0.61 37 21 1 1 8.1 50 21 2 1 63.2 64 21 2 1 13.1 77 21 2 1 20.4 9 23 0 1 4 23 23 2 1 0.70 37 23 1 1 7.1 50 23 2 1 57.8 64 23 2 1 14.1 77 23 2 1 17.5 9 25 0 1 5 23 25 2 1 0.46 37 25 1 1 7.1 50 25 2 1 63.8 64 25 2 1 14.9 77 25 2 1 17.2 9 27 0 1 6 23 27 2 1 0.68 37 27 2 1 14.9 50 27 2 1 30.5 64 27 2 1 11.0 77 27 2 1 15.5 9 29 0 1 7 23 29 2 1 0.52 37 29 2 1 14.5 50 29 2 1 30.9 64 29 2 1 10.8 77 29 2 1 15.0 9 31 0 1 8 23 31 2 1 0.58 37 31 2 1 11.6 50 31 2 1 28.0 64 31 2 1 13.8 77 31 2 1 18.5 9 32 0 1 9 23 32 2 1 0.62 37 32 1 1 8.5 50 32 2 1 51.2 64 32 2 1 13.4 77 32 2 1 15.1 9 34 1 1 10 23 34 2 1 0.60 37 34 2 1 13.5 50 34 2 1 32.2 64 34 2 1 13.3 77 34 2 1 17.3 9 36 1 1 11 23 36 2 1 0.66 37 36 1 1 9.8 50 36 2 1 56.1 64 36 1 1 8.5 77 36 2 1 12.8 9 38 1 1 12 23 38 2 1 0.58 37 38 2 1 10.4 50 38 2 1 17.7 64 38 1 1 9.6 77 38 2 1 22.8 9 40 1 1 13 23 40 2 1 0.51 37 40 2 1 13.0 50 40 2 1 37.6 64 40 1 1 9.0 77 40 2 1 14.0 9 42 1 1 14 23 42 2 1 0.56 37 42 2 1 16.0 50 42 2 1 43.9 64 42 1 1 6.7 77 42 2 1 13.3 9 44 1 1 15 23 44 2 1 0.49 37 44 2 1 11.8 50 44 2 1 36.4 64 44 2 1 10.7 77 44 2 1 13.6 9 46 3 1 Mean 23 46 2 1 0.59 37 46 2 1 11.5 50 46 2 1 41.6 64 46 2 1 11.3 77 46 2 1 16.2 9 48 1 1 SD 23 48 2 1 0.07 37 48 1 1 2.9 50 48 2 1 14.4 64 48 1 1 2.5 77 48 1 1 2.9 9 50 2 1 Crs, 73 50 1 1 Rint 12 50 67 1 Rrs, end-inspiratory static compliance of the respiratory system (ml/cmH2O); Fr, ventilator frequency (breaths/min); 75 50 2 1 and 80 50 7 1 minimum and 9 51 39 1 VT, maximum inspiratory resistance (cmH2O/l per second), respectively; 49 51 10 1 tidal volume (litres). 9 55 4 1 Table 2 9 57 14 1 Model study: protocol A 9 59 5 1 Parameter 21 59 8 1 Normal pattern 44 59 10 1 Restrictive pattern 68 59 10 1 Obstructive pattern 21 61 1 1 V' 28 61 1 1 V' 36 61 1 1 V' 44 61 1 1 V' 52 61 1 1 V' 60 61 1 1 V' 68 61 1 1 V' 75 61 1 1 V' 22 61 62 1 V' = 1 30 61 3 1 = 0.8 38 61 3 1 = 0.6 45 61 1 1 = 1 53 61 3 1 = 0.8 61 61 3 1 = 0.6 69 61 1 1 = 1 77 61 3 1 = 0.8 85 61 3 1 = 0.6 9 63 5 1 Large area 16 65 2 1 (ml) 21 65 4 1 191 ± 7 28 65 4 1 196 ± 6 36 65 4 1 190 ± 4 44 65 5 1 190 ± 13 52 65 5 1 190 ± 15 60 65 4 1 190 ± 6 68 65 4 1 196 ± 5 75 65 4 1 185 ± 6 11 65 76 1 187 ± 6 Leakpause 11 67 6 1 Leakconv (ml) 21 67 4 1 298 ± 6 28 67 5 1 315 ± 3a 36 67 5 1 339 ± 4ab 44 67 4 1 303 ± 6 52 67 5 1 330 ± 2a 60 67 5 1 358 ± 2ab 68 67 4 1 308 ± 7 75 67 4 1 309 ± 5 83 67 6 1 320 ± 10ab 9 69 5 1 Small area 16 71 2 1 (ml) 21 71 4 1 146 ± 2 28 71 4 1 135 ± 5 36 71 4 1 135 ± 4 44 71 4 1 147 ± 8 52 71 5 1 148 ± 12 60 71 4 1 137 ± 4 68 71 4 1 146 ± 9 75 71 4 1 139 ± 6 11 71 77 1 141 ± 11 Leakpause 11 73 6 1 Leakconv (ml) 21 73 4 1 239 ± 7 28 73 4 1 228 ± 3 36 73 5 1 244 ± 4ab 44 73 5 1 249 ± 10 52 73 4 1 243 ± 4 60 73 5 1 269 ± 7ab 68 73 5 1 243 ± 14 75 73 4 1 234 ± 4 83 73 5 1 254 ± 6ab 9 75 15 1 V', Results are means ± SD. 25 75 63 1 constant inspiratory flow (litre/s); Leakconv, cuff-leak volume measured when the cuff remained deflated during both 9 76 71 1 inspiration and expiration; Leakpause, cuff-leak volume measured when the cuff was deflated at the end of 3 s of inspiratory pause. 9 77 31 1 V'I aSignificantly different from the corresponding value at 41 77 5 1 = 1 litre/s. 9 78 31 1 V'I bSignificantly different from the corresponding value at 41 78 6 1 = 0.8 litre/s. 9 80 49 1 Protocol B increasing the size of the cross-sectional area around the 9 82 33 1 ∆Leak V'I endotracheal tube (Fig. 2). The effect of 36 82 1 1 on 43 82 5 1 was sig- 51 82 39 1 Similarly to protocol A, and independently of model mechan- 9 83 39 1 nificantly higher with simulated restrictive respiratory system 51 83 39 1 ics, Leakconv was significantly higher than Leakpause (Table 3). 9 85 51 1 R, disease and large cross-sectional area around the endotra- 51 85 7 1 For a given 60 85 30 1 Leakconv increased significantly with decreasing 51 86 1 1 C, 9 86 77 1 C, cheal tube (Fig. 2). 53 86 31 1 whereas Leakpause remained constant. For a given 87 86 3 1 Leak- 54 88 36 1 and Leakconv tended to increase slightly with the highest 51 88 2 0 pause 51 89 39 1 resistance, the difference being significant only for Leakpause. 2 92 2 1 R27

9 5 8 1 Critical Care 19 5 9 1 February 2005 30 5 6 1 Vol 9 No 1 39 5 10 1 Prinianakis et al. 9 10 4 1 Table 3 9 12 14 1 Model study: protocol B 21 15 0 1 R 44 15 0 1 R 9 15 59 1 R Parameter 22 15 1 1 = 8 45 15 2 1 = 16 69 15 2 1 = 32 21 16 1 1 C 28 16 1 1 C 36 16 1 1 C 44 16 1 1 C 52 16 1 1 C 60 16 1 1 C 68 16 1 1 C 75 16 1 1 C 22 16 62 1 C = 20 30 16 2 1 = 50 38 16 3 1 = 100 45 16 2 1 = 20 53 16 2 1 = 50 61 16 3 1 = 100 69 16 2 1 = 20 77 16 2 1 = 50 85 16 3 1 = 100 9 19 7 1 Leakpause (ml) 21 19 3 1 96 ± 9 28 19 3 1 99 ± 6 36 19 3 1 96 ± 9 44 19 5 1 105 ± 10 52 19 5 1 103 ± 11 60 19 4 1 110 ± 8 68 19 5 1 123 ± 12c 75 19 4 1 115 ± 9 83 19 5 1 118 ± 12c 9 21 6 1 Leakconv (ml) 21 21 5 1 275 ± 11a 28 21 4 1 257 ± 9 36 21 4 1 245 ± 8 44 21 5 1 278 ± 6ab 52 21 5 1 261 ± 10 60 21 4 1 253 ± 9 68 21 6 1 287 ± 13ab 75 21 4 1 268 ± 7 83 21 4 1 255 ± 6 9 23 15 1 C, Results are means ± SD. 25 23 62 1 model compliance (ml/cmH2O); Leakconv, cuff-leak volume measured when the cuff remained deflated during both 9 24 72 1 R, inspiration and expiration; Leakpause, cuff-leak volume measured when the cuff was deflated at the end of 3 s of inspiratory pause; 83 24 3 1 model 9 25 20 1 resistance (cmH2O/litre per second). 9 26 31 1 C aSignificantly different from the corresponding value at 41 26 10 1 = 100 ml/cmH2O. 9 27 31 1 C bSignificantly different from the corresponding value at 41 27 9 1 = 50 ml/cmH2O. 9 28 31 1 R cSignificantly different from the corresponding value at 41 28 16 1 = 8 cmH2O/litre per second. 9 34 81 1 ence in expired volume with and without a deflated cuff should Figure 3 51 36 39 1 be entirely due to gas leak around the endotracheal tube 51 37 39 1 during expiration (pause cuff leak). In contrast, when the cuff- 51 39 39 1 leak volume was measured with the conventional method, a 51 40 39 1 fraction of gas volume delivered by the ventilator might leak 51 42 39 1 around the endotracheal tube during inspiration. In that case 51 43 39 1 the measured cuff-leak volume is the total leak consisting of an 51 45 39 1 inspiratory and expiratory component. The design of this study 51 46 39 1 did not permit us to measure with accuracy the inspiratory 51 48 39 1 leak. This is because pause cuff leak is not similar to expiratory 51 49 39 1 leak obtained with the conventional method because end- 51 51 39 1 inspiratory lung volume and thus elastic recoil pressure at the 51 52 39 1 beginning of expiration differ substantially between the two 51 54 39 1 methods of cuff leak determination. The pause cuff leak should 51 55 39 1 be higher than the expiratory component of the total leak, 51 57 39 1 because end inspiratory lung volume and elastic recoil pres- 51 58 39 1 sure were considerably higher when pause cuff leak was 51 60 5 1 obtained. 51 63 39 1 Both in clinical and model study the cuff-leak volume deter- 51 64 39 1 mined with the conventional method (Leakconv) was always 51 66 39 1 higher than that obtained by cuff deflation at end-inspiratory 51 67 39 1 pause, which eliminated the inspiratory component of total 51 69 39 1 leak (Leakpause). It follows that the inspiratory component is an 51 70 39 1 important determinant of the cuff-leak test. It is of interest to 51 72 39 1 note that in patients Leakconv was about threefold Leakpause 51 73 24 1 whatever the amount of the total leak. 51 76 39 1 In Protocol A of the lung model study, for a given cross-sec- 51 77 39 1 tional area, the system mechanics and inspiratory flow consid- 51 79 39 1 erably affected Leakconv; Leakconv increased significantly with 25 80 3 1 ∆Leak 9 80 38 1 and Lung Lung model model study, study, protocol protocol II. II 29 80 15 1 (difference between Leakconv 51 80 39 1 decreasing compliance and inspiratory flow. In contrast, nei- 9 81 39 1 Leakpause) is shown at constant inspiratory flow as a function of respira- 51 82 39 1 ther system compliance nor inspiratory flow influenced Leak- 9 83 39 1 tory system mechanics in a simulated model of constant cross-sectional 51 83 39 1 ∆Leak pause, which remained relatively constant. As a result 9 84 21 1 R, area around the endotracheal tube. 30 84 13 1 model airway resistance 9 85 81 1 increased significantly with decreasing compliance and inspir- C, (cmH2O/litre per second); 25 85 22 1 model compliance (ml/cmH2O). *, Signifi- 9 86 27 1 C cantly different from the corresponding value at 36 86 11 1 = 100 ml/cmH2O. +, 51 86 39 1 atory flow. The constancy of Leakpause suggested that the 9 88 31 1 C Significantly different from the corresponding value at 40 88 4 1 = 50 ml/ 51 88 39 1 expiratory component of the total leak was also unaffected by 9 89 4 1 cmH2O. 51 89 39 1 changes in system compliance and inspiratory flow. It follows 51 91 39 1 that respiratory system compliance and inspiratory flow have 2 92 2 1 R29

54 5 36 1 http://arthritis-research.com/content/7/2/R349 Available online 9 10 4 1 Table 1 9 12 66 1 Clinical and demographic characteristics of patients with rheumatoid arthritis (RA) and of healthy controls 47 15 1 1 (n 66 15 1 1 (n 40 15 42 1 (n Erosive RA 48 15 3 1 = 88) 57 15 8 1 Non-erosive RA 67 15 3 1 = 43) 77 15 4 1 Controls 83 15 3 1 = 34) 9 17 26 1 Age (years) [mean ± standard deviation (range)] 41 17 8 1 63 ± 2 (28–85) 59 17 8 1 53 ± 3 (19–83) 77 17 9 1 42 ± 1.8 (18–67) 9 19 9 1 Sex (male/female) 44 19 3 1 26/62 62 19 3 1 12/31 80 19 3 1 12/22 9 21 17 1 Duration of the disease (years) 43 21 6 1 12.7 ± 1.2 62 21 4 1 8 ± 1.4 81 21 0 1 - 9 23 13 1 Rheumatoid factor (+/-) 45 23 2 1 80/8 62 23 3 1 10/33 81 23 1 1 n.a. 9 25 13 1 Treatment with DMARDs 11 27 8 1 (n Methotrexate 20 27 3 1 = 25) 45 27 1 1 18 63 27 0 1 7 81 27 0 1 - 11 29 10 1 (n Other DMARDs 22 29 3 1 = 13) 46 29 0 1 9 63 29 0 1 4 11 30 3 1 TNF-α 15 30 6 1 (n blockade 22 31 3 1 = 47) 44 31 4 1 42 (37*) 62 31 3 1 5 (5*) 11 32 4 1 (n None 16 32 3 1 = 45) 45 32 1 1 18 63 32 1 1 27 9 34 41 1 TNF-α, n.a., not assesed; DMARD, disease modifying anti-rheumatic drug; 51 35 35 1 tumour necrosis factor alpha. *In combination with methotrexate. 9 37 4 1 Table 2 9 40 65 1 Clinical comparison of patients with rheumatoid arthritis (RA) expressing high* and low levels of survivin 45 42 0 1 P 27 42 44 1 P Survivin high, erosive RA 53 42 11 1 Survivin low, erosive RA 78 42 11 1 Survivin low, non-erosive 31 43 0 1 (n 56 43 0 1 (n 32 43 50 1 (n = 27) 58 43 2 1 = 61) 81 43 1 1 RA 84 43 3 1 = 42)† 9 45 10 1 Survivin levels (pg/ml) 11 47 2 1 Blood 30 47 6 1 1180 ± 309 44 47 4 1 <0.0001 57 47 3 1 97 ± 9 70 47 2 1 0.013 82 47 4 1 127 ± 5 11 49 6 1 Synovial fluid 30 49 6 1 1039 ± 523 44 49 2 1 0.016 57 49 4 1 132 ± 4 70 49 1 1 n.s. 82 49 4 1 124 ± 2 9 51 11 1 Disease duration (years) 30 51 5 1 15.5 ± 2.4 45 51 1 1 n.s. 56 51 5 1 13.6 ± 1.2 69 51 3 1 0.0002 81 51 4 1 8.3 ± 1.4 9 52 5 1 Age (years) 31 52 3 1 58 ± 3 45 52 1 1 n.s. 57 52 3 1 60 ± 2 70 52 2 1 0.05 82 52 3 1 53 ± 3 9 54 14 1 (n) Rheumatoid factor-positive 32 54 1 1 25 45 54 1 1 n.s. 58 54 1 1 53 69 54 4 1 <0.0001 83 54 1 1 10 9 56 11 1 C-reactive protein (mg/l) 31 56 3 1 45 ± 9 44 56 2 1 0.035 57 56 3 1 29 ± 5 70 56 1 1 n.s. 82 56 3 1 39 ± 7 9 58 16 1 White blood cell count (× 109/ml) 11 60 2 1 Blood 31 60 4 1 8.7 ± 0.6 44 60 2 1 0.038 56 60 4 1 7.2 ± 0.3 70 60 1 1 n.s. 81 60 4 1 7.1 ± 0.3 11 62 6 1 Synovial fluid 30 62 5 1 10.8 ± 1.9 45 62 1 1 n.s. 56 62 5 1 11.2 ± 2.9 70 62 1 1 n.s. 81 62 5 1 13.1 ± 2.8 9 64 78 1 Continuous parameters are presented as the mean ± standard error of the mean. n.s., not significant. *Level of survivin above 300 pg/ml was 9 65 39 1 †One patient having a high survivin level is excluded. considered 'high'. 9 68 39 1 significant correlation to the serum levels of C-reactive pro- 51 68 39 1 patients having high and low levels of survivin (Table 2) 9 69 39 1 tein and WBC count, and neither to the synovial fluid leu- 51 69 39 1 revealed, beside erosivity, an association between high lev- 9 71 19 1 kocyte count and IL-6 levels. 51 71 39 1 els of survivin and increased circulating C-reactive protein 51 72 39 1 as well as elevated WBC counts. In contrast, age, gender, 9 73 39 1 The RA patients were further stratified as having 'high' 51 73 39 1 RF-positivity, and duration of the disease were similar in the 9 75 39 1 (>300 pg/ml) or 'low' (<300 pg/ml) levels of survivin, 51 75 39 1 ERA patients with high levels of survivin as compared with 9 76 39 1 departing from the level of survivin that corresponded to a 51 76 14 1 those with low levels. 9 78 39 1 mean + three standard deviations of the control group as a 9 79 39 1 cut-off. The difference in the mean survivin level between 51 79 39 1 The level of survivin was also studied in RA synovial fluid 9 81 39 1 the 'high' and the 'low' groups was about 10-fold (1180 ± 51 81 39 1 samples separated with respect to the cell pellet and the 9 82 63 1 (n 309 pg/ml versus 97 ± 9 pg/ml). High levels of survivin 51 82 19 1 supernatant by centrifugation 73 82 17 1 = 9). Survivin levels found 9 84 39 1 were detected in 28 of 131 patients (21%). All but one 51 84 39 1 in supernatants and in the lysates of synovial fluid cells 9 85 39 1 (96%) of the patients with a high survivin level displayed 51 85 39 1 obtained from the same sample revealed a strong correla- 54 87 0 1 (r 9 87 53 1 P erosive RA. A dominance of a high survivin level among the 51 87 2 1 tion 56 87 5 1 = 0.87, 63 87 27 1 < 0.0001). These data indicate that sur- 9 88 39 1 ERA patients was consequently found both in plasma and 51 88 39 1 vivin is produced and secreted locally in the joints of RA 9 90 39 1 in synovial fluid samples. Comparison between the ERA 51 90 5 1 patients. 93 92 3 1 R353

8 5 15 1 BMC Cancer 2005, 5:30 60 5 30 1 http://www.biomedcentral.com/1471-2407/5/30 8 10 39 1 supportive care was applied consisting of antiemetic med- 51 10 27 1 Table 1: Characteristics of Patients (N = 180) 8 12 39 1 ication with 5HT3 antagonist on the days of chemo, anti- 51 13 12 1 Median age (range), yrs 77 13 5 1 64 (60–73) 8 13 39 1 biotics in case of known bacterial infection associated 51 14 10 1 Age category, n (%) 8 15 39 1 with granulocyte colony-stimulating factor (G-CSF) in 55 15 10 1 >60 and < 65 years 78 15 4 1 100 (56) 8 16 39 1 case of febrile neutropenia; G-CSF was also sporadically 55 16 5 1 ≥ 65 years 78 16 3 1 80 (44) 8 18 39 1 used in case of grade 4 neutropenia without fever, and 8 19 39 1 prophylactically in cycles following febrile or grade 4 51 19 10 1 pT-category, n (%) 55 20 2 1 pT1 8 20 73 1 73 (41) neutropenia. 55 22 2 1 pT2 78 22 3 1 91 (51) 55 23 2 1 pT3 79 23 2 1 5 (3) 8 24 39 1 Pre-chemotherapy evaluation included medical history, 55 24 2 1 pT4 79 24 2 1 3 (2) 8 25 39 1 physical examination, haematology, serum biochemistry 55 26 5 1 unknown 79 26 2 1 8 (4) 8 27 39 1 tests, and tumor staging with chest radiography, abdomi- 8 28 39 1 nal ultrasonography, nuclidic bone scan with X-ray of 51 28 10 1 pN-category, n (%) 8 30 39 1 bone hot-spots, ECG and, in selected cases, an evaluation 55 30 2 1 pN0 78 30 3 1 66 (37) 8 31 39 1 of left ventricular ejection fraction. Toxicity was usually 55 31 2 1 pN1 78 31 3 1 97 (54) 55 32 2 1 pN2 8 32 73 1 10 (6) assessed with blood counts preceding each administration 55 34 5 1 unknown 8 34 73 1 7 (4) of chemotherapy, and complete serology before day 1 8 35 39 1 administration of chemotherapy, unless clinical need for 51 36 12 1 Receptor status, n (%) 8 37 39 1 more frequent control. Reviewing the charts, we codified 55 38 10 1 ER or PgR positive 78 38 3 1 82 (46) 8 38 39 1 toxicity according to National Cancer Institute Common 55 39 4 1 Negative 78 39 3 1 78 (43) 8 40 39 1 Toxicity Criteria (version 2.0, 1998). Toxicity is reported 55 40 5 1 Unknown 78 40 3 1 20 (11) 8 41 39 1 with details of type and grade, including all grades. Any- 8 43 39 1 way, for descriptive purposes we also summarised toxicity 51 43 8 1 Histotype, n (%) 8 44 39 1 into three major groups: haematological (leukopenia, 55 44 3 1 Ductal 78 44 4 1 132 (73) 55 46 4 1 Lobular 8 46 73 1 33 (18) neutropenia, anemia, thrombocytopenia, febrile neutro- 55 47 3 1 Other 8 47 73 1 15 (8) penia, bleeding, infection), other (including all the other 8 49 39 1 toxicities) and any (including both haematological and 51 50 12 1 Histologic grade, n (%) 8 50 39 1 other) and we summed up grades 0 to 2 as non severe and 55 51 1 1 G1 79 51 2 1 5 (3) 8 52 15 1 grades 3 or 4 as severe. 55 52 1 1 G2 78 52 3 1 53 (29) 55 54 1 1 G3 78 54 3 1 96 (53) 8 55 39 1 To analyse compliance, the distribution of several indices 55 55 5 1 Unknown 78 55 3 1 26 (14) 8 56 39 1 was compared between the two age subgroups. A cycle 8 58 39 1 was considered as delivered if day 1 chemotherapy had 51 58 18 1 Concomitant radiotherapy, n (%) 55 59 1 1 No 8 59 74 1 114 (63) been administered. Indices used for compliance descrip- 55 60 1 1 Yes 8 60 73 1 66 (37) tion were: a) the number of delivered cycles, b) the occur- 8 62 39 1 rence of withdrawal of chemotherapy on day 8, c) the 8 64 39 1 occurrence and the percentage of dose-reductions, d) the 8 65 39 1 occurrence of treatment delay that was expressed as rela- 8 67 39 1 tive duration by dividing actual and planned duration (1 8 68 39 1 means no delay, higher values mean progressively longer 8 70 39 1 delays), e) the need for G-CSF treatment according to pre- 51 70 39 1 Disease-free survival was defined as the time elapsed from 8 71 39 1 viously reported guidelines, and f) the cause of treatment 51 71 39 1 surgery to the date of assessment of local or distant or con- 8 73 39 1 discontinuation if earlier than planned. After description 51 73 39 1 tralateral breast cancer or the date of death for patients 8 74 39 1 of single indices, a binary index was built, weak compli- 51 74 39 1 dying without disease recurrence. The Kaplan-Meier 8 76 39 1 ance being defined as the occurrence of at least one of the 51 76 38 1 method was applied to draw disease-free survival curves. 8 77 39 1 following events: less than 6 cycles administered, with- 8 78 39 1 drawal of more than one day-8 treatment, dose reduction 51 78 39 1 Statistical significance of associations among variables 8 80 39 1 in more than 25% of cycles, relative duration higher than 51 80 39 1 was tested by the Fisher's exact test, or Wilcoxon-Mann 8 81 39 1 1.25, need of G-CSF treatment, treatment interruption 51 81 39 1 Whitney test for naturally ordered variables (i.e. number 8 83 39 1 because of patient's refusal or toxicity. Because of possible 51 83 39 1 of cycles, no. of day 8 omitted, grades of toxicity). Disease- 8 84 39 1 confounding effect of radiotherapy associated to CMF, 51 84 39 1 free survival curves were compared by the Log-rank test. 8 86 39 1 both toxicity and compliance were analysed not only by 51 86 39 1 Analysis were done with S-PLUS (6.0 Professional Release 8 87 39 1 age subgroups but also according to presence or absence 51 87 39 1 1, Insightful Corporation) and StatXact-5 (release 5.0.3, 8 89 20 1 of concomitant radiotherapy. 51 89 19 1 Cytel software Corporation). 82 95 8 1 Page 3 of 10 70 97 20 0 (page number not for citation purposes)

8 5 15 1 BMC Cancer 2005, 5:30 60 5 30 1 http://www.biomedcentral.com/1471-2407/5/30 8 10 37 1 Table 2: Worst degree of toxicity by age group (% of patients) 9 12 56 1 ≥ Type of toxicity 36 13 15 1 >60 and <65 years (n = 100) 66 13 9 1 65 years (n = 80) 86 13 1 1 P* 34 15 0 1 0 40 15 0 1 1 45 15 0 1 2 50 15 0 1 3 54 15 0 1 4 59 15 0 1 0 65 15 0 1 1 71 15 0 1 2 76 15 0 1 3 81 15 0 1 4 9 18 4 1 Anemia 34 18 1 1 81 39 18 1 1 10 45 18 0 1 8 50 18 0 1 0 54 18 0 1 1 59 18 1 1 86 65 18 0 1 6 71 18 0 1 4 76 18 0 1 4 81 18 0 1 0 86 18 2 1 0.39 9 19 6 1 Leukopenia 34 19 1 1 54 39 19 1 1 15 44 19 1 1 19 50 19 0 1 7 54 19 0 1 5 59 19 1 1 45 65 19 1 1 11 70 19 1 1 26 76 19 1 1 11 81 19 0 1 6 86 19 2 1 0.13 9 21 6 1 Neutropenia 34 21 1 1 73 39 21 1 1 10 45 21 0 1 8 50 21 0 1 5 54 21 0 1 4 59 21 1 1 68 65 21 0 1 9 71 21 0 1 7 76 21 1 1 10 81 21 0 1 6 86 21 2 1 0.32 9 22 11 1 Febrile Neutropenia 33 22 2 1 100 40 22 0 1 - 45 22 0 1 - 50 22 0 1 0 54 22 0 1 0 59 22 1 1 96 65 22 0 1 - 71 22 0 1 - 76 22 0 1 4 81 22 0 1 0 86 22 2 1 0.05 9 23 4 1 Infection 34 23 1 1 97 40 23 0 1 0 45 23 0 1 1 50 23 0 1 2 54 23 0 1 0 59 23 1 1 95 65 23 0 1 0 70 23 1 1 2.5 76 23 1 1 2.5 81 23 0 1 0 86 23 2 1 0.50 9 25 3 1 Fever 34 25 1 1 98 40 25 0 1 0 45 25 0 1 2 50 25 0 1 0 54 25 0 1 0 59 25 1 1 98 65 25 0 1 2 71 25 0 1 0 76 25 0 1 0 81 25 0 1 0 86 25 2 1 0.83 9 26 10 1 Thrombocytopenia 34 26 1 1 90 40 26 0 1 7 45 26 0 1 2 50 26 0 1 0 54 26 0 1 1 59 26 1 1 91 65 26 0 1 4 71 26 0 1 1 76 26 0 1 3 81 26 0 1 1 86 26 2 1 0.83 9 27 4 1 Bleeding 34 27 1 1 98 40 27 0 1 1 45 27 0 1 0 50 27 0 1 1 54 27 0 1 0 59 27 2 1 100 65 27 0 1 0 71 27 0 1 0 76 27 0 1 0 81 27 0 1 0 86 27 2 1 0.21 9 28 4 1 Nausea 34 28 1 1 81 40 28 0 1 8 45 28 0 1 6 50 28 0 1 5 54 28 0 1 0 59 28 1 1 69 65 28 1 1 11 70 28 1 1 12 76 28 0 1 8 81 28 0 1 0 86 28 2 1 0.06 9 30 4 1 Vomiting 34 30 1 1 71 40 30 0 1 5 44 30 1 1 11 50 30 1 1 13 54 30 0 1 0 59 30 1 1 69 65 30 0 1 9 70 30 1 1 16 76 30 0 1 6 81 30 0 1 0 86 30 2 1 0.98 9 31 5 1 Mucositis 34 31 1 1 84 40 31 0 1 5 45 31 0 1 8 50 31 0 1 3 54 31 0 1 0 59 31 1 1 92 65 31 0 1 1 71 31 0 1 5 76 31 0 1 1 81 31 0 1 0 86 31 2 1 0.09 9 32 2 1 Skin 34 32 1 1 99 40 32 0 1 1 45 32 0 1 0 50 32 0 1 0 54 32 0 1 0 59 32 1 1 98 65 32 0 1 0 71 32 0 1 2 76 32 0 1 0 81 32 0 1 0 86 32 2 1 0.43 9 34 4 1 Diarrhea 34 34 1 1 96 40 34 0 1 1 45 34 0 1 0 50 34 0 1 3 54 34 0 1 0 59 34 1 1 96 65 34 0 1 3 71 34 0 1 1 76 34 0 1 0 81 34 0 1 0 86 34 2 1 0.91 9 35 7 1 Constipation 34 35 1 1 99 40 35 0 1 0 45 35 0 1 0 50 35 0 1 1 54 35 0 1 0 59 35 1 1 98 65 35 0 1 1 71 35 0 1 1 76 35 0 1 0 81 35 0 1 0 86 35 2 1 0.45 9 36 4 1 Hepatic 34 36 1 1 88 40 36 0 1 3 45 36 0 1 7 50 36 0 1 2 54 36 0 1 0 59 36 1 1 90 65 36 0 1 6 71 36 0 1 4 76 36 0 1 0 81 36 0 1 0 86 36 2 1 0.60 9 38 4 1 Cardiac 33 38 2 1 100 40 38 0 1 0 45 38 0 1 0 50 38 0 1 0 54 38 0 1 0 59 38 1 1 96 65 38 0 1 0 70 38 1 1 2.5 76 38 0 1 1 81 38 0 1 0 86 38 2 1 0.09 9 39 3 1 Fatigue 34 39 1 1 96 40 39 0 1 1 45 39 0 1 3 50 39 0 1 0 54 39 0 1 0 59 39 1 1 95 65 39 0 1 0 70 39 1 1 2.5 76 39 1 1 2.5 81 39 0 1 0 86 39 2 1 0.72 9 40 5 1 Pulmonary 33 40 2 1 100 40 40 0 1 0 45 40 0 1 0 50 40 0 1 0 54 40 0 1 0 59 40 1 1 99 65 40 0 1 1 71 40 0 1 0 76 40 0 1 0 81 40 0 1 0 86 40 2 1 0.27 9 41 8 1 Abdominal pain 34 41 1 1 99 40 41 0 1 0 45 41 0 1 0 50 41 0 1 1 54 41 0 1 0 59 41 1 1 99 65 41 0 1 0 71 41 0 1 0 76 41 0 1 1 81 41 0 1 0 86 41 2 1 0.88 9 43 5 1 Chest pain 34 43 1 1 99 40 43 0 1 0 45 43 0 1 0 50 43 0 1 1 54 43 0 1 0 59 43 2 1 100 65 43 0 1 0 71 43 0 1 0 76 43 0 1 0 81 43 0 1 0 86 43 2 1 0.37 9 44 5 1 Heartburn 34 44 1 1 99 40 44 0 1 0 45 44 0 1 1 50 44 0 1 0 54 44 0 1 0 59 44 1 1 98 65 44 0 1 0 71 44 0 1 2 76 44 0 1 0 81 44 0 1 0 86 44 2 1 0.44 9 45 5 1 Headache 34 45 1 1 99 40 45 0 1 0 45 45 0 1 0 50 45 0 1 1 54 45 0 1 0 59 45 2 1 100 65 45 0 1 0 71 45 0 1 0 76 45 0 1 0 81 45 0 1 0 86 45 2 1 0.38 9 47 5 1 Confusion 34 47 1 1 99 40 47 0 1 1 45 47 0 1 0 50 47 0 1 0 54 47 0 1 0 59 47 2 1 100 65 47 0 1 0 71 47 0 1 0 76 47 0 1 0 81 47 0 1 0 86 47 2 1 0.38 9 48 6 1 Depression 33 48 2 1 100 40 48 0 1 0 45 48 0 1 0 50 48 0 1 0 54 48 0 1 0 59 48 1 1 99 65 48 0 1 0 71 48 0 1 0 76 48 0 1 1 81 48 0 1 0 86 48 2 1 0.27 9 49 9 1 Thrombophlebitis 34 49 1 1 97 40 49 0 1 0 45 49 0 1 0 50 49 0 1 3 54 49 0 1 0 59 49 1 1 94 65 49 0 1 0 71 49 0 1 0 76 49 0 1 6 81 49 0 1 0 86 49 2 1 0.30 9 50 4 1 Dysuria 34 50 1 1 99 40 50 0 1 0 45 50 0 1 0 50 50 0 1 1 54 50 0 1 0 59 50 1 1 99 65 50 0 1 0 71 50 0 1 0 76 50 0 1 1 81 50 0 1 0 86 50 2 1 0.88 9 52 3 1 Allergy 34 52 1 1 99 40 52 0 1 0 45 52 0 1 0 50 52 0 1 1 54 52 0 1 0 59 52 2 1 100 65 52 0 1 0 71 52 0 1 0 76 52 0 1 0 81 52 0 1 0 86 52 2 1 0.37 9 53 6 1 Weight gain 33 53 2 1 100 40 53 0 1 0 45 53 0 1 0 50 53 0 1 0 54 53 0 1 0 59 53 1 1 99 65 53 0 1 0 71 53 0 1 1 76 53 0 1 0 81 53 0 1 0 86 53 2 1 0.27 9 54 6 1 Weight loss 34 54 1 1 99 40 54 0 1 0 45 54 0 1 1 50 54 0 1 0 54 54 0 1 0 59 54 1 1 99 65 54 0 1 1 71 54 0 1 0 76 54 0 1 0 81 54 0 1 0 86 54 2 1 0.89 9 57 17 1 * Wilcoxon-Mann Whitney test 8 63 5 1 Results 51 63 39 1 tropenia, nausea, cardiac toxicity and thrombophlebitis 8 64 39 1 From March 1991 to March 2002, 180 patients were iden- 51 64 39 1 tended to be more frequent or severe among elderlies, 8 66 39 1 tified, 80 aged 65 or older, and 100 older than 60 and 51 66 39 1 while mucositis tended to be more evident among 8 67 39 1 younger than 65. The baseline characteristics of patients 51 67 39 1 younger patients. As shown in table 3, haematologic tox- 8 69 39 1 are reported in table 1. Overall, 87 patients received radi- 51 69 39 1 icity was adversely affected by concomitant radiotherapy, 8 70 39 1 otherapy on the residual breast; among these, 21 patients, 51 70 39 1 leukopenia, neutropenia, febrile neutropenia and fever 8 72 39 1 who received radiation therapy after the end of CMF, have 51 72 39 1 being significantly worse among patients who received 8 73 39 1 been considered in the no radiotherapy subgroup for all 51 73 39 1 concomitant radiotherapy. As reported in table 4, inci- 8 75 39 1 subsequent analyses. Sixty-six patients receiving concom- 51 75 39 1 dence of grade 3–4 toxicity (either haematological or 8 76 39 1 itant radiotherapy were equally distributed in the two age 51 76 39 1 other or any type) did not vary according to age, while 8 78 39 1 subgroups, 36 (36%) being younger than 65 and 30 51 78 39 1 severe haematological toxicity was more frequent among 8 79 13 1 (38%) being older. 51 79 39 1 patients who received concomitant radiotherapy (35% vs 51 81 39 1 8%, p < 0.0001); almost one half (47%) of the older 8 82 39 1 Treatment toxicity scattered by age subgroups is reported 51 82 39 1 patients receiving concomitant radiotherapy experienced 8 84 39 1 in table 2. There was no toxic death. The only type of tox- 51 84 22 1 grade 3–4 hematological toxicity. 8 85 39 1 icity that significantly differed between the two age groups 8 86 39 1 was febrile neutropenia that occurred in 3 cases among 51 86 39 1 Focusing on compliance, we found that CMF scheduling 8 88 39 1 older patients and never in the younger group (p = 0.05). 51 88 39 1 was modified (from days 1 and 8 every 4 weeks to day 1 8 89 39 1 Among toxicities that did not show a statistically signifi- 51 89 39 1 every 3 weeks) because of toxicity in 7 patients (2 in the 8 91 39 1 cant difference between the two groups, leukopenia, neu- 51 91 39 1 older and 5 in the younger group, p = 0.46); none of these 82 95 8 1 Page 4 of 10 70 97 20 0 (page number not for citation purposes)

8 5 15 1 BMC Cancer 2005, 5:30 60 5 30 1 http://www.biomedcentral.com/1471-2407/5/30 8 10 48 1 Table 3: Worst degree of hematological toxicity by radiotherapy (% of patients) 9 13 8 1 Type of toxicity 38 13 7 1 No (n = 114) 63 13 6 1 Yes (n = 66) 84 13 1 1 P* 33 15 0 1 0 38 15 0 1 1 43 15 0 1 2 47 15 0 1 3 51 15 0 1 4 55 15 0 1 0 61 15 0 1 1 66 15 0 1 2 71 15 0 1 3 77 15 0 1 4 9 18 4 1 Anemia 32 18 1 1 87 38 18 0 1 6 43 18 0 1 4 47 18 0 1 3 51 18 0 1 0 55 18 1 1 77 60 18 1 1 12 66 18 0 1 9 71 18 0 1 0 76 18 1 1 1.5 84 18 2 1 0.12 9 19 6 1 Leukopenia 32 19 1 1 58 38 19 1 1 15 43 19 1 1 22 47 19 0 1 3 51 19 0 1 3 55 19 1 1 36 60 19 1 1 11 66 19 1 1 23 71 19 1 1 20 76 19 1 1 11 83 19 3 1 0.0001 9 21 6 1 Neutropenia 32 21 1 1 82 38 21 0 1 8 43 21 0 1 6 47 21 0 1 2 51 21 0 1 2 55 21 1 1 50 60 21 1 1 12 66 21 1 1 11 71 21 1 1 17 76 21 1 1 11 83 21 4 1 <0.0001 9 22 11 1 Febrile Neutropenia 32 22 2 1 100 38 22 0 1 0 43 22 0 1 0 47 22 0 1 0 51 22 0 1 0 55 22 1 1 95 61 22 0 1 0 66 22 0 1 0 71 22 0 1 5 77 22 0 1 0 84 22 2 1 0.05 9 23 4 1 Infection 32 23 1 1 96 38 23 0 1 0 43 23 0 1 2 47 23 0 1 2 51 23 0 1 0 55 23 1 1 95 61 23 0 1 0 66 23 0 1 2 71 23 0 1 3 77 23 0 1 0 84 23 2 1 0.89 9 25 3 1 Fever 32 25 2 1 100 38 25 0 1 0 43 25 0 1 0 47 25 0 1 0 51 25 0 1 0 55 25 1 1 94 61 25 0 1 6 66 25 0 1 0 71 25 0 1 0 77 25 0 1 0 84 25 2 1 0.02 9 26 10 1 Thrombocytopenia 32 26 1 1 90 38 26 0 1 6 43 26 0 1 2 47 26 0 1 1 51 26 0 1 1 55 26 1 1 91 60 26 1 1 4.5 65 26 1 1 1.5 71 26 1 1 1.5 76 26 1 1 1.5 84 26 2 1 0.96 9 27 4 1 Bleeding 32 27 1 1 99 38 27 0 1 0 43 27 0 1 0 47 27 0 1 1 51 27 0 1 0 55 27 1 1 98 61 27 0 1 2 66 27 0 1 0 71 27 0 1 0 77 27 0 1 0 84 27 2 1 1.00 9 30 17 1 * Wilcoxon-Mann Whitney test 8 36 65 1 Table 4: Summary of toxicity by age, radiotherapy and combined subgroups expressed as row percentages) 39 38 8 1 Haematologic 61 38 3 1 Other 80 38 2 1 Any 9 41 6 1 Subgroup 35 41 3 1 G0-2 41 41 3 1 G3-4 50 41 1 1 p* 56 41 3 1 G0-2 62 41 3 1 G3-4 69 41 1 1 p* 74 41 3 1 G0-2 80 41 3 1 G3-4 87 41 1 1 p* 9 44 2 1 Age 49 44 2 1 0.17 68 44 2 1 0.49 86 44 2 1 1.00 9 45 11 1 >60 and <65 years (n 22 45 2 1 100) 35 45 2 1 86% 42 45 2 1 14% 56 45 2 1 72% 63 45 2 1 28% 74 45 2 1 62% 21 45 61 1 38% = 9 46 7 1 ≥ 65 years (n 18 46 1 1 80) 35 46 2 1 78% 42 46 2 1 22% 56 46 2 1 78% 63 46 2 1 22% 74 46 2 1 61% 17 46 66 1 39% = 9 49 8 1 Radiotherapy 48 49 4 1 <0.0001 68 49 2 1 0.21 86 49 2 1 0.27 9 50 3 1 No (n 14 50 2 1 114) 35 50 2 1 92% 42 50 1 1 8% 56 50 2 1 71% 63 50 2 1 29% 74 50 2 1 65% 13 50 70 1 35% = 9 52 3 1 Yes (n 14 52 1 1 66) 35 52 2 1 65% 42 52 2 1 35% 56 52 2 1 80% 63 52 2 1 20% 74 52 2 1 56% 13 52 69 1 44% = 9 55 13 1 Age and radiotherapy 48 55 4 1 <0.0001 68 55 2 1 0.42 86 55 2 1 0.08 9 56 12 1 >60 and < 65, no RT (n 23 56 1 1 64) 35 56 2 1 92% 42 56 1 1 8% 56 56 2 1 67% 63 56 2 1 33% 74 56 2 1 59% 22 56 60 1 41% = 9 57 8 1 ≥ 65, no RT (n 19 57 1 1 50) 35 57 2 1 92% 42 57 1 1 8% 56 57 2 1 76% 63 57 2 1 24% 74 57 2 1 72% 17 57 65 1 28% = 9 58 11 1 >60 and < 65, RT (n 22 58 1 1 36) 35 58 2 1 75% 42 58 2 1 25% 56 58 2 1 81% 63 58 2 1 19% 74 58 2 1 67% 20 58 62 1 33% = 9 60 6 1 ≥ 65, RT (n 17 60 1 1 30) 35 60 2 1 53% 42 60 2 1 47% 56 60 2 1 80% 63 60 2 1 20% 74 60 2 1 43% 16 60 67 1 57% = 9 63 16 1 G = grade; RT = radiotherapy. 8 70 39 1 patients did receive concomitant radiotherapy (p = 0.05). 51 70 39 1 about 10% each (p = 0.39). Eighty-eight percent of the 66 8 71 39 1 Distribution of compliance indices is summarized in table 51 71 39 1 patients who received concomitant radiotherapy were 8 73 39 1 5. Seventy-six percent of older patients received six cycles 51 73 39 1 able to receive 6 cycles of CMF and completed treatment 8 74 39 1 of chemotherapy as compared with 81% of those in the 51 74 39 1 according to the protocol, as compared with 74% of those 8 76 39 1 younger group. Day 8 of chemotherapy was withdrawn at 51 76 39 1 not irradiated concomitantly; concomitant irradiation 8 77 39 1 least once in 36% and 31%, and 18% and 14% of the 51 77 39 1 prolonged significantly duration of CMF in 5% of patients 8 79 39 1 patients had more than 25% of cycle at reduced doses, in 51 79 39 1 and produced a more frequent need of G-CSF. Overall, a 8 80 39 1 the older and younger groups, respectively. Treatment was 51 80 39 1 weak compliance (i.e. the occurrence of at least one of the 8 82 39 1 administered without significant delay (i.e. 24 weeks 51 82 39 1 following events: less than 6 cycles received, more than 1 8 83 39 1 without concomitant radiotherapy or 30 weeks with) in 51 83 39 1 day 8 omission, dose reduction in more than 25% of 8 85 39 1 almost all the patients. Haematopoietic support with G- 51 85 39 1 cycles, relative duration higher than 1.25, use of G-CSF or 8 86 39 1 CSF was required by 18% and 9% of older and younger 51 86 39 1 treatment interruption because of toxicity or refusal) was 8 88 39 1 patients. Causes of treatment discontinuation were simi- 51 88 39 1 more frequent among older patients (58% vs 46%, p = 8 89 39 1 lar in the two groups, protocol completion accounting for 51 89 39 1 0.02) and did not vary according to concomitant 8 91 39 1 more than three-fourths, and refusal and toxicity for 51 91 9 1 radiotherapy. 82 95 8 1 Page 5 of 10 70 97 20 0 (page number not for citation purposes)

8 5 15 1 BMC Cancer 2005, 5:30 60 5 30 1 http://www.biomedcentral.com/1471-2407/5/30 8 10 45 1 Table 5: Pattern of compliance to chemotherapy by age and radiotherapy 9 13 4 1 Variable 49 13 2 1 Age 74 13 7 1 Radiotherapy 36 15 16 1 ≥ >60 and < 65 (n 47 15 2 1 100) 54 15 2 1 65 (n 58 15 1 1 80) 63 15 1 1 P§ 67 15 3 1 No* (n 72 15 2 1 114) 78 15 3 1 Yes (n 83 15 1 1 66) 46 15 42 1 P§ = 57 15 0 1 = 71 15 0 1 = 81 15 0 1 = 9 18 18 1 N. of cycles administered, no. (%) 62 18 2 1 0.42 87 18 2 1 0.01 12 19 4 1 6 cycles 41 19 3 1 81 (81) 54 19 3 1 61 (76) 69 19 3 1 84 (74) 79 19 3 1 58 (88) 12 21 4 1 5 cycles 42 21 2 1 5 (5) 55 21 2 1 4 (5) 70 21 2 1 4 (3) 80 21 2 1 5 (8) 12 22 4 1 4 cycles 42 22 2 1 4 (4) 55 22 2 1 4 (5) 70 22 2 1 6 (5) 80 22 2 1 2 (3) 12 23 4 1 3 cycles 42 23 2 1 5 (5) 55 23 2 1 6 (8) 69 23 3 1 10 (9) 80 23 2 1 1 (1) 12 25 4 1 2 cycles 42 25 2 1 4 (4) 55 25 2 1 4 (5) 70 25 2 1 8 (7) 81 25 0 1 0 12 26 3 1 1 cycle 42 26 2 1 1 (1) 55 26 2 1 1 (1) 70 26 2 1 2 (2) 81 26 0 1 0 9 29 15 1 N. of day 8 omitted, no. (%) 62 29 2 1 0.37 87 29 2 1 0.86 12 30 0 1 0 41 30 3 1 69 (69) 54 30 3 1 51 (64) 69 30 3 1 77 (68) 79 30 3 1 43 (65) 12 31 0 1 1 41 31 3 1 23 (23) 54 31 3 1 18 (22) 69 31 3 1 24 (21) 79 31 3 1 17 (26) 12 33 0 1 2 42 33 2 1 5 (5) 55 33 3 1 8 (10) 70 33 2 1 9 (8) 80 33 2 1 4 (6) 12 34 0 1 3 42 34 2 1 3 (3) 55 34 2 1 3 (4) 70 34 2 1 4 (3) 80 34 2 1 2 (3) 9 37 21 1 Rate of cycles at reduced dose, no. (%) 62 37 2 1 0.54 87 37 2 1 0.83 12 38 3 1 ≤ 25% 41 38 3 1 86 (86) 54 38 3 1 66 (82) 69 38 3 1 97 (85) 79 38 3 1 55 (83) 12 39 3 1 >25% 41 39 3 1 14 (14) 54 39 3 1 14 (18) 69 39 3 1 17 (15) 79 39 3 1 11 (17) 9 42 17 1 Actual/planned duration, no. (%) 62 42 2 1 0.59 87 42 2 1 0.05 12 43 3 1 ≤ 1.25 41 43 3 1 99 (99) 54 43 3 1 78 (98) 68 43 5 1 114 (100) 79 43 3 1 63 (95) 12 45 3 1 >1.25 42 45 2 1 1 (1) 55 45 2 1 2 (2) 71 45 0 1 0 80 45 2 1 3 (5) 9 47 13 1 G-CSF utilization, no. (%) 62 47 2 1 0.12 79 47 3 1 <.0001 12 49 1 1 No 41 49 3 1 91 (91) 54 49 3 1 66 (82) 69 49 4 1 109 (96) 79 49 3 1 48 (73) 12 50 1 1 Yes 42 50 2 1 9 (9) 54 50 3 1 14 (18) 70 50 2 1 5 (4) 79 50 3 1 18 (27) 9 53 19 1 Treatment discontinuation, no. (%) 62 53 2 1 0.52 87 53 2 1 0.10 12 54 11 1 protocol completion 41 54 3 1 81 (81) 54 54 3 1 61 (76) 69 54 3 1 84 (74) 79 54 3 1 58 (88) 12 55 8 1 patient's refusal 42 55 2 1 6 (6) 55 55 3 1 9 (11) 69 55 3 1 12 (11) 80 55 2 1 3 (4) 12 57 9 1 treatment toxicity 41 57 3 1 11 (11) 55 57 2 1 7 (9) 69 57 3 1 13 (11) 80 57 2 1 5 (8) 12 58 3 1 other 42 58 2 1 2 (2) 55 58 2 1 3 (4) 70 58 2 1 5 (4) 81 58 0 1 0 9 61 13 1 Low compliance†, no. (%) 62 61 2 1 0.02 87 61 2 1 0.88 12 62 1 1 No 41 62 3 1 61 (61) 54 62 3 1 34 (42) 69 62 3 1 61 (54) 79 62 3 1 34 (52) 12 63 1 1 yes 41 63 3 1 39 (39) 54 63 3 1 46 (58) 69 63 3 1 53 (46) 79 63 3 1 32 (48) 9 66 44 1 * including 21 patients who received radiotherapy after the end of chemotherapy 9 67 45 1 § Fisher exact test or Wilcoxon-Mann Whitney test for naturally ordered variables) 9 69 80 1 † at least one of the following features: less than 6 cycles administered, withdrawal of more than one day-8 treatment, dose reduction in more than 9 70 74 1 25% of cycles, relative duration higher than 1.25, use of G-CSF treatment, treatment interruption because of patient's refusal or toxicity 8 77 39 1 As of March 2004, 41 patients had an event and disease- 51 77 39 1 time period of interest for this study was prevalently 8 78 39 1 free survival at 5 and 8 years was 76% and 71%. As shown 51 78 39 1 conservative, foreseeing adjuvant chemotherapy in elderly 8 80 39 1 in figure 1, there was no statistically significant difference 51 80 39 1 patients when either a very high risk of relapse or a very 8 81 38 1 in disease-free survival in the two age groups (p = 0.84). 51 81 39 1 strong patient's motivation existed, no severe or untreata- 51 83 39 1 ble comorbid conditions being evident. Overall, we antic- 8 84 8 1 Discussion 51 84 39 1 ipated that, under such conditions and because of 8 86 39 1 The present study was planned being aware that a selec- 51 86 39 1 selection bias, our study should produce underestimation 8 87 39 1 tion bias, typical of retrospective data collection, could 51 87 39 1 of toxicity and lack of compliance, this effect possibly 8 89 39 1 affect the results; namely, outside clinical trials, the 51 89 39 1 being as more pronounced as older the patients because 8 90 39 1 therapeutic strategy applied in our Institution during the 51 90 30 1 of the application of more restrictive criteria. 82 95 8 1 Page 6 of 10 70 97 20 0 (page number not for citation purposes)

8 5 28 1 BMC Health Services Research 2005, 5:22 60 5 30 1 http://www.biomedcentral.com/1472-6963/5/22 8 10 39 1 added to wait lists, they were not included in analyses of 51 10 38 1 Table 1: Characteristics of 9,231 subjects registered for isolated 8 11 81 1 coronary artery bypass surgery in British Columbia, 1991–2000. wait-list times. 51 14 8 1 Characteristic 82 14 1 1 N 87 14 2 1 (%) 8 15 39 1 There were 9,366 records of registration for isolated CABG 8 16 39 1 added to the Registry between January 1991 and Decem- 51 17 7 1 Age group (y) 8 18 39 1 ber 2000. We excluded 135 records of patients who were: 53 18 2 1 <50 83 18 2 1 732 87 18 2 1 (7.9) 53 19 3 1 50–59 82 19 2 1 2005 8 19 81 1 (21.7) emergency cases (30), removed on the registration date 53 20 3 1 60–69 82 20 2 1 3530 8 20 81 1 (38.2) (101), and had missing operating room reports (4). All 53 22 3 1 70–79 82 22 2 1 2770 8 22 81 1 (30.0) remaining 9,231 records had either the surgery date or the 53 23 2 1 ≥ 80 83 23 2 1 194 87 23 2 1 (2.1) 8 24 39 1 date and reason of removal from the list without surgery. 51 24 1 1 Sex 8 25 39 1 We restricted the analyses to the first 52 weeks after regis- 53 26 4 1 Women 82 26 2 1 1634 86 26 3 1 (17.7) 8 27 39 1 tration so that 475 (5%) patients remaining on the lists at 53 27 2 1 Men 82 27 2 1 7597 86 27 3 1 (82.3) 8 28 39 1 12 months were censored. Of those, 167 eventually 51 28 12 1 Urgency at registration 53 29 5 1 Priority 1 83 29 2 1 659 8 29 81 1 (7.1) underwent surgery; seven died; 78 received medical treat- 53 31 5 1 Priority 2 82 31 2 1 6496 8 31 81 1 (70.4) ment; 104 declined surgery; 17 were transferred to 53 32 5 1 Priority 3 82 32 2 1 1963 86 32 3 1 (21.3) 8 32 39 1 another surgeon or hospital; and 102 were removed for 53 33 5 1 Unknown 83 33 2 1 113 87 33 2 1 (1.2) 8 34 9 1 other reasons. 51 35 18 1 Major comorbidity at registration 53 36 3 1 None 82 36 2 1 4769 86 36 3 1 (51.7) 8 37 10 1 Priority groups 53 37 10 1 Minor comorbidity 82 37 2 1 2450 86 37 3 1 (26.5) 8 38 39 1 When assigning priority, all cardiac surgeons in BC apply 53 39 24 1 CHF, diabetes, COPD, rheumatoid arthritis, 82 39 2 1 2012 86 39 3 1 (21.8) 8 40 48 1 cancer common guidelines developed in 1990 [see Additional 8 41 53 1 Registration period file 1]. Using the location and degree of affected coronary 53 42 6 1 1991–1992 82 42 2 1 1724 86 42 3 1 (18.7) 8 43 39 1 anatomy and symptoms, the guidelines help to: identify 53 44 6 1 1993–1994 82 44 2 1 1889 86 44 3 1 (20.5) 8 44 39 1 patients for whom CABG can increase survival or improve 53 45 6 1 1995–1996 82 45 2 1 2010 86 45 3 1 (21.8) 8 46 39 1 quality of life[15]; classify patients according to urgency 53 46 6 1 1997–1998 82 46 2 1 1888 86 46 3 1 (20.5) 8 47 39 1 of treatment; and assign a maximum recommended wait- 53 47 6 1 1999–2000 82 47 2 1 1720 86 47 3 1 (18.6) 8 49 39 1 ing time (MRWT). Patients are assigned priority 1 if they 8 50 77 1 Abbreviations: CHF – congestive heart failure COPD – chronic require CABG urgently (eg, left main coronary artery sten- 8 51 59 1 obstructive pulmonary disease osis greater than 70%, MRWT three days); priority 2 if 8 53 39 1 there is moderate urgency (eg persistent unstable angina, 8 55 39 1 MRWT six weeks); or priority 3 if there is less urgency (eg 8 56 39 1 intractable chronic angina, MRWT 12 weeks). These 51 57 39 1 method[16]. Patients removed from the list for reasons 8 58 39 1 guidelines did not undergo any major revisions through 51 59 39 1 other than surgery were treated as censored observations. 8 59 20 1 the entire period under study. 51 62 39 1 Primary comparisons were done across synthetic cohorts 8 62 8 1 Comorbidity 51 63 39 1 of patients defined by two-year periods of registration on 8 64 39 1 Using the administrative data, coexisting medical condi- 51 65 39 1 the wait lists. Differences in the distributions of wait-list 8 65 39 1 tions were identified using all primary and secondary dis- 51 66 39 1 times across cohorts were examined using the log rank- 8 67 39 1 charge diagnoses recorded in all hospital discharge 51 67 39 1 test[17]. The average weekly surgery rate was calculated by 8 68 39 1 abstracts within one year prior to registration[13]. This 51 69 39 1 dividing the number of operations by the total number of 8 70 39 1 time frame was chosen in order to capture the presence of 51 70 39 1 patient-weeks on the list. The effect size for each registra- 8 71 39 1 chronic diseases that could have affected the waiting 51 72 39 1 tion period was estimated by hazard ratios for surgery 8 73 6 1 time[14]. 51 73 39 1 derived from a Cox proportional hazards model[18]. Haz- 51 75 39 1 ard ratios (HR) associated with registration periods evalu- 8 76 39 1 For each patient, we identified the presence of major and 51 76 39 1 ated the conditional weekly probability of undergoing 8 77 4 1 minor 15 77 6 1 comorbid 23 77 5 1 medical 31 77 7 1 conditions 40 77 5 1 present 47 77 1 1 at 51 78 39 1 CABG relative to the 1991–92 period. The priority groups 8 78 8 1 registration. 51 79 39 1 and the presence of comorbidity at registration were 51 81 39 1 included as independent variables in the Cox model to 8 81 13 1 Statistical methods 51 82 39 1 estimate adjusted effects. Age and sex were entered into 8 83 39 1 Waiting times were analyzed as prospective observations 51 84 39 1 the regression models as strata variables to avoid the pro- 8 84 39 1 beginning at the time of registration. Each subject had a 51 85 39 1 portionality assumption on these factors while using the 8 86 39 1 wait-list time calculated in calendar weeks from registra- 51 87 19 1 proportional hazards model. 8 87 39 1 tion to surgery or removal for other reasons. The cumula- 8 89 39 1 tive probability of undergoing surgery as a function of 51 90 39 1 The Clinical Research Ethics Board of the University of 8 90 39 1 wait-list time was estimated using the Kaplan-Meier 51 91 32 1 British Columbia approved the study protocol. 82 95 8 1 Page 3 of 10 70 97 20 0 (page number not for citation purposes)

8 5 28 1 BMC Health Services Research 2005, 5:22 60 5 30 1 http://www.biomedcentral.com/1472-6963/5/22 8 10 79 1 Table 2: Distribution of subjects registered for isolated coronary artery bypass surgery in British Columbia, 1991–2000, by priority 8 11 18 1 group and registration period 9 14 7 1 Registration 29 14 5 1 Priority 1 52 14 5 1 Priority 2 76 14 5 1 Priority 3 9 16 4 1 period 25 16 1 1 N 37 16 2 1 (%) 49 16 1 1 N 60 16 2 1 (%) 72 16 1 1 N 83 16 2 1 (%) 9 19 6 1 1991–1992 29 19 2 1 116 40 19 2 1 (6.7) 52 19 2 1 1221 63 19 3 1 (70.8) 76 19 2 1 334 86 19 3 1 (19.4) 9 21 6 1 1993–1994 29 21 2 1 110 40 21 2 1 (5.8) 52 21 2 1 1381 63 21 3 1 (73.1) 76 21 2 1 388 86 21 3 1 (20.5) 9 22 6 1 1995–1996 29 22 2 1 249 40 22 3 1 (12.4) 52 22 2 1 1363 63 22 3 1 (67.8) 76 22 2 1 374 86 22 3 1 (18.6) 9 23 6 1 1997–1998 29 23 2 1 117 40 23 2 1 (6.2) 52 23 2 1 1327 63 23 3 1 (70.3) 76 23 2 1 428 86 23 3 1 (22.7) 9 25 6 1 1999–2000 30 25 1 1 67 40 25 2 1 (3.9) 52 25 2 1 1204 63 25 3 1 (70.0) 76 25 2 1 439 86 25 3 1 (25.5) 9 27 28 1 Note: Excludes 113 subjects with unknown priority 8 34 5 1 Results 51 34 39 1 This can be seen in Figure 1, which shows access probabil- 8 35 39 1 In BC in the 1990s, 9,231 patients were registered on wait 51 35 39 1 ities for CABG in each priority group. The abscissa shows 8 37 39 1 lists for CABG and spent a total of 137,126 person-weeks 51 37 39 1 the number of weeks on the waiting list and the ordinate 8 38 39 1 waiting. Over the same period, 9,433 patients underwent 51 38 39 1 shows the probability of undergoing operation by that 8 40 39 1 isolated CABG without registration on wait lists. The most 51 40 39 1 week. Higher probabilities correspond to shorter wait list 8 41 39 1 prevalent groups at registration were men (82%), those 51 41 39 1 times. While all patients were removed at 52 weeks, for 8 43 39 1 without major comorbidities (52%), those registered in 51 43 39 1 graphical simplicity we show the first 36 weeks. Access 8 44 39 1 priority group 2 (70%), patients aged 60–69 (38%) and 51 44 39 1 probabilities in priority 3 (blue) were systematically lower 8 46 39 1 70–79 (30%) years, and those registered in 1995–96 51 46 39 1 indicating longer wait list times than among priority 2 8 47 39 1 (22%), Table 1. The proportion of patients registered in 51 47 24 1 (red line) or priority 1 (green line). 8 49 39 1 priority group 1 was lowest in 1999–2000 and highest in 8 50 69 1 Access to surgery by registration period the 1995–96 cohort, Table 2. The opposite pattern was 8 52 81 1 The differences in the proportion of patients undergoing observed in priority group 3. Of 8,756 patients who left 8 53 81 1 CABG were significant across registration periods (log the lists within 52 weeks: 7,991 underwent surgery; 90 8 55 81 1 rank test = 97.3, P < 0.0001), with longer wait-list times died while waiting; 176 received medical treatments; 188 8 56 79 1 for those registered between 1995 and 1998, Figure 2. declined surgery; and 311 were removed due to other 8 58 5 1 reasons. 51 59 39 1 Table 3 shows the number of weeks required for a speci- 8 61 81 1 fied proportion of patients to undergo the operation In all registration cohorts combined, the average weekly 8 62 81 1 across registration periods. Wait-list times in 1995–96 number of operations was 5.8 (95% confidence interval 8 64 81 1 were such that 10%, 25%, 50%, and 75% patients under- 5.7–6.0) per 100 patients listed, the median time on the 8 65 81 1 went surgery within 1, 6, 15, and 26 weeks, respectively, list was 11 weeks (25th percentile 5 weeks; 75th percentile 8 67 81 1 whereas half of the 1991–92 cohort underwent surgery 22 weeks), and the probability of undergoing surgery after 8 68 81 1 within 9 weeks, and 75% did so within 19 weeks. Com- 26 weeks on the list, twice the MRWT for priority group 3, 8 70 81 1 paring the 1995–96, 1997–98 and 1999–2000 cohorts we was 20%. 51 71 39 1 observed a compression in access to surgery, i.e., reduc- 8 73 81 1 tion in the length of wait-list interval required for a speci- As expected, there were significant differences among pri- 8 74 81 1 fied proportion to undergo the operation. As measured by ority groups, with larger proportions undergoing CABG at 8 75 81 1 and 50th percentiles of the wait the difference between 90th every week among more urgent patients (log rank test = 8 77 81 1 time distributions, 40% of the 1995–96 cohort under- 1611.9, P < 0.0001). The average weekly number of 8 78 81 1 went surgery within 33 weeks following the median time, operations per 100 patients on the list differed from 20.6 8 80 76 1 while it took 29 weeks for the 1999–2000 cohort. (19.0–22.2) in group 1 to 6.7 (6.5–6.8) in group 2 to 3.3 8 81 39 1 (3.1–3.4) in group 3. However, considerable variation in 8 83 81 1 While the median wait-list time was 11 weeks (the MRWT wait-list times was observed within each priority group. 8 84 81 1 of priority group 3) in all cohorts combined, 15% of the For instance, although half of group 1 underwent surgery 8 86 81 1 1991–92 and 1993–94 cohorts, 22% of the 1995–96 within two weeks and 90% underwent surgery by 12 8 87 81 1 cohort, 19% of the 1997–98 cohort, and 14% of the weeks, the remaining 10% waited another 1 to 32 weeks 8 89 81 1 1999–2000 cohort experienced an excessive wait, defined (total 13 to 44 weeks). 51 90 39 1 as longer than 26 weeks (data not shown). The average 82 95 8 1 Page 4 of 10 70 97 20 0 (page number not for citation purposes)

8 5 28 1 BMC Health Services Research 2005, 5:22 60 5 30 1 http://www.biomedcentral.com/1472-6963/5/22 8 10 80 1 Table 3: Percentiles of wait-list time (weeks) for subjects registered for isolated coronary artery bypass surgery in British Columbia 8 11 20 1 1991–2000 by registration period 9 14 12 1 Registration period 53 14 6 1 Percentile 27 16 2 1 10th 41 16 2 1 25th 55 16 2 1 50th 68 16 2 1 75th 82 16 2 1 90th 9 19 6 1 1991–1992 28 19 0 1 1 42 19 0 1 3 56 19 0 1 9 69 19 1 1 19 83 19 1 1 44 9 21 6 1 1993–1994 28 21 0 1 2 42 21 0 1 4 56 21 0 1 9 69 21 1 1 18 83 21 1 1 45 9 22 6 1 1995–1996 28 22 0 1 1 42 22 0 1 6 55 22 1 1 15 69 22 1 1 26 83 22 1 1 48 9 23 6 1 1997–1998 28 23 0 1 2 42 23 0 1 6 55 23 1 1 14 69 23 1 1 25 83 23 1 1 43 9 25 6 1 1999–2000 28 25 0 1 3 42 25 0 1 6 55 25 1 1 10 69 25 1 1 19 83 25 1 1 39 9 26 5 1 All periods 28 26 0 1 2 42 26 0 1 5 55 26 1 1 11 69 26 1 1 22 83 26 1 1 44 9 29 49 1 Note: probability of undergoing outpatient surgery within 26 weeks of registration is 0.804 8 35 76 1 Table 4: Average weekly rate of undergoing the operation from wait list for isolated coronary artery bypass surgery in British 8 37 41 1 Columbia 1991–2000 and adjusted rate ratios by registration period 9 39 7 1 Registration 22 39 6 1 Number of 33 39 9 1 Total wait time, 44 39 9 1 Crude Rate, per 60 39 1 1 SE 69 39 8 1 Hazard ratio 82 39 4 1 95% CI* 9 40 4 1 period 23 40 6 1 operations 36 40 3 1 weeks 48 40 2 1 100 9 43 6 1 1991–1992 25 43 2 1 1504 36 43 3 1 23047 48 43 1 1 6.5 60 43 1 1 0.2 71 43 2 1 1.00 82 43 4 1 referent 9 44 6 1 1993–1994 25 44 2 1 1646 36 44 3 1 25480 48 44 1 1 6.5 60 44 1 1 0.2 71 44 2 1 1.00 82 44 5 1 0.93, 1.08 9 45 6 1 1995–1996 25 45 2 1 1727 36 45 3 1 34186 48 45 1 1 5.1 60 45 1 1 0.1 71 45 2 1 0.70 82 45 5 1 0.65, 0.76 9 47 6 1 1997–1998 25 47 2 1 1613 36 47 3 1 30384 48 47 1 1 5.3 60 47 1 1 0.1 71 47 2 1 0.77 82 47 5 1 0.71, 0.83 9 48 6 1 1999–2000 25 48 2 1 1501 36 48 3 1 24029 48 48 1 1 6.2 60 48 1 1 0.2 71 48 2 1 0.94 82 48 5 1 0.88, 1.02 9 49 5 1 All periods 25 49 2 1 7991 36 49 4 1 137126 48 49 1 1 5.8 60 49 1 1 0.1 72 49 0 1 - 84 49 0 1 - 9 52 33 1 Abbreviations: SE = standard error; CI = confidence interval 9 53 38 1 *adjusted for priority group and comorbidity; stratified by age and sex 9 55 30 1 **0 patients were on the wait list on December 31 2001 8 61 39 1 5.6, P = 0.0183, 1 df; priority 2: chi-square = 58.2, P < 51 61 39 1 the 1991–92 cohort to 5.3 (5.0–5.6) in the 1995–96 8 62 39 1 0.0001, 1 df; priority 3: chi-square = 20.5, P < 0.0001, 1 51 62 39 1 cohort to 7.3 (6.9–7.7) in the 1999–2000 cohort. The 8 64 39 1 df). By 1999–2000, the pattern of change was different 51 64 39 1 adjusted HR in 1999–2000 was 0.99 (0.90–1.08) relative 8 65 16 1 between priority groups. 51 65 39 1 to 1991–92 (Table 5, columns 4 and 5). There was no dif- 51 67 39 1 ference in the distribution of wait-list times between 8 68 39 1 In priority group 1, the average weekly number of opera- 51 68 39 1 1991–92 and 1999–2000 cohorts (chi-square = 0.5, P = 8 70 39 1 tions per 100 patients listed declined from 42.4 (34.6– 51 70 6 1 0.5, 1 df). 8 71 39 1 50.2) in the 1991–92 cohort to 20.3 (17.7–22.9) in the 8 73 39 1 1995–96 cohort to 12.2 (9.2–15.3) in the 1999–2000 51 73 39 1 In priority group 3, the average weekly number of opera- 8 74 39 1 cohort (data not shown). Corresponding HRs and 95% 51 74 39 1 tions per 100 patients listed changed from 3.9 (3.4–4.3) 8 76 39 1 CIs are shown in Table 5, columns 2 and 3. The condi- 51 76 39 1 in 1991–92 to 2.8 (2.4–3.1) in 1995–96 to 4.0 (3.6–4.5) 8 77 39 1 tional weekly probabilities of undergoing surgery were 51 77 39 1 in 1999–2000. The adjusted HR associated with the 8 78 39 1 34% lower for the 1995–96 cohort, HR = 0.66 (0.50– 51 78 39 1 1999–2000 registration period was 1.07 (0.90–1.26) rela- 8 80 39 1 0.87), and 53% lower for the 1999–2000 cohort, HR = 51 80 39 1 tive to 1991–92 (Table 5, columns 6 and 7). There was no 8 81 39 1 0.47 (0.33–0.68), relative to 1991–92. There was a differ- 51 81 39 1 difference between the between 1991–92 and 1999–2000 8 83 39 1 ence in the distribution of wait-list times between 1995– 51 83 27 1 cohorts (chi-square= 0.6, P = 0.4, 1 df). 8 84 39 1 96 and 1999–2000 cohorts (chi-square = 9.9, P = 0.0017, 8 86 50 1 Discussion 1 df). 51 87 39 1 In this paper we studied the amount of time that patients 8 89 81 1 with CAD spent on CABG wait lists before and after the In priority group 2, the average weekly number of opera- 8 90 81 1 provincial government started providing supplementary tions per 100 patients listed varied from 7.3 (6.8–7.7) in 82 95 8 1 Page 7 of 10 70 97 20 0 (page number not for citation purposes)

8 5 28 1 BMC Health Services Research 2005, 5:22 60 5 30 1 http://www.biomedcentral.com/1472-6963/5/22 8 10 81 1 Table 5: Access to surgery by registration period and priority group for subjects registered for isolated coronary artery bypass surgery 8 11 43 1 in British Columbia 1991–2000, as measured by adjusted hazard ratios* 9 14 7 1 Registration 29 14 5 1 Priority 1 52 14 5 1 Priority 2 76 14 5 1 Priority 3 9 15 4 1 period 25 18 2 1 HR 35 18 5 1 (95% CI) 48 18 2 1 HR 58 18 5 1 (95% CI) 72 18 2 1 HR 82 18 5 1 (95% CI) 9 20 6 1 1991–1992 29 20 2 1 1.00 35 20 4 1 referent 52 20 2 1 1.00 59 20 4 1 referent 76 20 2 1 1.00 82 20 4 1 referent 9 22 6 1 1993–1994 29 22 2 1 0.79 34 22 6 1 (0.57, 1.09) 52 22 2 1 1.10 58 22 6 1 (1.01, 1.20) 76 22 2 1 0.78 81 22 6 1 (0.65, 0.92) 9 23 6 1 1995–1996 29 23 2 1 0.66 34 23 6 1 (0.50, 0.87) 52 23 2 1 0.71 58 23 6 1 (0.65, 0.78) 76 23 2 1 0.69 81 23 6 1 (0.58, 0.82) 9 24 6 1 1997–1998 29 24 2 1 0.49 34 24 6 1 (0.36, 0.67) 52 24 2 1 0.82 58 24 6 1 (0.75, 0.89) 76 24 2 1 0.75 81 24 6 1 (0.63, 0.88) 9 26 6 1 1999–2000 29 26 2 1 0.47 34 26 6 1 (0.33, 0.68) 52 26 2 1 0.99 58 26 6 1 (0.90, 1.08) 76 26 2 1 1.07 81 26 6 1 (0.90, 1.26) 9 28 32 1 Abbreviations: HR = hazard ratio; CI = confidence interval 9 30 38 1 *adjusted for priority group and comorbidity; stratified by age and sex 8 35 39 1 funding to increase the annual number of CABG opera- 51 35 39 1 list time in groups 2 and 3 decreased later, such that there 8 37 39 1 tions. We sought to determine whether the period of reg- 51 37 39 1 were no differences between the 1999–2000 and 1991–92 8 38 39 1 istration had an effect on the wait-list time and whether 51 38 5 1 cohorts. 8 40 39 1 the period effect was similar across priority groups. Using 8 41 39 1 the population-based registry, we compared the number 51 41 39 1 Studies examining access to elective care in Canada and 8 43 39 1 of weeks from registration to surgery for equal propor- 51 43 39 1 elsewhere often report median or mean times [22-25]. We 8 44 39 1 tions of patients across different registration periods. In 51 44 39 1 found that reporting the probability of undergoing CABG 8 46 39 1 these comparisons, we accounted for the priority mix at 51 46 39 1 as a function of wait-list time helps overcome some limi- 8 47 39 1 registration. We used prospective follow-up of all patients 51 47 39 1 tations of using single-value statistics in understanding 8 49 39 1 registered to avoid biases inherent in wait-list statistics 51 49 39 1 differences between periods [26-29]. For instance, we 8 50 37 1 based on patients undergoing the procedure only[19]. 51 50 39 1 were able to conclude that not only did changes in waits 51 52 39 1 reduce the median delay from 15 to 10 weeks in the 8 53 39 1 We found that the registration period had an effect on the 51 53 39 1 1995–96 and 1999–2000 cohorts, respectively, but also 8 55 39 1 amount of time that patients spent awaiting CABG in BC 51 55 39 1 provided 20% compression in access for 40% patients 8 56 39 1 in the 1990s. Wait-list times in the 1995–96 cohort were 51 56 39 1 staying on the lists longer than the median time. Studying 8 58 39 1 such that 50% and 75% patients underwent surgery 51 58 39 1 the distributions of wait-list times, we also were able to 8 59 39 1 within 15 and 26 weeks, respectively, whereas one-half of 51 59 39 1 compare the conditional weekly probability of undergo- 8 61 39 1 the 1991–92 cohort underwent surgery within nine weeks 51 61 39 1 ing CABG across registration periods while adjusting for 8 62 39 1 and three quarters did so within 19 weeks. This trend was 51 62 23 1 priority, comorbidity, age and sex. 8 64 39 1 reversed later, such that the 1999–2000 patients waited 8 65 39 1 no longer than did their 1991–92 counterparts. Relative 51 65 39 1 The lack of information on hospitals or surgeons could be 8 67 39 1 to the 1991–92 cohort, the conditional weekly probabili- 51 67 39 1 a limitation of this study as we were not able to adjust for 8 68 39 1 ties of undergoing surgery were 30% lower in 1995–96 51 68 39 1 the volume of CABG between the four tertiary care 8 70 39 1 patients, and 23% lower in 1997–98 patients, while there 51 70 39 1 hospitals where the operation was performed or for the 8 71 39 1 were no differences between periods 1991–92 and 1999– 51 71 24 1 wait lists between cardiac surgeons. 8 73 3 1 2000. 51 74 9 1 Conclusion 8 75 81 1 Our results provide evidence for a significant reduction in We also found that the effect of registration period was 8 77 81 1 wait-list time after supplementary funding was provided different across priority groups. In priority group 1, the 8 78 81 1 on an annual basis to tertiary care hospitals within a single wait-list time increased by the middle of the decade and 8 80 81 1 publicly funded health system. While system-level factors increased even further by the end of the decade. This may 8 81 81 1 such as changes in the organization or delivery of services reflect changes in queuing patients with more severe CAD 8 83 81 1 may have affected the wait-list time, one plausible reason including lessened concern about safety of delaying 8 84 81 1 for the observed reduction was that the hospitals had patients with left main stenosis[20] as well as increasing 8 86 81 1 capacity to increase the number of operations. Compared use of angioplasty to treat patients who would have for- 8 87 81 1 to 1995–96, there was a 12% increase (from 3,696 to merly been treated surgically[21]. In priority groups 2 and 8 89 81 1 4,174) in the total number of CABG operations in 1999– 3, the wait-list time also increased by the middle of the 8 90 81 1 2000, Table 6. Also, between 1995–96 and 1999–2000, decade. In contrast to priority group 1, however, the wait- 82 95 8 1 Page 8 of 10 70 97 20 0 (page number not for citation purposes)

8 5 28 1 BMC Health Services Research 2005, 5:22 60 5 30 1 http://www.biomedcentral.com/1472-6963/5/22 8 10 78 1 Table 6: Distributions of patients who were registered on wait lists or operated without delay in British Columbia 1991–2000, by 8 11 11 1 registration period 9 14 12 1 Registration Period 46 14 22 1 Patients identified as needing CABG 34 16 14 1 Registered on wait lists 66 16 15 1 Operated without delay 32 19 1 1 N 48 19 2 1 (%) 65 19 1 1 N 81 19 2 1 (%) 9 22 6 1 1991–1992 38 22 2 1 1724 54 22 3 1 (49.3) 70 22 2 1 1770 86 22 3 1 (50.7) 9 23 6 1 1993–1994 38 23 2 1 1889 54 23 3 1 (55.3) 70 23 2 1 1526 86 23 3 1 (44.7) 9 25 6 1 1995–1996 38 25 2 1 2010 54 25 3 1 (54.4) 70 25 2 1 1686 86 25 3 1 (45.6) 9 26 6 1 1997–1998 38 26 2 1 1888 54 26 3 1 (48.6) 70 26 2 1 1997 86 26 3 1 (51.4) 9 27 6 1 1999–2000 38 27 2 1 1720 54 27 3 1 (41.2) 70 27 2 1 2454 86 27 3 1 (58.8) 8 34 58 1 Additional material there was a 13% decrease (from 54% to 41%) in the pro- 8 35 39 1 portion of patients accessing the operation through wait 8 37 39 1 lists, indicating that supplementary funding was used to 52 38 13 1 Additional File 1 8 38 27 1 provide more operations without delay. 52 39 37 1 The Microsoft® Word 2002 file "BC consensus guidelines for CABG prior- 52 41 37 1 ity.doc" shows the guidelines used by British Columbian cardiac surgeons 8 41 39 1 The relatively short time frame following the funding 52 42 35 1 for assigning priority to patients registered for coronary artery bypass 8 43 39 1 increase is a limitation of our study. As discussed else- 52 43 4 1 grafting. 8 44 39 1 where, supplementary funding may not result in shorten- 52 44 9 1 Click here for file 8 46 79 1 [http://www.biomedcentral.com/content/supplementary/1472- ing wait lists if hospitals function near full capacity [30], 8 47 53 1 6963-5-22-S1.doc] or, if it is expected that funding will be withdrawn after 8 49 39 1 wait lists are reduced[10]. Reducing wait lists may require 8 50 39 1 investing in new health services facilities. In Denmark, 8 52 39 1 rates of open-heart surgery increased by 70% and the 51 52 16 1 Acknowledgements 8 53 39 1 median waiting times declined by half since 1994 when 51 54 39 1 The authors gratefully acknowledge the contributions of the following indi- 8 55 39 1 additional capacity for cardiac surgical care was estab- 51 55 39 1 viduals: Rita Sobolyeva, Lisa Kuramoto, Laurie Kilburn, Christopher Buller, 8 56 39 1 lished by increasing the number of operating theatres, 51 57 14 1 Min Gao, and Gordon Pate. 8 58 39 1 equipment and personnel [30]. On-going study of wait- 8 59 39 1 list times for CABG in BC will help determine the perma- 51 59 37 1 The following cardiac surgeons are contributors to the BCCR Surgical 8 61 79 1 Research Committee: Drs. James Abel, Richard Brownlee, Larry Burr, nence of the impact of supplementary funding. 51 62 39 1 Anson Cheung, James Dutton, Guy Fradet, Virginia Gudas, Robert Hayden, 8 63 81 1 Eric Jamieson, Michael Janusz, Shahzad Karim, Tim Latham, Jacques LeBlanc, Competing interests 51 65 39 1 Sam Lichtenstein, Hilton Ling, John Ofiesh, Michael Perchinsky, Peter Skars- 8 65 39 1 The author(s) declare that they have no competing 51 66 11 1 gard and Frank Tyers 8 67 6 1 interests. 51 69 38 1 This study received financial support from the: St Paul's Hospital Founda- 8 69 18 1 Authors' contributions 51 70 36 1 tion (ARL), Vancouver Coastal Health Research Institute (BGS, JMF), 8 71 39 1 ARL conceived and designed the study, acquired the data, 51 71 39 1 Michael Smith Foundation for Health Research (ARL), Canada Foundation 8 73 80 1 for Innovation (ARL, BGS), and Canada Research Chairs program (BGS). interpreted the results, and drafted the manuscript. BGS 8 74 80 1 None of the sponsors had any role in the study design; in the collection, conceived and designed the study, analysed the data, 8 75 81 1 analysis, and interpretation of data; in the writing of the report; or the deci- interpreted the results, and drafted the manuscript. RH 51 76 21 1 sion to submit the paper for publication. 8 77 39 1 participated in the design of the study, helped acquire the 8 78 39 1 data, and interpreted the results. MK helped acquire the 51 79 9 1 References 8 80 39 1 data, and interpreted the results. JMF participated in the 51 81 39 1 Explaining Waiting Times Variations for 1. 54 81 10 1 Siciliani L, Hurst J: 8 81 39 1 design of the study and interpreted the results. MTS partic- 54 82 36 1 Elective Surgery across OECD Countries. Volume OECD Health 8 83 81 1 7. Paris, Organisation for Economic Co-operation Working Papers No. ipated in the design of the study and interpreted the 54 84 13 1 and Development; 2003. 8 84 4 1 results. 15 84 1 1 All 18 84 5 1 authors 25 84 2 1 read 29 84 2 1 and 33 84 6 1 approved 41 84 2 1 the 45 84 3 1 final 51 85 39 1 British Columbia sends patients 2. 54 85 16 1 Katz SJ, Mizgala HF, Welch HG: 8 86 8 1 manuscript. 54 86 36 1 to Seattle for coronary artery surgery. Bypassing the queue 54 87 6 1 in Canada. 60 87 14 1 266:1108-1111. 1991, JAMA 51 88 39 1 Waiting for coronary 3. 54 88 23 1 Naylor CD, Sykora K, Jaglal SB, Jefferson S: 54 89 36 1 artery bypass surgery: population-based study of 8517 con- 54 90 36 1 secutive patients in Ontario, Canada. The Steering Commit- 82 95 8 1 Page 9 of 10 70 97 20 0 (page number not for citation purposes)

8 5 20 1 BMC Neuroscience 2005, 6:20 60 5 30 1 http://www.biomedcentral.com/1471-2202/6/20 8 10 15 1 Table 1: Antibodies used 9 13 3 1 Marker 32 13 4 1 Species 43 13 9 1 Working dilution 57 13 11 1 Vendor or productor 9 15 8 1 Proliferation 9 17 2 1 BrdU 32 17 5 1 Mouse IgG 43 17 2 1 1:500 57 17 16 1 Novo Castra (Newcastle, UK) 9 18 2 1 BrdU 32 18 4 1 Rat IgG 43 18 3 1 1:500 57 18 21 1 Accurate Chemicals (New York, USA) 9 19 14 1 Magnocellular neurons 9 20 1 1 OT 32 20 5 1 Rabbit IgG 43 20 3 1 1:2000 57 20 12 1 Produced by G. Alonso 9 22 1 1 VP 32 22 5 1 Rabbit IgG 43 22 3 1 1:2000 57 22 12 1 Produced by G. Alonso 9 23 6 1 Astrocytes 9 24 3 1 GFAP 32 24 5 1 Rabbit IgG 43 24 3 1 1:5000 57 24 15 1 Dakko (Glostrup, Denmark) 9 26 19 1 Microglia and endothelial cells 9 27 12 1 Lectin from bandeirae 43 27 3 1 1:2000 57 27 3 1 Sigma 9 28 9 1 Simplifolia (IsoB4) 9 30 18 1 Oligodendrocyte progenitors 9 31 2 1 NG2 32 31 5 1 Rabbit IgG 43 31 3 1 1:500 57 31 15 1 Chemicon (Temecula, USA) 9 32 4 1 Vessels 9 33 16 1 Rat IgG-Alexa 488 conjugated 32 33 5 1 Goat IgG 43 33 3 1 1:2000 57 33 17 1 Molecular Probes (Eugene, USA) 9 35 2 1 EBA 32 35 6 1 Mouse IGM 43 35 3 1 1:5000 57 35 24 1 Sternberger Monoclonals (Lutherville, USA) 9 36 3 1 Nestin 32 36 5 1 Mouse IgG 43 36 2 1 1:100 57 36 18 1 Hybridoma Bank (Iowa city, USA) 9 37 4 1 Vimentin 32 37 5 1 Mouse IgG 43 37 3 1 1:5000 57 37 3 1 Sigma 9 40 76 1 BrdU: bromodeoxy uridine; EBA: endothelial brain antigen; GFAP: glial fibrillary acidic protein; IgG-IgM: immunoglobulins type G and M; OT: 9 41 14 1 oxytocin; VP: vasopressin. 8 46 60 1 Imaging and quantification Table 2: Quantitative evaluation of the nature of SON 8 48 10 1 proliferative cells 51 48 39 1 After rinsing in PBS, labeled sections were mounted in 51 49 39 1 Mowiol and observed under a Biorad MRC 1024 confocal 27 50 19 1 % of BrdU-labeled nuclei associated 30 51 60 1 laser scanning microscope equipped with a krypton/argon with specific cell markers 51 52 39 1 mixed gas laser. Two laser lines emitting at 488 nm and 11 54 79 1 568 nm were used for exciting GFP and the Alexa-488 or Cell type 20 54 4 1 Marker 51 55 39 1 Cy3-conjugated secondary markers. The background 12 57 4 1 Neurons 20 57 4 1 VP + OT 37 57 0 1 0 51 57 39 1 noise of each confocal image was reduced by averaging 11 58 5 1 Astrocytes 21 58 3 1 GFAP 37 58 0 1 6 51 58 2 1 four 56 58 4 1 image 62 58 4 1 inputs. 69 58 2 1 The 74 58 8 1 organization 84 58 1 1 of 88 58 2 1 the 9 59 9 1 Oligo. Precursors 21 59 2 1 NG2 36 59 1 1 18 51 60 39 1 immunostained structures was studied 1) on single confo- 11 61 4 1 Microglia 21 61 3 1 IsoB4 37 61 0 1 8 51 61 39 1 cal images or 2) on bi-dimensional reconstructed images 12 62 3 1 Vessels 21 62 3 1 IsoB4 36 62 1 1 60 51 63 39 1 obtained by collecting 5 to 10 consecutive confocal 12 63 3 1 Vessels 21 63 3 1 Nestin 36 63 1 1 67 51 64 9 1 µm images 1 12 64 78 1 apart through the whole vibratome section Vessels 22 65 2 1 EBA 36 65 1 1 55 51 66 39 1 thickness, and by projecting on the same plane. Unaltered 9 67 81 1 digitalized images were transferred to a PC type computer BrdU was administrated to osmotically stimulated rats drinking 2% 9 68 35 1 saline during 6 days, and the animals were fixed 5 hours after the 51 69 38 1 and Adobe Photoshop was used to prepare final figures. 9 70 36 1 BrdU administration. The relative proportion of each BrdU-labeled 9 71 37 1 cell type was calculated from at least 4 sections per rat in 4 stimulated 9 72 81 1 Quantitative analysis was performed on series of sections rats. More than 500 BrdU-labeled nuclei were scored for most 9 73 4 1 markers. 51 73 39 1 passing through the middle portions of the SON (i.e. the 9 74 35 1 GFAP: glial fibrillary acidic protein; EBA: endothelial brain antigen; 51 75 39 1 largest SON areas): 1) The cell proliferation was quanti- 9 75 31 1 NG2: proteoglycan NG2; OT: oxytocin; VP: vasopressin. 51 76 39 1 fied by counting the BrdU-labeled nuclei detected in four 51 78 39 1 SON areas per rat, in at least three rats per experiment, 2) 51 79 39 1 The nature of proliferative cells was evaluated on sections 51 81 39 1 double-labeled for BrdU and a specific cell type marker. 51 82 39 1 The number of double-labeled cells was pooled among at 8 84 39 1 and (3) exciting each fluorochrome by the inappropriate 51 84 39 1 least 4 SON areas per rat, in 4 rats, and 3) The anatomical 8 85 39 1 illumination. This allowed us to confirm that the two sec- 51 85 39 1 organization of SON vessels was evaluated on sections of 8 87 39 1 ondary antibodies used in double immunostaining exper- 51 87 39 1 rat brains fixed by immerssion and double immunos- 8 88 39 1 iments did not induce artifactual fluorescent labelling and 51 88 39 1 tained for Rat-IgG and for VP + OT. The density of SON 8 89 39 1 that there was no overlap of the emission spectra of the 51 89 39 1 vessels was quantified according to the point-counting 8 91 13 1 two fluorochromes. 51 91 39 1 method [53], by scoring the number of point intersections 81 95 9 1 Page 17 of 19 70 97 20 0 (page number not for citation purposes)

8 5 15 1 BMC Urology 2005, 5:5 61 5 29 1 http://www.biomedcentral.com/1471-2490/5/5 8 10 81 1 Table 1: Highly expressed Casodex responsive genes in the LNCaP model. Data show fold-increase in gene expression over reference 8 11 51 1 samples. Each datum point represents mean ± SE from 3 independent experiments. 55 14 5 1 LNCaP-R 76 14 7 1 LNCaP-UR 9 16 6 1 GenBank # 18 16 3 1 Name 49 16 8 1 Control ± SE 59 16 8 1 Treated ± SE 70 16 8 1 Control ± SE 81 16 8 1 Treated ± SE 9 19 5 1 AA663884 18 19 17 1 synaptosomal-associated protein 50 19 6 1 15.18 ± 3.35 61 19 6 1 20.67 ± 9.33 72 19 6 1 20.92 ± 1.07 83 19 6 1 11.68 ± 0.61 9 21 5 1 AA700604 18 21 12 1 sorbitol dehydrogenase 50 21 6 1 13.73 ± 3.23 61 21 6 1 19.60 ± 6.50 72 21 6 1 21.08 ± 1.58 83 21 6 1 11.08 ± 1.39 9 22 4 1 T73556 18 22 15 1 fatty-acid-Coenzyme A ligase 52 22 5 1 5.11 ± .34 62 22 6 1 2.45 ± 0.74 73 22 6 1 3.94 ± 0.46 83 22 6 1 1.07 ± 0.25 9 23 5 1 AA028987 18 23 2 1 EST 51 23 6 1 5.38 ± 6.95 62 23 6 1 8.43 ± 5.68 73 23 6 1 6.22 ± 2.05 83 23 6 1 11.00 ± 3.74 9 25 4 1 R98851 18 25 18 1 membrane metallo-endopeptidase 50 25 6 1 10.28 ± 5.90 61 25 6 1 55.90 ± 30.2 71 25 7 1 79.29 ± 16.62 81 25 8 1 153.50 ± 56.07 9 26 4 1 N68465 18 26 25 1 UDP-N-acteylglucosamine pyrophosphorylase 51 26 6 1 4.37 ± 3.73 62 26 6 1 3.99 ± 1.44 72 26 6 1 12.47 ± 2.03 83 26 6 1 9.03 ± 2.41 9 27 4 1 H84113 18 27 19 1 retinal outer segment membrane 1 51 27 6 1 8.03 ± 2.06 61 27 6 1 14.73 ± 4.93 73 27 6 1 9.79 ± 1.39 83 27 6 1 7.09 ± 0.65 9 28 5 1 AA041499 18 28 12 1 cell division cycle 4-like 50 28 6 1 11.66 ± 7.07 61 28 6 1 10.67 ± 1.84 73 28 6 1 3.47 ± 0.44 83 28 6 1 3.64 ± 1.14 9 30 5 1 AA456695 18 30 17 1 H2B histone family, member Q 51 30 6 1 4.67 ± 3.02 62 30 6 1 4.23 ± 1.14 73 30 6 1 2.56 ± 0.27 83 30 6 1 4.67 ± 1.13 9 31 4 1 R82299 18 31 21 1 S-adenosylmethionine decarboxylase 1 51 31 6 1 8.35 ± 2.23 61 31 6 1 14.84 ± 4.06 72 31 6 1 11.11 ± 2.19 83 31 6 1 6.98 ± 0.51 9 32 5 1 AA063521 18 32 15 1 BCL2/adenovirus E1B 19 kD 51 32 6 1 3.59 ± 3.34 61 32 6 1 13.14 ± 3.94 73 32 6 1 6.76 ± 1.79 83 32 6 1 12.46 ± 2.32 9 34 5 1 AA419164 18 34 14 1 retinoic acid receptor, beta 51 34 6 1 9.90 ± 3.40 61 34 6 1 11.14 ± 2.23 73 34 6 1 7.44 ± 0.98 83 34 6 1 5.01 ± 0.46 9 35 5 1 AA459039 18 35 20 1 serine protease inhibitor, Kunitz type, 51 35 6 1 2.89 ± 2.90 61 35 6 1 10.44 ± 1.46 73 35 6 1 6.28 ± 0.99 83 35 6 1 9.94 ± 1.44 9 36 5 1 AA489752 18 36 5 1 cyclin G2 51 36 6 1 2.86 ± 2.67 61 36 6 1 10.97 ± 1.10 73 36 6 1 6.92 ± 1.62 83 36 6 1 15.14 ± 4.80 9 37 5 1 AA412053 18 37 10 1 CD9 antigen (p24) 51 37 6 1 4.80 ± 1.05 62 37 6 1 8.78 ± 2.72 73 37 6 1 5.44 ± 0.96 83 37 6 1 6.12 ± 0.84 8 44 49 1 Table 2: Gene expression fold change in the LNCaP model treated with Casodex. 63 46 7 1 Fold Change 9 49 6 1 GenBank# 18 49 3 1 Name 45 49 18 1 LNCaP-R Control vs. Treated 68 49 19 1 LNCaP-UR Control vs. Treated 9 52 4 1 T73556 18 52 15 1 fatty-acid-Coenzyme A ligase 53 52 3 1 - 2.09 76 52 3 1 - 3.68 9 53 4 1 R98851 18 53 18 1 membrane metallo-endopeptidase 53 53 3 1 + 5.44 76 53 3 1 + 1.94 9 54 5 1 AA063521 18 54 15 1 BCL2/adenovirus E1B 19 kD 53 54 3 1 + 3.66 76 54 3 1 + 1.84 9 56 5 1 AA459039 18 56 20 1 serine protease inhibitor, Kunitz type, 53 56 3 1 + 3.61 76 56 3 1 + 1.58 9 57 5 1 AA489752 18 57 5 1 cyclin G2 53 57 3 1 + 3.84 76 57 3 1 + 2.19 9 60 24 1 (+, fold up regulated and -, fold downregulated) 8 68 80 1 Hypoxia regulated gene expression in the LNCaP model Table 1 shows the list of genes whose expression was sig- 8 69 56 1 treated with Casodex nificantly increased compared to the reference RNA and 8 71 81 1 HIF-1 is a transcriptional activator involved in oxygen Table 2 shows the fold change of these genes in response 8 72 81 1 homeostasis in cells [25]. In tumors where hypoxic condi- to treatment with Casodex. These genes were overex- 8 74 81 1 tions are common, HIF-1 levels are increased and stimu- pressed in all three hybridizations. After normalization 8 75 81 1 late gene expression that promotes angiogenesis [26]. and filtering, a total of 160 genes from the samples were 8 76 81 1 HIF-1α is the inducibly expressed subunit that forms a analyzed using hierarchical clustering to show the similar- 8 78 81 1 HIF-1β. heterodimer with the constitutively expressed ities of expression. The dendogram separated the samples 8 79 81 1 HIF-1α has been implicated in androgen-independent according to overall similarity and shows that control and 8 81 81 1 prostate cancer [13] and is thought to mediate angiogen- treated LNCaP-R samples reside on the same branch and 8 83 81 1 esis in tumors by transcriptional up regulation of vascular control and treated LNCaP-UR samples reside on a paral- 8 84 81 1 endothelial growth factor [27]. BNIP3 was shown to be lel branch (Fig. 1). Of the 160 genes clustered were MME, 51 85 14 1 HIF-1α regulated by 8 86 81 1 [28,29] which led us to examine cyclin G2, and BNIP3 all of which showed differential reg- 8 87 81 1 whether the high levels of BNIP3 detected in our hybridi- ulation between the LNCaP-R and UR cells and all of 51 88 36 1 HIF-1α. zations were accompanied by expression of 8 89 81 1 Fig- which clustered in a same node (Fig. 1). The correlation 8 90 81 1 ure 2 (panel A) shows the results of an RT-PCR and reveals coefficient of this node was 0.85 which suggests that the 51 91 18 1 HIF-1α the expression of 8 91 81 1 in normal human prostate expression pattern of these genes was 85% similar. 82 95 8 1 Page 5 of 15 70 97 20 0 (page number not for citation purposes)

8 5 15 1 Filaria Journal 2005, 4:1 63 5 27 1 http://www.filariajournal.com/content/4/1/1 8 10 55 1 Table 1: Number of nodules palpated vs. number of nodules detected by ultrasound (USG). 33 12 13 1 Male patients N = 44 52 12 14 1 Female patients N = 17 76 12 7 1 Total N = 61 9 15 2 1 USG 38 15 2 1 303 59 15 1 1 98 79 15 2 1 401 9 17 4 1 Palpation 38 17 2 1 272 59 17 1 1 84 79 17 2 1 356 8 27 64 1 Measurement of onchocercomas USG, 13 27 1 1 in 16 27 4 1 female 22 27 5 1 patients 29 27 1 1 84 32 27 4 1 (geom. 38 27 3 1 mean 43 27 4 1 4.487) 8 28 81 1 The size of the nodules measured by USG varied from onchocercomas were palpated and 98 (geom. mean 8 29 81 1 0.18 cm2 to 9.44 cm2 (geom. mean 1.823, median 1.78). 5.210) were detected by USG (table 1). 51 31 39 1 of all Evaluation of the total cross-sectional area (cm2) 8 32 23 1 Correlation of palpation and USG 51 32 39 1 nodules measured per patient showed that nodules were 8 34 39 1 Onchocercomas could be clearly differentiated from 51 34 39 1 larger in older patients (figure 6). There was no significant 8 35 39 1 lymph nodes, where a sinus was visible with afferent and 51 35 39 1 relation between age and the occurrence of cystic nodules 8 37 39 1 efferent lymphatic vessels, and from lipomas, which 51 37 39 1 where worm movements could be documented by USG 8 38 39 1 appear brighter, more echo dense and homogeneous than 51 38 39 1 (Mann-Whitney U test: p = 0.1989). Volume measure- 8 40 39 1 onchocercomas. Foreign bodies such as pieces of metal, 51 40 39 1 ments of the onchocercomas were not done since this 8 41 39 1 stones or thorns could be differentiated from onchocerco- 51 41 39 1 would have been inaccurate with a 2-dimensional device. 8 43 39 1 mas due to their sharp signals, which result from their 51 43 6 1 Therefore 59 43 10 1 3-dimensional 71 43 7 1 ultrasound 80 43 5 1 scanners 88 43 2 1 are 8 44 39 1 well-defined borders and surrounding granuloma tissues. 51 44 6 1 necessary. 8 47 78 1 Patients with onchocercomas containing motile adult All palpated nodules could be detected by USG. In 23 8 49 47 1 filariae patients the number of nodules detected by palpation 8 50 81 1 In 18 out of 61 (14 male, 4 female) participants (table 2), matched with the number of onchocercomas detected by 8 52 81 1 22 onchocercomas with moving adult filariae were USG. In 26 patients more onchocercomas were detected 8 53 81 1 observed by USG. 15 patients had one nodule, 2 patients by USG than by palpation. These onchocercomas were of 8 55 81 1 had 2 nodules, and 1 patient had 3 nodules with living a larger size and tightly packed in a surrounding capsule 8 56 81 1 filariae. Anatomical sites where cystic nodules could be of connective tissue. By USG the examiner was able to dif- 8 58 81 1 detected were thorax, trochanter, iliac crest, crena analis, ferentiate more precisely the exact number of nodules, 8 59 81 1 knee and foot (heel). The frequency distribution of divided by septulae of connective tissue inside larger 51 61 10 1 onchocercomas 63 61 4 1 where 68 61 4 1 worm 74 61 8 1 movements 83 61 3 1 could 8 61 81 1 be onchocercomas. 21 61 3 1 USG 26 61 1 1 of 29 61 1 1 12 32 61 5 1 patients 39 61 5 1 showed 45 61 2 1 less 8 62 81 1 observed is shown in table 3. The living adult worms were onchocercomas than palpated: 2 lipomas, 2 foreign bod- 8 64 81 1 visible as acoustic enhancement (bright echo) reflected ies and 8 lymph nodes had been erroneously judged as 8 65 81 1 from tissue moving in echo-free areas of onchocercomas onchocercomas by palpation. The 2 painful foreign bod- 8 67 81 1 (additional file 1,3) and in one case in hypodense nodular ies, which were seen by USG, were excised and verified. 51 68 39 1 tissue (additional file 2). The echo-free parts of the nodule 8 70 21 1 Reproducibility of USG findings 51 70 39 1 provided a so-called "acoustic window" where worm 8 71 39 1 3 patients underwent USG examination on 3 consecutive 51 71 39 1 movements could easily be seen (figure 2A, 2B, 3B, 4A, 8 73 39 1 days at the first time-point (onset of the study). All 51 73 39 1 additional file 1,3) in comparison to nodules where adult 8 74 39 1 onchocercomas were detected at the same locations and 51 74 39 1 worms are packed tightly in a capsule of connective tissue 8 75 39 1 judged as onchocercomas. 45 of the 61 study patients 51 76 39 1 (figure 1A, 1B, 3A, additional file 2). Within the echo-free 8 77 39 1 underwent a second confirmatory ultrasound examina- 51 77 39 1 areas moving adult filariae appeared as coiled and twisted 8 78 39 1 tion 4 months after study start. All nodules judged as 51 78 39 1 structures moving around each other in the surrounding 8 80 39 1 onchocercomas by USG at the beginning of the study 51 80 39 1 fluid. The echo-free areas differed from a very small part 8 81 39 1 could be rediscovered and still proved to be onchocerco- 51 81 39 1 of the nodule to areas comprising the whole nodule (fig- 8 83 39 1 mas. 30 out of the 45 USG re-examined patients under- 51 83 28 1 ure 2A, 2B, 3B, 4A, 4Badditional file 1,3). 8 84 3 1 went 14 84 10 1 nodulectomies 26 84 1 1 at 30 84 2 1 this 35 84 7 1 time-point. 45 84 3 1 After 8 86 39 1 nodulectomies no more onchocercomas were visible by 51 86 39 1 Beside motile adult filariae, all cystic nodules contained 8 87 39 1 USG at locations known to be positive for these findings 51 87 39 1 parts of probably degenerated fragments of adult worms, 8 89 4 1 before. 51 89 39 1 seen as particles where the ultrasound beam cannot pass 82 95 8 1 Page 3 of 11 70 97 20 0 (page number not for citation purposes)

8 5 15 1 Filaria Journal 2005, 4:1 63 5 27 1 http://www.filariajournal.com/content/4/1/1 8 10 81 1 Table 2: Total numbers of male and female patients with onchocercomas, in which motile or non-motile adult worms were detected 8 11 13 1 by ultrasound (USG). 49 14 8 1 Male patients 64 14 10 1 Female patients 82 14 3 1 Total 53 16 1 1 14 68 16 0 1 4 9 16 77 1 18 (29.5%) Patients with onchocercomas where adult motile filariae could be 9 18 9 1 detected by USG 53 19 1 1 30 68 19 1 1 13 9 19 77 1 43 (71.5%) Patients with onchocercomas where no adult motile filariae could 9 20 11 1 be detected by USG 9 21 3 1 Total 53 21 1 1 44 68 21 1 1 17 81 21 5 1 61(100%) 8 30 50 1 Table 3: Anatomical sites where onchocercomas with living adult filariae occurred: 9 32 73 1 Total Anatomical site 20 32 6 1 Crena analis 29 32 6 1 Trochanter 38 32 5 1 Iliac crest 48 32 4 1 Thorax 57 32 2 1 Knee 66 32 2 1 Foot 23 35 0 1 7 32 35 0 1 5 41 35 0 1 3 50 35 0 1 3 58 35 0 1 3 67 35 0 1 1 9 35 65 1 22 Number 75 35 13 1 (observed in 18 patients) 8 42 39 1 through and is reflected and visible as double layers mov- 51 42 39 1 ulectomies. In cystic onchocercomas adult filariae were 8 44 34 1 ing in the cystic fluid (figure 4A, additional file 3). 51 44 39 1 found by histology (figure 5 A–C). A nucleus is visible in 51 45 39 1 the hypodermis of the worms and intact microfilariae are 8 47 39 1 The Pulse Wave Doppler Technique as described for the 51 47 39 1 visible inside the uterus as sign of vitality at the time-point 8 48 20 1 detection of adult filariae of 44 48 4 1 [8,14] 29 48 32 1 of nodulectomy. Wuchereria bancrofti 8 50 39 1 was used to visualise movements of living worms of O. 8 51 50 1 Discussion as a function of time and frequency, which how- volvulus 8 53 81 1 In the present study, onchocerciasis patients were exam- ever appears slow and rare in comparison to adult filariae 8 54 81 1 ined by ultrasonography to determine the frequency of in lymphatic vessels. Thus the relevance of the Pulse Wave 51 56 17 1 detection of living adult 8 56 81 1 in onchocercomas. Doppler technique appears less important for the observa- 69 56 7 1 O. volvulus 8 57 81 1 The current study is an improvement over previous tion and confirmation in onchocerciasis in comparison to 8 59 81 1 attempts to detect worms in onchocercomas as it provides examinations in bancroftian filariasis. 51 60 39 1 video documentation of worm movements and shows 8 62 16 1 Correlation to histology 51 62 39 1 that, with the availability of higher resolution transducers, 8 63 39 1 A preliminary analysis of onchocercomas by histology, 51 63 39 1 the frequency of detection of worms containing moving 8 65 39 1 confirmed the results on the proportion of live adult 51 65 39 1 adult worms is surprisingly high. In 18 out of 61 (29.5%) 8 66 39 1 worms from an earlier study (85%)[10], which was based 51 66 39 1 of the examined patients, 22 onchocercomas with moving 8 68 39 1 in the same endemic area. The data clearly show that most 51 68 18 1 adult filariae were detected. 8 69 39 1 of non-motile worms are vital but that the adult worms 8 71 39 1 are not detectable and do not appear as motile worms by 51 71 39 1 Clinical trials to verify potentially macrofilaricidal effects 8 72 39 1 USG. The reasons may be that i) worms are too tightly 51 72 39 1 of drugs currently lack methods for repeatable long-term 8 73 39 1 packed in host tissue to appear motile; ii) the connective 51 73 39 1 observation of adult living filariae in onchocercomas in 8 75 39 1 tissue surrounding the onchocercomas is thick and thus 51 75 39 1 vivo. The use of ultrasonography to observe onchocerco- 8 76 39 1 reflects the ultrasound beam of the capsule but not of the 51 76 39 1 mas has been rare, since movements of adult worms 8 78 9 1 worms inside. 51 78 39 1 embedded in nodules surrounded by connective tissue 51 79 39 1 were difficult to image. In fact there is only one case listed 8 81 39 1 Out of the 18 patients presenting cystic nodules with 51 81 39 1 in the literature where moving adult worms were detected 8 82 39 1 motile worms, 8 patients were re-examined after 4 51 82 39 1 [5]. Homeida et al. [11] first described the use of USG to 8 84 39 1 months before the above-mentioned nodulectomies. All 51 84 39 1 detect onchocercomas in patients. Poltera and Zak [16] 8 85 39 1 cystic nodules could be rediscovered at the same location 51 85 6 1 published 60 85 1 1 in 63 85 3 1 1988 69 85 1 1 an 73 85 7 1 ultrasound 82 85 3 1 study 88 85 1 1 on 8 87 28 1 and still contained moving adult filariae. 51 87 39 1 onchocercomas, where they described the ability to use 51 88 39 1 USG to observe alterations in nodule tissue after treat- 8 90 39 1 To correlate these findings in the histology 3 patients pre- 51 90 39 1 ment in vitro. Leichsenring et al. [12] evaluated in 1990 8 91 39 1 senting nodules with visible movements underwent nod- 51 91 39 1 the use of ultrasound examinations on patients with 82 95 8 1 Page 5 of 11 70 97 20 0 (page number not for citation purposes)

8 5 18 1 Nutrition Journal 2005, 4:12 64 5 25 1 http://www.nutritionj.com/content/4/1/12 11 11 44 1 120 1.1 11 13 1 0 1.0 54 14 1 0 100 12 15 0 0 .9 55 17 1 0 80 12 17 0 0 .8 9 18 1 3 Survival 52 18 1 3 patients 12 19 0 0 .7 55 20 1 0 60 84 21 2 1 SEX 9 21 1 4 Cumulative 12 21 0 0 .6 52 22 1 0 of 52 23 4 3 40 Number 86 23 3 0 Female 12 24 0 0 .5 12 25 76 1 Male 20 .4 58 26 6 0 Jan-July 2003 67 26 6 0 Aug-Jan 2004 12 26 69 1 Feb-July 2004 -.2 16 26 1 0 0.0 20 26 0 0 .2 24 26 0 0 .4 27 26 0 0 .6 31 26 0 0 .8 34 26 1 0 1.0 38 26 1 0 1.2 42 26 1 0 1.4 45 26 1 0 1.6 13 28 54 1 Periods of the study survival interval (years) 51 31 10 1 Figure 3 Sex Bangwe distribution study 61 31 27 1 of patients enrolled in three periods of the 8 31 35 1 Figure 1 Survival analysis of all home based care patients – Kaplan Meier 51 32 38 1 Sex distribution of patients enrolled in three periods of the 8 33 35 1 Survival of all home based care patients – Kaplan Meier 51 34 9 1 Bangwe study. 8 34 5 1 analysis. 12 42 1 0 60 51 43 36 1 Table 1: Clinical staging by period of recruitment – Bangwe 12 45 1 0 50 66 45 8 1 Clinical staging 87 45 2 1 Total 12 48 1 0 40 52 48 4 1 PERIOD 65 48 0 1 2 73 48 0 1 3 81 48 0 1 4 12 50 1 0 30 57 51 0 1 1 64 51 1 1 14 72 51 1 1 40 80 51 2 1 111 10 51 79 3 165 patients 57 52 0 1 2 65 52 0 1 3 72 52 1 1 30 80 52 1 1 87 88 52 2 1 120 57 53 0 1 3 65 53 0 1 1 72 53 1 1 17 80 53 1 1 46 12 53 77 1 64 20 51 55 2 1 Total 64 55 1 1 18 72 55 1 1 87 80 55 2 1 244 10 55 79 1 349 of 40 56 7 0 Std. Dev = 7.77 57 56 0 1 % 64 56 1 1 5% 71 56 2 1 25% 79 56 2 1 70% 10 56 79 3 100% Number 10 40 57 6 0 Mean = 32.0 40 59 5 0 N = 350.00 13 59 53 1 Chi sq = 7.2, df 4, p = 0.13 0 14 60 74 1 period 1 – patients recruited before food supplementation started 15.0 20.0 25.0 30.0 35.0 40.0 45.0 50.0 15 61 70 1 period 2 – patients recruited in the first six months after food 17.5 22.5 27.5 32.5 37.5 42.5 47.5 51 62 13 1 supplementation started 14 63 74 1 period 3 – patients recruited in the second six months period after Age in years 51 65 16 1 food supplementation started 8 66 29 1 Figure 2 Age distribution of home based care patients 8 67 29 1 Age distribution of home based care patients. 51 71 39 1 2). Females were younger (mean age of 30.9 years with 51 73 39 1 standard deviation (SD) of 7.8) than males (mean age of 51 74 39 1 33.4 years with SD of 7.5). More females than males were 8 75 39 1 service of which 97 were still alive. It appears that about 51 76 39 1 enrolled in the first six months of the study (Figure 3). 8 77 39 1 half the chronic sick in the study area were enrolled at the 51 77 39 1 However there was no apparent difference in case severity 8 78 3 1 time. 51 78 39 1 of patients enrolled in the different periods of the study 51 80 39 1 based on symptoms of fever, cough, lower limb pain and 8 81 39 1 Of the 360 patients, one third died within 6 months (Fig- 51 81 39 1 thrush using discriminant analysis (Wilks' Lambda Test = 8 83 39 1 ure 1). The median survival was 1.2 years up to July 2004. 51 83 32 1 0.98, p= 0.803), or body mass index (p = 0.63). 8 84 39 1 Since the start of the study 199 of the 360 patients have 8 86 62 1 Clinical stage at presentation died (56%). 51 87 39 1 The majority of patients presented in an advanced stage of 8 89 81 1 disease, with 70% in stage 4 (Table 1). None presented The age distribution of patients follows one well recog- 8 90 81 1 with stage 1 disease. There is no statistical difference in the nised in AIDS patients with a mean age of 32 years (Figure 83 95 7 1 Page 3 of 8 70 97 20 0 (page number not for citation purposes)

8 5 18 1 Nutrition Journal 2005, 4:12 64 5 25 1 http://www.nutritionj.com/content/4/1/12 51 10 39 1 stages of presentation in the three periods (Chi2 = 7.2, df 13 11 18 1 Nutritional status on presentation 51 12 9 1 = 4, p = 0.13). 11 13 1 0 60 51 15 19 1 Nutritional status of patients 11 16 1 0 50 51 16 39 1 The mean BMI at presentation was 18.5 kg/m2 (SD 3.1). 51 18 39 1 Half the patients were malnourished with a body mass 11 19 78 1 on enrolment. A index (BMI) of less than 18.5 kg/m2 40 51 21 39 1 quarter was severely malnourished below 16 kg/m2 (Fig- 11 22 78 1 ure 4). The nutritional status of the group presenting 30 51 24 39 1 before July 2003 is similar to that of the group presenting 9 24 1 3 patients 51 25 10 1 later (Table 2). 11 26 1 0 20 9 28 1 0 of 51 28 39 1 The three follow up nutrition surveys provide some evi- 11 29 1 0 10 9 29 38 3 Std. Dev = 3.07 Number 51 30 39 1 dence of the changing status of the patients who survived. 40 30 5 0 Mean = 18.5 51 31 39 1 No one received food before the first follow up survey in 12 31 33 1 N = 310.00 0 14 32 76 2 June 2003. During this time nutritional status remained 31.0 19.0 21.0 23.0 25.0 27.0 29.0 17.0 15.0 13.0 11.0 51 34 39 1 constant (Table 3). In subsequent surveys the mean BMI 13 35 77 1 per 100 days by the of those still alive rises by 0.49 kg/m2 Body mass index - kg/metre2 51 37 39 1 by the time of the third second survey and 0.46 kg/m2 8 38 35 1 Figure 4 presentation Nutritional status of home based care patients on first 51 38 39 1 survey, not quite statistically significant (Anova – F = 2.58, 8 40 35 1 Nutritional status of home based care patients on first 51 40 11 1 df = 2, p = 0.08). 8 41 8 1 presentation. 51 43 39 1 There is a small group of patients, 22 in number, who 51 44 39 1 have survived through all three surveys (Table 4). For 51 46 39 1 them there is an increase in the rate of change of BMI 51 47 39 1 between the pre-food and the first post food period but 8 49 81 1 not the second period (Wilks' Lambda = 0.84, p = 0.17, Table 2: Nutritional status of patients enrolled before and after 8 50 15 1 start of food distribution 51 50 18 1 partial eta squared = 0.16). 9 52 3 1 Period 19 52 19 1 Mean BMI (kg/m2) Lower 95% CI 39 52 7 1 Upper 95% CI 51 53 14 1 Oil supplementation 9 55 81 1 The addition of oil to the food package has no effect on Before July 2003 19 55 2 1 18.4 30 55 2 1 17.8 39 55 2 1 19.0 9 56 81 1 nutritional status as measured by a change in mean BMI After June 2003 19 56 2 1 18.6 30 56 2 1 18.1 39 56 2 1 19.1 51 58 15 1 per 100 days (Table 5). 8 61 77 1 Table 3: Mean change in BMI per 100 days for patients who survived from initial assessment to one of three surveys in Bangwe 32 63 4 1 Number 43 63 2 1 Mean 51 63 7 1 Std. Deviation 62 63 5 1 Std. Error 71 63 18 1 95% Confidence Interval for Mean 72 66 7 1 Lower Bound 82 66 7 1 Upper Bound 9 69 7 1 Change in BMI 37 69 1 1 61 47 69 1 1 .07 57 69 2 1 1.64 67 69 1 1 .21 77 69 2 1 -.35 88 69 1 1 .49 9 70 6 1 to June 2003 9 71 7 1 Change in BMI 37 71 1 1 70 47 71 1 1 .49 57 71 1 1 .82 67 71 1 1 .10 78 71 1 1 .29 88 71 1 1 .68 9 72 7 1 to Nov 2003 9 74 7 1 Change in BMI 37 74 1 1 81 47 74 1 1 .46 57 74 1 1 .96 67 74 1 1 .11 78 74 1 1 .25 88 74 1 1 .67 9 75 6 1 to July 2004 9 76 2 1 Total 36 76 2 1 212 47 76 1 1 .36 57 76 2 1 1.17 67 76 1 1 .08 78 76 1 1 .20 88 76 1 1 .52 9 77 3 1 Model 19 77 6 1 Fixed Effects 57 77 2 1 1.16 67 77 1 1 .08 78 77 1 1 .20 88 77 1 1 .51 19 79 8 1 Random Effects 67 79 1 1 .13 77 79 2 1 -.20 88 79 1 1 .91 9 80 4 1 ANOVA 30 80 8 1 Sum of Squares 48 80 1 1 df 52 80 7 1 Mean Square 68 80 0 1 F 78 80 1 1 Sig. 13 81 4 1 Between 26 81 2 1 6.95 48 81 0 1 2 57 81 2 1 3.48 67 81 2 1 2.58 78 81 1 1 .08 14 82 4 1 Groups 10 84 8 1 Within Groups 24 84 3 1 281.96 47 84 2 1 209 57 84 2 1 1.35 15 85 2 1 Total 24 85 3 1 288.91 47 85 2 1 211 83 95 7 1 Page 4 of 8 70 97 20 0 (page number not for citation purposes)

8 5 18 1 Nutrition Journal 2005, 4:12 64 5 25 1 http://www.nutritionj.com/content/4/1/12 8 10 54 1 Table 4: Rate of change of nutritional status of 22 patients who survived all three surveys 56 13 2 1 Mean 69 13 7 1 Std. Deviation 86 13 1 1 N 9 15 39 1 change in BMI per 100 days from initial assessment to June 2003 survey 57 15 1 1 .30 72 15 2 1 1.80 86 15 1 1 22 9 17 32 1 change in BMI per 100 days from June to Nov 2003 surveys 57 17 1 1 .84 72 17 1 1 .92 86 17 1 1 22 9 18 35 1 change in BMI per 100 days from Nov 2003 to July 2004 surveys 57 18 1 1 .17 72 18 2 1 1.09 86 18 1 1 22 9 21 31 1 Wilks' Lambda = 0.84, p = 0.17, partial eta squared = 0.16 8 27 81 1 Table 5: Comparison of change in nutritional status (change in BMI per 100 days) between those who did or did not receive oil as part 8 28 14 1 of food supplementation 49 30 10 1 oil actually received 61 30 4 1 Number 67 30 2 1 Mean 72 30 7 1 Std. Deviation 81 30 8 1 Std. Error Mean 9 33 35 1 rate of BMI change per 100 days from initial assessment to latest 53 33 2 1 OIL 63 33 1 1 50 68 33 1 1 .26 75 33 1 1 .67 84 33 1 1 .09 9 34 3 1 survey 52 36 4 1 NO OIL 63 36 1 1 14 68 36 1 1 .39 74 36 2 1 1.03 84 36 1 1 .27 9 38 19 1 Student t = -0.59, df = 62, p = 0.56 15 45 28 1 Survival of patients recruited during three different periods 57 45 26 1 Survival of Home based care patients 13 46 1 0 1.1 57 47 13 1 oil supplementation 13 49 1 0 1.0 55 49 1 0 1.1 14 51 42 0 1.0 .9 38 53 4 1 PERIOD 56 53 0 0 .9 14 54 0 0 .8 42 55 0 0 3 56 55 0 0 .8 11 56 3 3 .7 Survival 42 56 4 0 3-censored 53 56 3 3 .7 Survival 83 57 6 1 oil allocation 42 58 0 0 2 14 58 42 1 .6 .6 84 59 1 0 OIL 11 59 1 4 Cumulative 42 60 4 0 2-censored 53 60 3 4 Cumulative .5 14 61 0 0 .5 42 61 0 0 1 56 62 0 0 .4 84 62 3 0 NO OIL 14 63 0 0 .4 42 63 4 0 1-censored 56 63 0 0 .3 14 64 1 0 -.2 17 64 1 0 0.0 19 64 0 0 .2 22 64 0 0 .4 24 64 0 0 .6 27 64 0 0 .8 29 64 6 0 1.0 1.2 1.4 37 64 1 0 1.6 56 64 1 0 -.2 59 64 1 0 0.0 62 64 0 0 .2 65 64 0 0 .4 68 64 0 0 .6 70 64 0 0 .8 73 64 1 0 1.0 76 64 1 0 1.2 79 64 1 0 1.4 81 64 1 0 1.6 15 66 10 1 survival interval (year) 57 66 11 1 survival interval (years) 9 68 35 0 Period 1 – Jan – June 2003; period 2 July – Nov 2003; period 3 – Dec 2003 – July 2004 51 69 13 0 Log Rank 2.24, df = 1, p = 0.13 9 70 13 0 Log Rank 0.04, df = 2, p = 0.98 51 71 39 1 Figure 6 Survival allocated of oil home supplementation based care patients – Kaplan who Meier were analysis or were not 8 72 38 1 Figure 5 survival during three of patients different enrolled periods in – the Kaplan home Meier based analysis care scheme 51 72 39 1 Survival of home based care patients who were or were not 8 73 38 1 survival of patients enrolled in the home based care scheme 51 74 34 1 allocated oil supplementation – Kaplan Meier analysis. 8 75 35 1 during three different periods – Kaplan Meier analysis. 51 81 31 1 actually received oil (Figure 6) and those who 83 81 7 1 oil (Figure 8 82 5 1 Survival 51 82 39 1 7) compared to those who did not, although results are 8 84 39 1 A third of patients died (112) within 4 months of being 51 84 39 1 only statistically significant for those who actually 8 85 39 1 first seen. Half (180) survived fourteen months (Figure 1). 51 85 39 1 received oil. Oil seems to have an effect but only for those 8 87 39 1 There was no difference in the survival patterns of those 51 87 38 1 who survive six months from time of initial assessment. 8 88 39 1 who in the first months after presentation did not receive 8 90 39 1 food compared to those who received food from the start 51 90 39 1 The survival of Period 1 patients prior to receiving food 8 91 32 1 allocated (Figure 5). Survival was better in those 41 91 6 1 to receive 51 91 39 1 can be compared to a group of patients in Period 2 over a 83 95 7 1 Page 5 of 8 70 97 20 0 (page number not for citation purposes)

8 5 18 1 Nutrition Journal 2005, 4:12 64 5 25 1 http://www.nutritionj.com/content/4/1/12 8 10 81 1 Table 6: Mean Body Mass Index of households all ages comparing two groups, one surveyed (in June 2003) before and the other (in July 8 11 17 1 2004) after food distribution 23 14 3 1 Group 42 14 1 1 N 55 14 2 1 Mean 66 14 7 1 Std. Deviation 79 14 8 1 Std. Error Mean 9 17 2 1 BMI 23 17 2 1 2003 41 17 2 1 506 55 17 2 1 21.86 69 17 2 1 6.54 83 17 1 1 .29 23 18 2 1 2004 41 18 2 1 343 55 18 2 1 19.38 69 18 2 1 4.22 83 18 1 1 .23 9 21 19 1 Student t = 6.2, df = 847, p < 0.001 8 27 76 1 Table 7: Mean BMI of households of residents over 4 years of age (in June 2003) before and the other (in July 2004) after food 23 29 3 1 Group 42 29 1 1 N 55 29 2 1 Mean 66 29 7 1 Std. Deviation 79 29 8 1 Std. Error Mean 9 32 2 1 BMI 23 32 2 1 2003 41 32 2 1 434 55 32 2 1 22.02 69 32 2 1 6.39 83 32 1 1 .31 23 33 2 1 2004 41 33 2 1 290 55 33 2 1 19.65 69 33 2 1 4.32 83 33 1 1 .25 9 36 18 1 Student t = 5.5, df = 722, p < 0.0 8 41 39 1 oil being eaten by the patient so providing a concentrated 51 41 39 1 females in the before-food group. It may be that males 8 43 39 1 source of energy for those patients who are not terminally 51 43 39 1 tend not to seek home based care until it is known that 8 44 39 1 ill. An alternative suggestion is that oil is a saleable com- 51 44 39 1 food is available. However, the severity of disease of these 8 46 39 1 modity and money so realised may be used to purchase 51 46 39 1 patients does not seem to be different from the severity of 8 47 39 1 essential commodities such as water and charcoal. This 51 47 39 1 disease of those presenting before the food handouts 8 49 39 1 possible explanation fits with the result of oil having no 51 49 39 1 started. It appears that the two groups at first presentation 8 50 39 1 demonstrable effect on nutritional status. There is a slight 51 50 11 1 are comparable. 8 52 39 1 suggestion which is not statistically significant, as seen in 8 53 39 1 the survival curves for clinical stage 4 patients that food 51 53 39 1 Disease progression without antiretroviral drugs is usually 8 55 39 1 may prolong survival after the first three months. No such 51 55 39 1 inevitable and insidious. Some patients who present with 8 56 26 1 difference is found in stage 3 patients. 51 56 39 1 terminal illnesses require palliative care such as opiates 51 58 39 1 and soon die. Others do stabilise with the majority 8 59 39 1 Food supplementation has not helped to maintain body 51 59 39 1 needing continuous or intermittent treatment to provide 8 61 39 1 mass in household members of home based care patients. 51 61 39 1 palliation of symptoms. But should food supplementa- 8 62 39 1 This apparently disappointing result needs interpretation. 51 62 29 1 tion be considered one such intervention? 8 64 39 1 The reduction in mean BMI of household members may 8 65 39 1 be attributable to the socio-financial catastrophe brought 51 65 39 1 The benefits of food, if they exist, may be outweighed by 8 67 39 1 on by loss of income and increase in expenditure due to 51 67 39 1 the costs, not just to donor organisations, but to the 8 68 39 1 the chronic ill health of one or two of the adults in the 51 68 39 1 patients and their families. Food distribution in urban 8 70 39 1 family. The longer the adult remains alive and ill, the 51 70 39 1 areas has problems and food may not get to the people 8 71 39 1 longer the loss of earnings, drain on resources and ensu- 51 71 39 1 intended. Indeed the WFP were hesitant about initiating 8 73 39 1 ing poverty. This may account for the reduction in BMI of 51 73 39 1 the programme because of the problems likely to be 8 74 39 1 PLWA households some of whose patients have survived 51 74 39 1 encountered in Bangwe. The social disruption and ani- 8 76 39 1 for 12 months or more. Perhaps food supplementation 51 76 39 1 mosities produced by the free but selective distribution of 8 77 39 1 has alleviated this tendency to malnutrition. It could have 51 77 39 1 food in a community with slender food security may be 8 78 20 1 been worse without the food. 51 78 39 1 substantial. The difficulty of families where the adults are 51 80 39 1 unable to get out of the house to collect the food is real. 8 81 39 1 An observational study of this sort is difficult to interpret. 51 81 39 1 Food is only one of the needs of the family. The catastro- 8 83 39 1 Bias can confuse interpretation if the groups which are 51 83 39 1 phe brought on by terminal illness in one or both caring 8 84 39 1 compared are not similar. The severity of case mix has 51 84 39 1 adults is economic not famine. A more direct help would 8 86 39 1 been compared using discriminant analysis of the 51 86 39 1 be the replacement of lost income. It is money in an urban 8 87 39 1 presence and severity of presenting symptoms. The before 51 87 39 1 area which is wanted, and not just food. Money is easier 8 89 39 1 and after food groups have similar case mix. Their BMIs 51 89 39 1 to distribute and replaces the actual loss experienced by 8 90 39 1 are similar. The main difference is the preponderance of 51 90 39 1 such a family. The WFP has been reviewing the place of 83 95 7 1 Page 7 of 8 70 97 20 0 (page number not for citation purposes)

8 5 19 1 BMC Medical Ethics 2005, 6:3 61 5 29 1 http://www.biomedcentral.com/1472-6939/6/3 8 10 80 1 Table 1: Summary of the ethical issues concerning cardiac report cards (adapted from Nast S, Richard SA, Martin DK. Ethical issues 8 11 19 1 related to cardiac report cards. 28 11 24 1 Cardiol. Mar 1 2004;20(3):325–328.) Can J 12 14 7 1 Ethical issue 53 14 7 1 Description 14 17 3 1 Quality 24 17 44 1 •Quality operationalizes the ethical principles of beneficence and non-maleficence 24 18 51 1 •Report cards may improve quality of care through external pressure from an informed public 24 19 49 1 •Report cards may impede improvements to quality by generating anger and defensiveness 11 22 10 1 Informed Consent 24 22 37 1 •Informed consent operationalizes the ethical principle of autonomy 24 23 52 1 •To make health care decisions, patients need and want information about their medical options 24 25 52 1 •Report cards have the potential to provide this information and thus facilitate informed consent 14 27 3 1 Equity 24 27 28 1 •Equity operationalizes the ethical principle of justice 24 29 52 1 •Health equity between regions is an important consideration in publicly funded health systems 24 30 46 1 •Report cards must address policy makers to affect regional inequities in health care 13 33 5 1 Legitimacy 24 33 60 1 •Legitimacy operationalizes the ethical principle of justice, in this case deliberative forms of democratic justice 24 34 47 1 •Report card authors must ensure that report cards are and perceived to be legitimate 24 35 57 1 •The legitimacy of report cards will depend on their ability to meet stakeholders' reasonable expectations 8 43 39 1 Health care report cards are most well developed in car- 51 43 39 1 phi method with a panel of stakeholders to synthesize 8 44 39 1 diac care. An ethical framework would provide necessary 51 44 27 1 these insights into an ethical framework. 8 46 39 1 guidance for those generating cardiac report cards (CRCs) 8 47 39 1 and may help them avoid a number of difficult issues 51 47 39 1 The Delphi method allows a panel of stakeholders to gen- 8 49 39 1 associated with existing report cards. 'Gaming' and uncer- 51 49 39 1 erate ideas on a given topic and to reach a consensus on 8 50 39 1 tainty about the quality of report card data have been cited 51 50 39 1 the relative importance of those ideas [13,14]. This study 8 52 39 1 as impediments to reliable outcomes measures and a rea- 51 52 39 1 used a three-round modified Delphi method to identify 8 53 39 1 son to limit the public release of report card data [5,6]. 51 53 39 1 and reach agreement on the elements of an ethical frame- 8 55 39 1 Uncertainty also exists as to whether report cards have 51 55 10 1 work for CRCs. 8 56 39 1 empowered patients and/or improved health care qual- 8 58 60 1 Participants and sampling ity[7-10]. Other ethical and practical issues, such as bal- 8 59 81 1 Participants were selected from two previous studies con- ancing the public's desire for provider-specific outcomes 8 61 81 1 ducted by this research team. The first study described the measures with cardiac care providers' desires to limit the 8 62 81 1 views of cardiac care administrators, cardiac surgeons, car- amount and type of information released to the public, 8 64 81 1 diac nurses, cardiac patients, cardiologists, members of affect the content and legitimacy of CRCs. 51 65 39 1 the media, and outcomes researchers about CRCs [12]. 8 67 81 1 The second study described the views of cardiac patients An ethical framework can identify points of ethical con- 51 68 1 1 (a 54 68 3 1 paper 59 68 1 1 on 62 68 2 1 this 66 68 3 1 study 71 68 2 1 has 75 68 3 1 been 80 68 7 1 submitted 8 68 81 1 for cern 13 68 1 1 for 16 68 9 1 practitioners, 26 68 5 1 patients, 33 68 4 1 policy 39 68 4 1 makers 45 68 2 1 and 8 70 51 1 publication). researchers. And it can aid in the development, imple- 8 71 39 1 mentation and improvement of future generations of 8 73 81 1 We selected participants for our Delphi panel so that the CRCs. 51 74 39 1 views and interests of each stakeholder group were repre- 8 76 81 1 sented. Our final panel consisted of 13 panelists: 5 cardiac The purpose of this study is to develop an ethical frame- 8 77 81 1 patients, 2 administrators, 2 cardiac nurses, 2 cardiolo- work for CRCs. 51 78 39 1 gists, 1 member of the media, and 1 outcomes researcher. 8 80 7 1 Methods 51 80 39 1 We included a critical mass of patients in order to balance 8 81 4 1 Design 51 81 39 1 potential or perceived power differentials and enhance 8 83 39 1 Forming this ethical framework has been a three step 51 83 28 1 the comfort level of participating patients. 8 84 39 1 process. First, we analyzed the relevant ethical issues in an 8 86 61 1 Data collection and analysis earlier article [11]. A summary of these issues is presented 8 87 81 1 The Delphi process consisted of three rounds. Round 1 in table 1. Next we described stakeholders' views in two 8 89 81 1 was conducted using electronic communication. We pro- previous papers [12] (a paper on patients' views has been 8 90 81 1 vided three papers based on previous research to each submitted for publication). Finally, this study used a Del- 83 95 7 1 Page 2 of 7 70 97 20 0 (page number not for citation purposes)

8 5 17 1 BMC Genomics 2005, 6:39 60 5 30 1 http://www.biomedcentral.com/1471-2164/6/39 8 10 48 1 Results bacterial growth [5]. In contrast, most organisms contain 8 12 53 1 The Rpf domain several rpf-like genes, whose products are functionally 8 13 81 1 Bacterial genome sequencing projects have uncovered redundant [3,6-8]. All the proteins so far tested show 51 15 26 1 many genes whose products share with 8 15 81 1 Rpf a ca. cross-species activity in bioassays using laboratory cul- 78 15 6 1 M. luteus 8 16 81 1 70-residue segment that we have called the Rpf domain. tures of several different organisms, including 41 16 6 1 luteus, M. 51 18 10 1 This segment of 17 18 73 1 Rpf is both necessary and suffi- rhodochrous, 36 18 7 1 tuberculosis, 8 18 59 1 M. luteus Rhodococcus 25 18 9 1 Mycobacterium 44 18 3 1 Myco- 8 19 81 1 cient for biological activity, indicating that it corresponds (BCG) and bacterium 16 19 3 1 bovis 30 19 9 1 Mycobacterium 41 19 6 1 smegmatis 8 21 81 1 to a functional protein domain [5]. The Rpf-like proteins [4,7,9,10]. Since they are active at minute concentrations, 8 22 81 1 appear to be restricted to several genera within the actino- it was suggested that they might be involved in inter-cel- 8 24 75 1 bacteria, including Corynebacterium, Micrococcus, lular signalling [1,3,4]. 84 24 6 1 Mycobac- 51 25 4 1 terium, 56 25 34 1 and Streptomyces, but they appear Saccharopolyspora 8 27 39 1 Rpf-like proteins are not found in firmicutes (low G+C 51 27 39 1 to be absent from some others, such as Bifidobacterium, 8 28 39 1 Gram-positive bacteria), although some distantly related 60 28 2 1 and 51 28 39 1 (Table 1). An alignment of Thermobifida 63 28 7 1 Tropheryma 8 30 15 1 proteins are found in 34 30 2 1 and 45 30 2 1 (see 24 30 66 1 44 Rpf-like domains revealed that a central region of Staphylococcus 37 30 7 1 Oenococcus 8 31 39 1 below). In this article we report the results of comparative 51 31 39 1 between 6 and 9 residues accounts for almost all of the 8 32 39 1 genomic and domain analyses indicating that the firmi- 51 32 39 1 observed variation in length of this domain (see addi- 8 34 39 1 cutes contain a cognate protein family related to the actin- 51 34 39 1 tional data file 1). SignalP [11] and TMHMM [12] predic- 8 35 39 1 obacterial Rpf proteins by a process of "non-orthologous 51 35 39 1 tions suggest that all of the Rpf-like gene products so far 8 37 39 1 domain displacement". The available evidence strongly 51 37 39 1 uncovered are either secreted, or membrane-associated, 8 38 39 1 suggests that both the firmicute and actinobacterial pro- 51 38 39 1 with the exception of one instance of an Rpf-like domain 8 40 39 1 teins have a catalytic function, which may be responsible 51 40 39 1 within a mycobacteriophage tape measure protein [13]. 8 41 39 1 for their observed activity in improving the culturability of 51 41 39 1 The Rpf domain also contains two highly conserved 8 43 23 1 the organisms that produce them. 51 43 39 1 cysteine residues. Modelling has suggested that they lie in 51 44 39 1 close proximity and may form a disulphide bridge (A. 51 46 27 1 Murzin, personal communication) [14]. 8 49 27 1 Table 1: Organisms containing rpf-like genes 9 51 31 1 Part A: genes encoding proteins containing a Rpf domain 9 54 5 1 Organism 29 54 10 1 Genome size (Mb) 50 54 7 1 No. of genes 70 54 15 1 Genome Accession Number 29 57 1 1 2.5 50 57 0 1 3 9 57 68 1 NC_002935 Corynebacterium diphtheriae 29 58 1 1 3.3 50 58 0 1 2 9 58 68 1 NC_003450 Corynebavterium glutamicum 29 59 1 1 3.1 50 59 0 1 2 9 59 68 1 NC_004369 Corynebacterium efficiens 29 61 1 1 2.3 50 61 0 1 1 70 61 6 1 Mukamolova 9 61 74 1 al, 1998 Micrococcus luteus 78 61 0 1 et 29 62 1 1 4.7 50 62 0 1 4 9 62 68 1 NC_002944 Mycobacterium avium 29 63 1 1 4.3 50 63 0 1 5 9 63 68 1 NC_002945 Mycobacterium bovis 29 64 1 1 3.3 50 64 0 1 3 9 64 68 1 NC_002677 Mycobacterium leprae 29 66 1 1 6.5 50 66 0 1 4 9 66 74 1 NC_004506 (unfinished) Mycobacterium marinum 29 67 1 1 7.0 50 67 0 1 4 9 67 74 1 NC_002974 (unfinished) Mycobacterium smegmatis 23 68 3 1 H37Rv 29 68 1 1 4.4 50 68 0 1 5 9 68 68 1 NC_000962 Mycobacterium tuberculosis 29 70 2 1 8. 7 50 70 0 1 5 9 70 68 1 NC_003888 Streptomyces coelicolor 29 71 1 1 9.0 50 71 0 1 6 9 71 68 1 NC_003155 Streptomyces avermitilis 9 74 48 1 Part B: genes encoding proteins containing a domain distantly related to the Rpf domain 21 77 5 1 NCC2705 29 77 1 1 2.3 50 77 0 1 3 9 77 68 1 NC_004307 Bifidobacterium longum 20 78 6 1 strain Twist 29 78 1 1 0.9 50 78 0 1 2 9 78 68 1 NC_004572 Tropheryma whipplei 29 79 2 1 8. 7 50 79 0 1 2 9 79 68 1 NC_003888 Streptomyces coelicolor 29 80 1 1 9.0 50 80 0 1 3 9 80 68 1 NC_003155 Streptomyces avermitilis 29 82 0 1 - 50 82 0 1 2 9 82 61 1 - Staphylococcus carnosus 21 83 3 1 N315 29 83 1 1 2.8 50 83 0 1 1 9 83 68 1 NC_002745 Staphylococcus aureus 29 84 1 1 2.6 50 84 0 1 1 9 84 68 1 NC_004461 Staphylococcus epidermidis 29 86 1 1 0.3 50 86 0 1 1 9 86 72 1 NZ_AABJ02000001 Oenococcus oeni 14 88 1 1 and 9 88 30 1 genomes are not yet sequenced M. luteus 16 88 5 1 S. carnosus 14 90 19 1 genome size taken from Murayama 9 90 29 1 al. [78] M. luteus 34 90 0 1 et 82 95 8 1 Page 2 of 14 70 97 20 0 (page number not for citation purposes)

8 5 17 1 BMC Genomics 2005, 6:39 60 5 30 1 http://www.biomedcentral.com/1471-2164/6/39 10 13 9 1 S. coelicolor 49 13 2 0 tatD 41 13 37 1 ksgA sco3152 62 13 2 0 rpfB 11 17 10 1 C. glutamicum 49 17 2 0 tatD 34 17 41 1 ksgA cgl0883 61 17 2 0 rpfB 11 20 11 1 M. tuberculosis 35 21 2 0 metS 48 21 2 0 tatD 61 21 13 0 ksgA rpfB 11 24 7 1 B.subtilis 33 24 2 0 metS 49 24 2 0 tatD 59 24 2 0 yabE 74 24 2 0 rnmV 82 24 2 0 ksgA 11 28 10 1 B. halodurans 31 28 2 0 metS 74 28 2 0 rnmV 48 28 36 0 ksgA tatD 59 28 2 0 yabE 11 31 10 1 C. perfringens 59 31 2 0 yabE 71 31 2 0 rnmV 35 31 46 1 ksgA cpe2521 48 31 2 0 tatD 11 35 7 1 L. innocua 47 35 2 0 tatD 32 35 44 1 rnmV lin0223 59 35 2 0 yabE 81 35 2 0 ksgA 8 38 17 1 Figure 2 Genomic context of some 29 38 2 1 and 26 38 12 1 genes rpfB 32 38 2 1 yabE 8 40 19 1 Genomic context of some 32 40 2 1 and 28 40 15 1 genes. rpfB 35 40 3 1 yabE 44 40 14 1 The sco3152, cgl0883, 64 40 2 1 and 58 40 29 1 genes represented by an cpe2521 67 40 4 1 lin0223 8 41 78 1 empty arrow are hypothetical proteins unrelated to each other. See the text for the designations of the remaining genes. 8 49 27 1 Table 2: Organisms containing sps-like genes 9 51 31 1 Part A: genes encoding proteins containing a Sps domain 9 54 5 1 Organism 29 54 10 1 Genome size (Mb) 50 54 7 1 No. of genes 70 54 15 1 Genome Accession Number 18 57 7 1 strain A2012 29 57 1 1 5.1 50 57 0 1 5 9 57 68 1 NC_003995 Bacillus anthracis 18 58 6 1 strain Ames 29 58 1 1 5.2 50 58 0 1 6 9 58 68 1 NC_003997 Bacillus anthracis 17 59 7 1 ATCC 10987 29 59 1 1 5.2 50 59 0 1 6 9 59 68 1 NC_003939 Bacillus cereus 17 61 7 1 ATCC 14579 29 61 1 1 5.4 50 61 0 1 5 9 61 68 1 NC_004722 Bacillus cereus 29 62 1 1 4.2 50 62 0 1 3 9 62 67 1 NC-002570 Bacillus halodurans 29 63 1 1 4.2 50 63 0 1 4 9 63 68 1 NC_000964 Bacillus subtilis 29 64 1 1 3.6 50 64 0 1 4 9 64 68 1 NC_004193 Oceanobacillus iheyensis 29 66 1 1 3.0 50 66 0 1 2 9 66 68 1 NC_003212 Listeria innocua 21 67 3 1 EGD-e 29 67 1 1 2.9 50 67 0 1 2 9 67 68 1 NC_003210 Listeria monocytogenes 20 68 2 1 V583 29 68 1 1 3.2 50 68 0 1 1 9 68 68 1 NC_004668 Enterococcus faecalis 18 70 3 1 subsp 29 70 1 1 2.4 50 70 0 1 1 9 70 68 1 NC_002662 Lactococcus lactis 22 70 2 1 lactis 29 71 1 1 3.9 50 71 0 1 2 9 71 68 1 NC_003030 Clostridium acetobutylicum 20 72 0 1 A 29 72 1 1 3.9 50 72 0 1 2 9 72 74 1 NC_003223 (unfinished) Clostridium botulinum 21 74 3 1 str 13 29 74 1 1 3.0 50 74 0 1 3 9 74 68 1 NC_003366 Clostridium perfringens 18 75 2 1 E88 29 75 1 1 2.8 50 75 0 1 2 9 75 68 1 NC_004557 Clostridium tetani 29 76 1 1 3.7 50 76 0 1 4 9 76 70 1 AABG03000000 Clostridium thermocellum 29 77 1 1 4.9 50 77 0 1 1 9 77 70 1 AAAW00000000 Desulfitobacterium hafniense 29 79 1 1 2.7 50 79 0 1 1 9 79 68 1 NC_003869 Thermoanaerobacter tengcongensis 9 80 6 1 SPβc2 Phage 29 80 1 1 0.1 50 80 0 1 1 70 80 6 1 NC_001884 9 83 48 1 Part B: genes encoding proteins containing a domain distantly related to the Sps domain 29 86 1 1 3.6 50 86 0 1 1 9 86 68 1 NC_004193 Oceanobacillus iheyensis 29 87 1 1 3.1 50 87 0 1 1 9 87 75 1 NC_001263, NC_001264 Deinococcus radiodurans 29 88 1 1 1.9 50 88 0 1 1 9 88 68 1 NC_000853 Thermotoga maritima 82 95 8 1 Page 5 of 14 70 97 20 0 (page number not for citation purposes)

8 5 16 1 BMC Pediatrics 2005, 5:4 61 5 29 1 http://www.biomedcentral.com/1471-2431/5/4 8 10 28 1 Table 1: Details of tests evaluated in the review 9 12 2 1 Test 29 12 4 1 Details 50 12 7 1 Advantages 70 12 9 1 Disadvantages 9 15 9 1 Urine sampling 29 18 14 1 Needle attached to syringe 50 18 14 1 Least risk of contamination 9 18 65 1 Invasive Suprapubic aspiration (SPA) 29 19 18 1 inserted through lower abdomen 29 20 6 1 into bladder. 29 22 16 1 Catheter inserted through the 50 22 12 1 Less invasive than SPA 9 22 79 1 Invasive, causes pain and distress Transurethral catheterisation 29 23 13 1 urethra into the bladder. 70 23 4 1 to child 29 24 16 1 Midstream sample collected in 50 24 15 1 Non-invasive, easy to obtain 9 24 76 1 Difficult in younger children Clean voided urine (CVU) 29 25 9 1 sterile container. 29 27 13 1 Bag applied to perineum. 50 27 16 1 Suitable for babies and infants 9 27 73 1 Risk of contamination Urine bags 9 28 37 1 Absorbent pad placed in nappy. Urine pads 9 31 5 1 Dipstick 9 34 3 1 Nitrite 29 34 16 1 Gram-negative bacteria reduce 50 34 17 1 Very easy and quick to perform, 70 34 14 1 Less accurate than culture 29 35 13 1 dietary nitrate to nitrites. 50 35 8 1 relatively cheap 29 36 17 1 Leukocyte esterase is an enzyme 9 36 78 1 Not commercially available, not Leukocyte esterase (LE) Glucose 29 37 15 1 that suggests the presence of 70 37 16 1 suitable for non-potty trained 29 38 18 1 leukocytes. Normal urine contains 70 38 4 1 children 29 40 18 1 small amount of glucose. Bacteria 29 41 18 1 metabolise glucose and so this test 29 42 17 1 tests for the absence of glucose. 29 43 16 1 Requires morning fasting urine 29 44 5 1 specimen. 9 47 7 1 Microscopy 9 50 3 1 Pyuria 29 50 13 1 Urine examined through 50 50 11 1 Quicker than culture 70 50 14 1 More time consuming than 29 51 18 1 microscope for presence of white 70 51 16 1 dipstick, more expensive than 29 52 15 1 blood cells. Samples may be 70 52 10 1 dipstick and culture 29 53 17 1 centrifuged before examination 9 55 37 1 Urine examined for presence of Bacteriuria 29 56 4 1 bacteria. 29 57 15 1 Urine may be Gram-stained. 9 60 4 1 Culture 29 63 17 1 Reference standard test for UTI. 50 63 7 1 Very accurate 9 63 80 1 Time consuming: takes 48 hours to Standard Culture 29 64 14 1 Involves streaking urine on 70 64 18 1 give a result, has to be performed 29 65 17 1 enrichment and selective media. 70 65 9 1 in the laboratory 8 73 39 1 Clinical history and examination is the first step in any 51 73 39 1 Analytes commonly tested by dipsticks include leukocyte 8 74 39 1 diagnosis and is the means of identifying children with 51 74 39 1 esterase, nitrite, blood and protein[4]. Dipstick tests have 8 75 39 1 suspected UTI. Elements of the clinical examination have 51 76 39 1 the advantage of being quick and easy to perform and can 8 77 39 1 also been evaluated as diagnostic tests for UTI but there is 51 77 39 1 be carried out in primary care giving an immediate result. 8 78 39 1 little data available on these. Urine tests are commonly 51 78 39 1 Microscopic examination of urine samples for leukocytes 8 80 20 1 used for the diagnosis of UTI. 51 80 39 1 or bacteria [4] is considerably more time consuming and 51 81 39 1 labour intensive than the dipstick method[5]. However, 8 83 39 1 The reference standard for the diagnosis of UTI in children 51 83 39 1 unlike culture, it can be used to give results within the pri- 8 84 39 1 is considered to be any bacterial growth on a culture of 51 84 39 1 mary care setting. An uncontaminated sample is necessary 8 86 39 1 urine obtained by suprapubic aspiration[4]. Culture has 51 86 39 1 to reach an accurate diagnosis. Obtaining this is a particu- 8 87 39 1 the disadvantage of taking at least 48 hours to give a 51 87 39 1 lar issue when investigating young children. Table 1 8 89 39 1 result. More rapid methods of UTI diagnosis are therefore 51 89 39 1 presents a summary of the advantages and disadvantages 8 90 39 1 desirable. The most widely used rapid tests are dipsticks. 51 90 9 1 of these tests. 82 95 8 1 Page 2 of 13 70 97 20 0 (page number not for citation purposes)

8 5 16 1 BMC Pediatrics 2005, 5:4 61 5 29 1 http://www.biomedcentral.com/1471-2431/5/4 23 12 8 1 Withdrawals 83 12 2 1 Yes 83 14 2 1 No 16 15 15 1 Uninterpretable results 83 16 7 1 Not stated 18 17 13 1 Clinical review bias 16 20 15 1 Diagnostic review bias 20 22 11 1 Test review bias 13 25 18 1 Reference execution details 17 28 14 1 Test execution details 19 30 12 1 Incorporation bias 13 33 18 1 Differential verification bias 16 35 15 1 Partial verification bias 14 38 16 1 Disease progression bias 10 41 21 1 Appropriate reference standard 20 43 11 1 Selection criteria 16 46 15 1 Spectrum composition 32 49 2 1 0% 40 49 3 1 20% 49 49 3 1 40% 58 49 3 1 60% 67 49 3 1 80% 75 49 4 1 100% 8 52 21 1 Figure 2 Results of the quality assessment 8 53 21 1 Results of the quality assessment. 8 59 33 1 Table 2: Summary of results for studies of dipstick tests 9 61 12 1 Dipstick positive for: 24 61 11 1 Number of studies 38 61 8 1 Range in LR+ 47 61 13 1 Pooled LR+ (95 % CI)* 63 61 7 1 Range in LR- 72 61 13 1 Pooled LR- (95 % CI)* 9 64 3 1 Nitrite 30 64 1 1 23 39 64 6 1 2.5 – 439.6 50 64 8 1 15.9 (10.7, 23.7) 64 64 6 1 0.12 – 0.86 75 64 8 1 0.51 (0.43, 0.60) 9 66 1 1 LE 30 66 1 1 14 39 66 5 1 2.6 – 32.2 51 66 6 1 5.5 (4.1, 7.3) 64 66 6 1 0.02 – 0.66 75 66 8 1 0.26 (0.18, 0.36) 9 67 7 1 Nitrite or LE 30 67 1 1 15 39 67 5 1 3.0 – 32.2 51 67 6 1 6.1 (4.3, 8.6) 64 67 6 1 0.03 – 0.39 75 67 8 1 0.20 (0.16, 0.26) 9 68 7 1 Nitrite and LE 30 68 0 1 9 39 68 6 1 6.3 – 197.1 50 68 8 1 28.2 (17.3–46.0) 64 68 6 1 0.07 – 0.86 75 68 8 1 0.37 (0.26, 0.52) 9 69 4 1 Glucose 30 69 0 1 4 38 69 6 1 25.2 – 156.1 50 69 9 1 66.3 (20.0, 219.6) 64 69 6 1 0.02 – 0.38 75 69 8 1 0.07 (0.01, 0.83) 9 71 4 1 Protein 30 71 0 1 2 39 71 4 1 1.7 & 1.8 54 71 1 1 na 63 71 6 1 0.78 & 0.96 79 71 1 1 na 9 72 3 1 Blood 30 72 0 1 1 41 72 1 1 2.3 54 72 1 1 na 65 72 2 1 0.84 79 72 1 1 na 9 73 8 1 LE and protein 30 73 0 1 1 41 73 2 1 17.4 54 73 1 1 na 65 73 2 1 0.12 79 73 1 1 na 9 75 9 1 Nitrite, blood, or 30 75 0 1 1 41 75 1 1 2.7 54 75 1 1 na 65 75 2 1 0.28 79 75 1 1 na 9 76 4 1 protein 9 77 11 1 Nitrite, blood, or LE 30 77 0 1 1 41 77 1 1 1.3 54 77 1 1 na 65 77 2 1 0.50 79 77 1 1 na 9 78 11 1 Nitite, blood and LE 30 78 0 1 1 41 78 1 1 3.5 54 78 1 1 na 65 78 2 1 0.19 79 78 1 1 na 9 80 12 1 Nitrite, LE and protein 30 80 0 1 2 39 80 5 1 3.1 & 69.2 54 80 1 1 na 63 80 6 1 0.05 & 0.17 79 80 1 1 na 9 81 12 1 Nitrite, LE, or protein 30 81 0 1 1 41 81 1 1 1.9 54 81 1 1 na 65 81 2 1 0.05 79 81 1 1 na 9 84 12 1 Nitrite, LE, protein, or 30 84 0 1 1 41 84 1 1 8.0 54 84 1 1 na 65 84 2 1 0.19 79 84 1 1 na 9 85 3 1 blood 9 88 61 1 * There was significant heterogeneity in all pooled estimates therefore these should be interpreted with caution 82 95 8 1 Page 5 of 13 70 97 20 0 (page number not for citation purposes)