TEXT
stringlengths 10
7.37k
| ID
int64 0
824
|
|---|---|
Microcalcifications 5–69 93–100
Cystic aspect 10–34 91–100
Peripheral vascularity 40–86 57–93
Hyperechogenicity 30–87 43–95
Round shape 37 70 | 193 |
aAdapted with permission from the European Thyroid Associa- tion guidelines for cervical ultrasound (20). | 194 |
cancer risk. However, given the higher pretest likelihood of thyroid cancer associated with these clinical risk factors, FNA can be considered at lower size cutoffs for all of the sonographic appearances described above.
Ultrasound elastography (USE) has similarly been investi- gated for its ability to modify thyroid cancer risk assessment among clinically relevant thyroid nodules. Elastography is a measurement of tissue stiffness. Performance requires an US machine, as well as an elastography computational module that most often must be purchased separately. An initial pro- spective study of 92 selected, nonrandomized patients sug- gested positive and negative predictive values (NPVs) near 100% (97). However, more recently, larger trials have reported substantially different results. Moon and colleagues retro- spectively studied 703 thyroid nodules in comparison to gray- scale US (78). Performance of USE was inferior to that of gray-scale US assessment. The largest prospective study of 706 patients with 912 thyroid nodules was recently published by Azizi et al. (98). In this study, the positive predictive value (PPV) of USE was only 36%, comparable to that of micro- calcifications. The NPV of USE was 97% in a population with cancer prevalence of 9%. Thus, while USE holds promise as a means by which to noninvasively assess cancer risk, its per- formance is highly variable and operator dependent. Perhaps most importantly, USE can only be effectively applied to solid nodules, thus excluding its utility for cystic or partially cystic nodules. Furthermore, to be amenable to direct pressure and determination of tissue strain, the index nodule must not
gray-scale or Doppler US examination. The committee therefore believes USE (when available) may prove to be a helpful tool for preoperative risk assessment in those patients in whom accurate assessment can be performed. However, the committee cannot presently recommend its universal use or widespread adoption. Importantly, the ability to perform (or not perform) USE should not modify the recommendation for traditional gray-scale sonographic evaluation.
Finally, while most thyroid nodules meeting the preceding sonographic patterns and sizes should undergo FNA, we acknowledge that a conservative approach of active surveil- lance management may be appropriate as an alternative to FNA in selected patients. These may include patients with very low-risk tumors (e.g., no clinical or radiographic evi- dence of invasion or metastases), patients at high surgical risk, or those with a relatively short life span expectancy in whom the benefits of intervention may be unrealized. | 195 |
[A11] What is the role of FNA, cytology interpretation, and molecular testing in patients with thyroid nodules? | 196 |
RECOMMENDATION 9
Thyroid nodule FNA cytology should be reported using diagnostic groups outlined in the Bethesda System for Reporting Thyroid Cytopathology.
(Strong recommendation, Moderate-quality evidence) | 197 |
To address a significant variability in the reporting of cy- tological findings in thyroid FNA samples, the 2007 National Cancer Institute Thyroid Fine-Needle Aspiration State of the Science Conference provided consensus recommenda- tions known as the Bethesda System for Reporting Thyroid Cytopathology (99,100). The Bethesda system recognizes six diagnostic categories and provides an estimation of cancer risk within each category based upon literature review and expert | 198 |
Diagnostic category
Table 8. The Bethesda System for Reporting Thyroid Cytopathology: Diagnostic Categories and Risk of Malignancya
Estimated/predicted risk of malignancy by the Bethesda system, %a | 202 |
Actual risk of malignancy in nodules surgically excised,
% median (range)b | 204 |
Nondiagnostic or unsatisfactory 1–4 20 (9–32)
Benign 0–3 2.5 (1–10) | 206 |
Atypia of undetermined significance or follicular lesion of undetermined significance
5–15 14 (6–48) | 207 |
aAs reported in The Bethesda System by Cibas and Ali (1076).
bBased on the meta-analysis of eight studies reported by Bongiovanni et al. (103). The risk was calculated based on the portion of nodules in each diagnostic category that underwent surgical excision and likely is not representative of the entire population, particularly of nondiagnostic and benign diagnostic categories. | 208 |
opinion (Fig. 1, Table 8). These categories are (i) nondiagnostic/ unsatisfactory; (ii) benign; (iii) atypia of undetermined signifi- cance/follicular lesion of undetermined significance (AUS/ FLUS); (iv) follicular neoplasm/suspicious for follicular neo- plasm (FN/SFN), a category that also encompasses the diag- nosis of Hu¨rthle cell neoplasm/suspicious for Hu¨rthle cell neoplasm; (v) suspicious for malignancy (SUSP), and (vi) malignant. Recent studies that applied the criteria and termi- nology of the Bethesda System to a large series of patients have shown a relatively good concordance in reporting FNA cytol- ogy, with 89%–95% of samples being satisfactory for inter-
pretation and 55%–74% reported as definitively benign and
2%–5% as definitively malignant (101–104). The remaining samples are cytologically indeterminate, including AUS/FLUS
in 2%–18% of nodules, FN in 2%–25%, and SUSP in 1%–6%. In these studies, the probability of malignancy for each Be- thesda category demonstrated significant variability, but was overall compatible with the range predicted by the Bethesda System, with the exception of the AUS/FLUS diagnosis, for which the risk of malignant outcome in some studies was sig- nificantly higher than predicted (Table 8) (103,105). Recently, a blinded prospective evaluation of inter-observer concordance using Bethesda classification was performed. These data con- firm an inherent limitation to the reproducibility of interpreting any cytology specimen (106). Specimens diagnosed as AUS/ FLUS and SUSP were associated with the highest discordance rates. Some studies suggest that the AUS/FLUS category should be further subdivided into AUS with cytologic atypia (higher risk for malignancy) and FLUS with architectural atypia (lower risk for malignancy), but this has not yet been widely adopted (107). Nonetheless, classification using the Bethesda system has proven highly beneficial, allowing practitioners to speak with the same terminology and better convey malignant risk. The risk of malignancy in each of the six diagnostic categories should be independently defined at each cytology center or institution to guide clinicians on risk estimates and help choose appropriate molecular testing for patients with indeterminate cytology. | 210 |
[A12] Nondiagnostic cytology | 211 |
RECOMMENDATION 10
For a nodule with an initial nondiagnostic cytology result, FNA should be repeated with US guidance and, if available, on-site cytologic evaluation
(Strong recommendation, Moderate-quality evidence)
Repeatedly nondiagnostic nodules without a high suspicion sonographic pattern require close observation or
surgical excision for histopathologic diagnosis
(Weak recommendation, Low-quality evidence)
Surgery should be considered for histopathologic di- agnosis if the cytologically nondiagnostic nodule has a high suspicion sonographic pattern, growth of the nodule (>20% in two dimensions) is detected during US surveil- lance, or clinical risk factors for malignancy are present
(Weak recommendation, Low-quality evidence) | 212 |
Nondiagnostic or unsatisfactory FNA biopsies are those that fail to meet the established quantitative or qualitative
criteria for cytologic adequacy (i.e., the presence of at least
six groups of well-visualized follicular cells, each group
containing at least 10 well-preserved epithelial cells, pref-
erably on a single slide) (99,108). Although an FNA speci- men found to have abundant colloid and few epithelial cells may be considered nondiagnostic by the above criteria, this is also likely a benign biopsy. After an initial nondiagnostic cytology result, repeat FNA with US guidance and, if avail- able, on-site cytologic evaluation, will substantially increase the rate of specimen adequacy (109–113). It has been sug- gested that repeat FNA should be performed no sooner than 3 months after the initial FNA to prevent false-positive inter- pretation due to biopsy-induced reactive/reparative changes (114). Two recent studies have questioned the necessity for a 3-month waiting period after the first FNA because they did not find a correlation between the diagnostic yield/accuracy of repeated FNA and the waiting time between the procedures (115,116). A 3-month waiting period after a nondiagnostic biopsy is likely not necessary. If clinical and US features are suspicious for malignancy, a shorter waiting period may be appropriate. Repeat FNA with US guidance will yield a di- agnostic cytology specimen in 60%–80% of nodules, partic- ularly when the cystic component is <50% (64,112,117). Nodules with larger cystic portion have a higher chance to yield nondiagnostic samples on the initial and repeated FNA. Most nodules with a nondiagnostic cytology interpretation are benign. In large series of patients classified based on the
Bethesda System, nondiagnostic samples constituted 2%–16%
of all FNA samples, of which 7%–26% were eventually resected (101–103). The frequency of malignancy among all initially nondiagnostic samples was 2%–4% and among those non- diagnostic samples that were eventually resected was 9%–32%. Sonographic features are also useful for identifying which nodules with repeat nondiagnostic FNA cytology results are more likely to be malignant. Of 104 nodules with two non-
diagnostic cytology results, thyroid cancer was found in 25%
of those with microcalcifications, irregular margins, a taller
than wide shape, or hypoechogenicity, but in only 4% lacking
these features (118).
In some studies, the use of thyroid core-needle biopsy (119)
and molecular testing for BRAF (120,121) or a panel of muta-
tions (122) helped to facilitate appropriate management of these
patients, although the full clinical impact of these approaches for nodules with nondiagnostic cytology remains unknown. Some studies have found that core biopsy offers a higher ade- quacy rate, but may be less sensitive for the detection of pap- illary cancer (123,124). Mutational testing may be informative in samples considered inadequate by qualitative criteria (i.e., due to poor preparation or poor staining of cells) but is unlikely to be contributory in samples with insufficient quantity of cells. | 213 |
[A13] Benign cytology | 214 |
RECOMMENDATION 11
If the nodule is benign on cytology, further immediate diagnostic studies or treatment are not required
(Strong recommendation, High-quality evidence) | 215 |
Accurate FNA cytology diagnosis depends upon a num- ber of factors including the skill of the operator, FNA tech- nique, specimen preparation, and cytology interpretation. | 216 |
Ultrasound-guided FNA with real-time visualization of needle placement in the target nodule decreases the false-negative rate of a benign cytology diagnosis (68,69,126,128). Although pro- spective studies are lacking, malignancy rates of only 1%–2% have been reported in large retrospective series that analyzed the utility of systematic repeat FNA in nodules with prior benign cytology results (129–133). A pooled analysis of 12 studies by Tee et al. (134) showed that of 4055 patients with benign cy- tology who underwent surgery, the rate of malignancy was 3.2%. Studies have also attempted to correlate nodule size with accuracy of FNA cytology. Several surgical series have re- ported higher malignancy rates in nodules >3–4 cm, but these suffer from both selection bias (only a subset of patients un- derwent preoperative FNA) and potential sampling error (FNA performed by palpation) (135,136). A recent study evaluated the accuracy of FNA and US features in patients with thyroid nodules ‡4 cm (137). This was a single-center study in which the practice is to offer thyroidectomy or lobectomy to all pa- tients with nodules ‡4 cm. The investigators identified thyroid cancer in 22% of 382 nodules. A subset of thyroid nodules underwent preoperative FNA, and of the 125 cytologically benign nodules, 10.4% were malignant on final histopathology. The investigators further showed that individual US char- acteristics were not predictive for malignancy, although they did not look at sonographic patterns. Two other recent reports of consecutive US-guided FNA evaluations in over 1400 nodules >3 cm with initial benign cytology followed for a mean of 3 years confirmed a lower false-negative rate of <1.5% (138,139). Interestingly, in both these studies 66% of the missed cancers were found in nodules with high suspicion sonographic pattern, despite initial benign cytol- ogy. A recent retrospective study analyzed the long-term follow-up of 2010 cytologically benign nodules from 1369 patients. Over a mean follow-up of 8.5 years 18 false- negative malignancies were detected. However, no deaths attributable to thyroid cancer were identified in this cohort. These data confirm that an initially benign FNA confers negligible mortality risk during long-term follow-up despite a low but real risk of false negatives in this cytologic cate- gory (140). Based on the evidence, it is still unclear if pa- tients with thyroid nodules ‡4 cm and benign cytology carry a higher risk of malignancy and should be managed differ-
ently than those with smaller nodules.
Follow-up for patients with benign cytology is discussed in section [A24] and Recommendation 23. | 218 |
[A14] Malignant cytology | 219 |
RECOMMENDATION 12
If a cytology result is diagnostic for primary thyroid ma- lignancy, surgery is generally recommended.
(Strong recommendation, Moderate-quality evidence) | 220 |
A cytology diagnostic for a primary thyroid malignancy will almost always lead to thyroid surgery. However, an ac-
tive surveillance management approach can be considered as
an alternative to immediate surgery in
patients with very low risk tumors (e.g., papillary microcarcinomas without clinically evident metasta- ses or local invasion, and no convincing cytologic evidence of aggressive disease),
patients at high surgical risk because of comorbid conditions,
patients expected to have a relatively short remaining
life span (e.g., serious cardiopulmonary disease, other malignancies, very advanced age), or
patients with concurrent medical or surgical issues that need to be addressed prior to thyroid surgery.
Following thyroid surgery for papillary thyroid micro-
carcinoma (PTMC), defined as a tumor 1 cm or less in size, disease-specific mortality rates have been reported to be
<1%, loco-regional recurrence rates are 2%–6%, and distant recurrence rates are 1%–2% (141,142). It is quite likely that these excellent outcomes are more related to the indolent nature of the disease rather than to the effectiveness of treatment, since two prospective clinical studies of active
surveillance from Japan reported similar clinical outcomes in
1465 patients with biopsy-proven PTMC that were not sur-
gically removed and were followed for up to 15 years (av- erage 5–6 years, range 1–17 years) (95,143). In the study by
Ito et al. (95), observation was offered to 1235 patients with
PTMC that did not have (i) location adjacent to the trachea or on the dorsal surface of the lobe close to the recurrent la- ryngeal nerve, (ii) FNA findings suggestive of high-grade malignancy; (iii) presence of regional lymph node metasta- ses; or (iv) signs of progression during follow-up. Of those, most patients showed stable tumor size on average follow-up of 60 months (range 18–227 months), whereas 5% showed
tumor enlargement (>3 mm) by US on 5-year follow-up, and
8% on 10-year follow-up. Furthermore, 1.7% and 3.8% of patients at 5-year and 10-year follow-up showed evidence for lymph node metastases. Of 1235 patients, 191 underwent
surgical treatment after observation, including those with tumor enlargement and new lymph node metastases. These patients have been followed an average of 75 months (range 1–246 months) after the surgical intervention. Only one of the patients treated with surgery after observation developed tumor recurrence. In the study by Sugitani et al. (143), 230 patients with asymptomatic PTMC were followed for 5 years on average. Of those patients, tumor size enlargement was observed in 7%, and 1% developed apparent lymph node metastasis. Seven percent of patients underwent surgery after 1–12 years of follow-up, and no recurrences were identified in those on limited follow-up, suggesting that the delayed surgery did not affect the outcome. A more recent study by
Ito and colleagues followed 1235 patients with PTMC under active surveillance for an average of 60 months (95). Only 43 patients (3.5%) had clinical progression of disease by their stated criteria (tumor growing to >12 mm or appearance of new lymph node metastases). Interestingly, the younger pa-
tients (<40 years old) had an 8.9% rate of clinical progres-
sion, while those 40–60 years old had a 3.5% rate of progression and those
clinical progression (1.6%).
Despite the evidence that cautious observation is a safe and effective alternative to immediate surgical resection, very few PTMC patients outside of those two centers in Japan are given the option of an active surveillance approach. This is in part due to reports in the literature of a small percentage of patients with PTMC presenting with clinically significant regional or distant metastases (141,142,144). Unfortunately, no clinical features (145–151) can reliably differentiate the relatively small number | 221 |
of PTMC patients destined to develop clinically significant progression from the larger population of people that harbor indolent PTMCs that will not cause significant disease.
Similarly, well-known thyroid cancer oncogenes, such as
BRAF, when taken in isolation, are not able to specifically
identify the microcarcinomas that will progress and spread
outside of the thyroid. The prevalence of BRAFV600E muta- tions in PTMC with lymph node metastases and tumor re- currence is higher than PTMC without LN metastases or recurrence, and in some studies the presence of a BRAF mu- tation was associated with lymph node metastasis from PTMC on multivariate analysis (150,152,153). These studies showed that although the presence of a BRAFV600E mutation identifies 65%–77% of patients with PTMC that develop lymph node metastases, the BRAF status taken in isolation has a low PPV
for detecting PTMC with extrathyroidal spread and therefore has a limited role for guiding patient management. However, recent data suggest that specific molecular profiles, such as the coexistence of BRAF with other oncogenic mutations (such as
PIK3CA, AKT1), TERT promoter, or TP53 mutations may
serve as more specific markers of less favorable outcome of
PTC. Therefore, it is likely that finding of one of these genetic profiles in a small tumor would suggest that it represents an early stage of a clinically relevant PTC (154–157). Future studies are expected to establish the impact of molecular pro- filing involving multiple mutations or other genetic alterations on clinical management of patients with PTMC. | 223 |
[A15] Indeterminate cytology (AUS/FLUS, FN, SUSP)* | 224 |
[A16] What are the principles of the molecular testing of FNA samples?
Molecular markers may be classified according to intended use; that is, diagnostic (classification of a disease state), prognostic, or predictive purposes (providing information on the estimated probability of therapeutic benefit or harm of a specific therapy) (158). Furthermore, companion use of pre- dictive molecular markers involves the identification of patient subgroups in which a therapeutic intervention is proven to be either beneficial or harmful, with intended implications for appropriate clinical stratification of therapies (158). Validation studies of molecular marker tests may include examination of
analytic validity (including test accuracy and reproduc- ibility in ascertaining the molecular event), (b) clinical validity (the performance of the test in distinguishing different groups of patients, based on biology or expected disease outcome, including measures of sensitivity and specificity or predictive values), and (c) clinical utility (examination of the test’s ability to improve outcomes, with direct clinical decision-making implications) (158). Furthermore, an NCCN Tumor Marker Task Force has indicated that the clinical utility of a molecular test should be founded in strong evidence proving that use of the marker ‘‘improves patient outcomes sufficiently to justify its incorporation into routine clinical practice.’’{
The principal proposed use of molecular markers in
indeterminate thyroid FNA specimens is diagnostic
(ruling out or in the presence of thyroid malignancy), with the implication of a companion use to inform decision-making on primary surgical treatment (i.e., the decision to perform surgery and if so, the extent of surgery). However, the focus of this section is restricted to the clinical validity of molecular testing of indeter- minate FNA specimens. It is important to note that long- term outcome data on companion use of molecular marker status to guide therapeutic decision-making is currently lacking, and therefore we do not know if im- plementation of molecular marker use in routine clinical practice would result in a significant overall benefit in health outcomes in patients with thyroid nodules. Sur- gical decision-making on indeterminate FNA specimens is reviewed in another section of these guidelines, with some reference to molecular marker testing (if performed). As summarized in a Disease State Commentary from the AACE Thyroid Scientific Committee and a consensus statement from the ATA Surgical Affairs Committee, use of molecular marker testing on indeterminate FNA spec- imens should not be intended to replace other sources of information or clinical judgment (159,160). The pretest probability of malignancy (based on clinical risk factors, cytology, US findings), feasibility considerations, and patient preferences are some additional factors that need to be considered in decision-making related to molecular marker testing of FNA specimens. Because this is a rap- idly evolving field of investigation, it will be important to perform interval evaluations of the published evidence to ensure that recommendations remain contemporary.
A number of molecular approaches have been studied in the clinical setting of indeterminate FNA cytologic inter- pretation (161). One could surmise that an ideal ‘‘rule-in’’ test would have a PPV for histopathologically proven ma- lignancy similar to a malignant cytologic diagnosis (98.6%), and an ideal ‘‘rule-out’’ test would have a NPV similar to a benign cytologic diagnosis (96.3%) (predictive value esti- mates based on a recent meta-analysis of performance of the Bethesda system) (103), and these would hold true with a reasonable degree of precision and reproducibility. | 225 |
RECOMMENDATION 13
If molecular testing is being considered, patients should be
counseled regarding the potential benefits and limitations of testing and about the possible uncertainties in the therapeutic and long-term clinical implications of results.
(Strong recommendation, Low-quality evidence) | 226 |
The largest studies of preoperative molecular markers in patients with indeterminate FNA cytology have respectively evaluated a seven-gene panel of genetic mutations and re- arrangements (162), | 227 |
a gene expression classifier (167 GEC; mRNA expression
of 167 genes) (163), and galectin-3 immunohistochemistry | 228 |
*The final draft for the sections (A15–A19) and recommenda- tions (13–17) were revised and approved by a subgroup of seven members of the task force with no perceived conflicts or competing interests in this area.
{From the NCCN Biomarkers Compendium (www.nccn.org/ professionals/biomarkers/default.aspx).
(cell blocks) (164). These respective studies have been sub- ject to various degrees of blinding of outcome assessment (162–164).
Mutational testing has been proposed for use as a rule-in test because of relatively high reported specificity (86%–100%) | 229 |
and PPV (84%–100%) (105,122,162,165–168). Although
BRAFV600E single mutation testing has been estimated to have a specificity of approximately 99% (pooled data from 1117 nodules with histopathologic confirmation from multiple studies), the sensitivity has been deemed to be too low to reliably rule out the presence of malignancy (169). Therefore, mutational panels have been expanded to include multiple mutations/translocations including BRAF, NRAS, HRAS, and KRAS point mutations, as well as RET/PTC1 and RET/PTC3, with or without PAX8/PPARc rearrangements (105,122,162, 165–168) and other gene rearrangements (170). In indeter- minate cytology thyroid nodules, the sensitivity of the seven- gene mutational panel testing is variable, with reports ranging from 44% to 100% (162,165,167). The reported variability in sensitivity of mutational analysis with the seven-gene panel in indeterminate nodules suggests that traditional limited muta- tion panels may not reliably rule out malignancy with a neg- ative test in this population. Next-generation sequencing of an expanded panel of point mutations, single base insertions/de- letions (indels), and gene rearrangements has been reported to have a sensitivity of 90% for FN/SFN FNA cytology speci- mens from a single-center study (170). A limitation was that the pathologists evaluating the gold standard surgical pathol- ogy specimens were aware of the results of earlier generation molecular tests previously conducted on the FNAs (170). It is not known to what extent differences in techniques used to perform mutational testing by various groups (162,167,170) may affect test performance, and direct, head-to-head compar- isons of these tests within the same population are lacking. The currently available seven-gene mutational panels have been proposed to be most useful when surgery is favored. However, this is based on the assumption that the surgical approach would be altered with a positive test, and long-term outcome data proving the overall benefit of this therapeutic strategy are nee- ded. It is important to acknowledge that algorithms employing seven-gene mutational testing as a means to inform decision- making on extent of primary thyroid surgery (i.e., lobectomy or total thyroidectomy) (162) were developed at a time when the ATA guidelines favored total thyroidectomy for most PTCs
>1 cm in diameter (25). However, this does not reflect recom- mendations in these guidelines (see Recommendation 35 on surgical management of malignant cytology nodules). Fur- thermore, long-term outcome data from a strategy of using molecular markers in indeterminate FNA specimens to stratify surgical approach are currently lacking.
Use of a 167 GEC has been proposed as a rule-out test due to relatively high sensitivity (92%) and NPV (93%), as reported in a prospective multicenter study (163). The relatively low specificity of the 167 GEC test (mean values 48%–53% in indeterminate nodules subject to histopatho- logic confirmation) suggests that the test cannot defini-
tively rule-in malignancy in indeterminate nodules. In a retrospective analysis of 167 GEC results from five insti- tutions, Alexander et al. (171) reported that the prevalence of 167 GEC benign readings by institution varied up to 29%, which was not statistically significant. The distribu- tion of recruitment from each of the five study sites was highly variable (total n = 339 nodules), with two sites contributing only 30 and 37 patients, and the other three sites accounting for the majority of the study population. The prevalence of malignancy confirmed by histopathology in 167 GEC ‘‘suspicious’’ nodules ranged from 33% to 80% in
the 48 nodules in the AUS/FLUS group across institutions (no 95% confidence interval [95% CI] reported, p = 0.11), and from 33% to 67% in the 65 nodules in the FN cytology group (no 95% CI reported, p = 0.87) (171). For the 174 patients with 167 GEC ‘‘benign’’ readings, four patients were advised to undergo surgery (2%), and 41% (71 patients) of this group had short documented follow-up for a mean of 8.5 months (me- dian 8 months, range 1–24 months); ultimately, 6% of patients in this group (11/174) had surgery, with one histopatho- logically confirmed malignancy (171). The reproducibility of 167 GEC NPV measures in different populations of pa- tients with indeterminate cytology thyroid nodules has re- cently been questioned in three smaller, independent, unblinded studies (172–174). None of these studies reported any degree of blinding. Furthermore, 95% CI of predictive estimates in indeterminate cytology nodules were not re- ported in two single-center studies (172,173), and one multi-center study was reported to be a retrospective anal- ysis (174), making it difficult to interpret the findings. However, such data highlight the need for additional inde- pendent research examining the reproducibility of diag- nostic efficacy of the 167 GEC in more institutions, and the importance of reporting precision estimates for diagnostic accuracy measures. Furthermore, as in the case of other molecular-based diagnostic tests in the field, long-term outcome data from clinical utility studies are needed to in- form potential future clinical practice implications of the 167 GEC.
Immunohistochemical stains such as galectin-3 and HBME-1 have been examined in multiple studies of histo- logically confirmed thyroid FNA samples with indeterminate cytology, with reports of relatively high rates of specificity, but low sensitivity, for cancer detection (164,175,176). Two of these studies were reported to be prospective (164,175), but only one of the studies reported any degree of blinding (blind central histopathologic review) (164). Immuno- histochemical stains require the availability of a cell block to perform the staining. Additional diagnostic molecular marker strategies are also under development. Specifically, mRNA markers (177–179), as well as miRNA markers (180–185), have shown initial diagnostic utility in FNA samples with in- determinate cytological diagnoses, but they have not been thoroughly validated. A recent study combining seven-gene mutational testing with expression of a set of 10 miRNA genes on preoperative FNA sampling from 109 patients with AUS/ FLUS or FN cytology, showed 89% sensitivity, 85% speci- ficity, with a 73% PPV and 94% NPV on this group with a 32% prevalence of malignancy (186). Finally, peripheral blood TSH receptor mRNA assay has been reported to have a 90% PPV and 39% NPV in FNA-based assessment of thyroid nodules with atypical or suspicious cytology in a single-center, pro- spective validation study (187).
In summary, there is currently no single optimal molecular
test that can definitively rule in or rule out malignancy in all
cases of indeterminate cytology, and long-term outcome data
proving clinical utility are needed. | 231 |
RECOMMENDATION 14
If intended for clinical use, molecular testing should be per- formed in Clinical Laboratory Improvement Amendments/ College of American Pathologists (CLIA/CAP)-certified molecular laboratories, or the international equivalent, | 232 |
because reported quality assurance practices may be superior compared to other settings.
(Strong recommendation, Low-quality evidence) | 234 |
Many molecular marker tests are available in hospital-based molecular pathology laboratories and in reference laboratories. Importantly, all molecular marker tests intended for clinical use should be performed only in CLIA/CAP-certified molecular laboratories after appropriate analytical and clinical validation of all assays in each laboratory (158). In a survey of American molecular genetic testing laboratory directors, laboratory pro- cess quality assurance score (for multiple relevant domains) was associated with the presence of CLIA certification (188). In a large, international survey of medical genetic testing labo- ratory directors, accreditation of the laboratory was associated with a higher quality assurance index score (189). | 235 |
[A17] AUS/FLUS cytology | 236 |
RECOMMENDATION 15
For nodules with AUS/FLUS cytology, after consider- ation of worrisome clinical and sonographic features, in- vestigations such as repeat FNA or molecular testing may be used to supplement malignancy risk assessment in lieu of proceeding directly with a strategy of either surveillance or diagnostic surgery. Informed patient preference and feasi- bility should be considered in clinical decision-making.
(Weak recommendation, Moderate-quality evidence)
If repeat FNA cytology, molecular testing, or both are
not performed or inconclusive, either surveillance or di-
agnostic surgical excision may be performed for an AUS/ FLUS thyroid nodule, depending on clinical risk factors, sonographic pattern, and patient preference.
(Strong recommendation, Low-quality evidence) | 237 |
Based on the Bethesda System, this diagnostic category is reserved for specimens that contain cells with architectural and/or nuclear atypia that is more pronounced than expected for benign changes but not sufficient to be placed in one of the higher risk diagnostic categories (99,190). Although this diagnostic category has been recommended for limited use and has an expected frequency in the range of 7%, recent reports based on the Be- thesda System have found this cytologic diagnosis to be used in 1%–27% of all thyroid FNA samples (105,191). In studies that utilized the criteria established by the Bethesda System, the risk of cancer for patients with AUS/FLUS nodules who underwent surgery was 6%–48%, with a mean risk of 16% (191).
A second opinion review of the cytopathology slides by a
high-volume cytopathologist may be considered for patients
with AUS/FLUS cytology. There is some evidence that this approach may reclassify many of these patients into the be- nign and nondiagnostic categories (106,192). Furthermore, the overall diagnostic accuracy may be improved. Unfortunately, there is a relatively high intra-observer variability in this dif- ficult diagnostic category (106).
A repeat FNA yields a more definitive cytologic diagnosis in many cases, whereas 10%–30% of nodules are repeatedly AUS/FLUS (104,193–195). The rate of malignancy on sur- gical follow-up has been shown to be similar for patients with
a single AUS/FLUS diagnosis (37/90, 41%), two successive AUS/FLUS diagnoses (22/51, 43%), and patients with a be- nign cytologic interpretation following the initial AUS/FLUS diagnosis (2/7, 29%) (196). Use of thyroid core-needle bi- opsy was reported by some to be more informative than re- peated FNA for sampling nodules that were AUS/FLUS on initial FNA (119), and it is reasonably well-tolerated (197). The refinement of risk stratification of nodules with AUS/ FLUS cytology using molecular testing has been examined in multiple studies. The interpretation of molecular testing is complex, however, and its utility is strongly influenced by the prevalence of cancer in the tested population of nodules. Only two molecular tests have been separately evaluated and val- idated for the individual AUS/FLUS, FN, and SUSP cate- gories. Mutational testing for BRAF in AUS/FLUS samples
has high specificity for cancer, but low sensitivity (198,199). Testing for a panel of mutations (BRAF, NRAS, HRAS, KRAS, RET/PTC1, RET/PTC3, PAX8/PPARc) offers a significantly higher sensitivity of 63%–80% (162,165). In the largest prospective study to date of nodules with AUS/FLUS cytol- ogy (653 consecutive nodules, of which 247 had surgical follow-up) from a single institution, detection of any of these mutations using RT-PCR with fluorescent melting curve analysis was reported to convey an 88% risk of cancer among nodules with surgical follow-up; 63% of cancers on final histopathology were identified with a positive mutation pre- operatively (22 of 35), and 94% of nodules that were negative on mutation analysis had a benign final histopathology (209 of 222) (162). Positive testing for BRAF, RET/PTC or PAX8/ PPARc was specific for a malignant outcome in 100% of cases, whereas RAS mutations had an 84% risk of cancer and a 16% chance of benign follicular adenoma. A recent study utilizing RT-PCR with liquid bead array flow cytometry with the seven- gene mutation panel reported on 11 AUS/FLUS cytology nodules that had malignant histopathologic confirmation, of which seven had a negative molecular test and four had a positive test (167). There were also 11 AUS/FLUS cytology nodules, which were benign on histopathologic evaluation; 9 of 11 had a negative molecular test result and 2 of 11 had a positive result. Interpretation of results from the AUS/FLUS subgroup is limited by the small reported sample size (167).
Molecular testing using the 167 GEC has been studied for its diagnostic use in nodules with AUS/FLUS cytology. A multi-institutional study of 129 FNA samples with AUS/ FLUS cytology and surgical follow-up reported a 90% sen- sitivity [95% CI 74%–98%] and 95% NPV [95% CI 85%–
99%], but only a 53% specificity [95% CI 43%–63%] and
38% PPV [95% CI 27%–50%] for cancer (163). Although the
specificity of the 167 GEC was low (53% [95% CI 43%– 63%]), the negative test result was reported to decrease the risk estimate of malignancy in AUS/FLUS nodules in this study from 24% to 5%. This observation has led to a clinical extrapolation suggesting that nodules that have a negative 167 GEC test results may be followed without surgery (163). In a recent single-center retrospective study including 68 cases of AUS/FLUS nodules, 16 AUS/FLUS cases were re- ported to have a 167 GEC suspicious result, and the PPV was 61% (11 of 18) for those with surgical pathology confirma- tion (200). There were insufficient data to confirm the NPV of the 167 GEC test for AUS/FLUS cytology, since the vast majority of patients with a benign 167 GEC test did not undergo surgery, and no long-term follow-up of such cases was reported | 238 |
(200). In three other recent studies, there were insufficient data for analysis in the AUS/FLUS subgroup to draw any mean- ingful conclusion on 167 GEC test performance in that sub- group (172–174). In addition, published follow-up for the 167 GEC is currently limited to a mean of 8.5 months in a subgroup of 71 patients (171), hence long-term outcome data are needed to ensure durability of benign 167 GEC findings with corre- lation to clinical and histologic outcomes.
Several recent studies (201–205) have examined the feasi- bility of using sonographic features to estimate risk of malig- nancy in nodules with AUS/FLUS cytology. The PPV of suspicious sonographic features has been estimated to range from 60% to 100% depending on the pretest probability of malignancy of AUS/FLUS cytology and the specific sono- graphic criteria selected in respective studies. A limitation of all of these studies is that a gold-standard surgical excisional di- agnosis was not required for confirmation of malignancy and long-term follow-up data were generally lacking. The preva- lence of suspicious sonographic features among studies of AUS/ FLUS cytology nodules ranged from 18% to 50%, assuming that one or more suspicious features were deemed to be suffi- cient to be categorized as a sonographically suspicious nodule. The findings of these studies must be interpreted in the context of each study’s overall risk of malignancy for this cytology classification because of its effect on the PPVs obtained by subsequent application of sonographic features to cytologically AUS/FLUS nodules. From the four Korean studies (overall malignancy rate 40%–55%), the reported cancer risk in AUS/ FLUS nodules with the high suspicion sonographic pattern is 90%–100% (201–204), and the presence of even one suspi- cious US feature (irregular margins, taller than wide shape, marked hypoechogenicity or microcalcifications) increases the cancer risk to 60%–90%. However, when the reported cancer rate in AUS/FLUS nodules is lower [e.g., 23% in a study from Brazil (205)], the high suspicion sonographic pattern still raises the risk of malignancy, but the PPV is lower at 70%. None- theless, the incidence of cancer in AUS/FLUS nodules with either the high suspicion pattern US or just one suspicious US feature is significantly higher than that generally accepted for this cytology category. In a secondary analysis of a retro- spective single-center study of indeterminate FNA specimens subject to 167 GEC testing, Lastra et al. studied whether re- examining US characteristics could assist in distinguishing malignancy in indeterminate cytology nodules with a 167 GEC suspicious result (200). The presence of ‘‘nodular cal- cifications’’ or hyper- versus hypo-echogenicity did not alter the prediction of malignancy. It is unclear whether this study was adequately powered because the analysis was limited to a subgroup of 48 cases for analysis of microcalcifications and 20 cases for analysis of echogenicity (200). Further research is needed to examine the impact of considering clinical and sonographic features on the potential utility and interpreta- tion of molecular testing of FNA specimens. | 240 |
[A18] Follicular neoplasm/suspicious for follicular neo- plasm cytology | 241 |
RECOMMENDATION 16
Diagnostic surgical excision is the long-established standard of care for the management of FN/SFN cytology nodules. However, after consideration of clinical and
sonographic features, molecular testing may be used to supplement malignancy risk assessment data in lieu of proceeding directly with surgery. Informed patient pref- erence and feasibility should be considered in clinical decision-making.
(Weak recommendation, Moderate-quality evidence)
If molecular testing is either not performed or incon-
clusive, surgical excision may be considered for removal and definitive diagnosis of an FN/SFN thyroid nodule.
(Strong recommendation, Low-quality evidence) | 242 |
This diagnostic category of the Bethesda System is used for cellular aspirates (i) comprised of follicular cells arranged in an altered architectural pattern characterized by cell crowding and/or microfollicle formation and lacking nuclear features of papillary carcinoma or (ii) comprised almost ex- clusively of oncocytic (Hu¨rthle) cells (99,206,207). This is an intermediate risk category in the Bethesda System, with a 15%–30% estimated risk of malignancy. Studies that applied the Bethesda System reported the use of this diagnostic cat- egory in 1%–25% (mean, 10%) of all thyroid FNA samples, with the risk of cancer on surgery found to range from 14% to 33% (mean, 26%) (191).
The refinement of risk stratification of nodules with FN/ SFN/Hu¨rthle cell neoplasm cytology has been examined using ancillary molecular testing. Testing for a seven-gene panel of mutations (including BRAF, RAS, RET/PTC, and/PPARc) in nodules with follicular or Hu¨rthle cell neoplasm or SFN has been reported to be associated with a sensitivity of 57%–75%, specificity of 97%–100%, PPV of 87%–100%, and NPV of 79%–86% (162,165). Many of these benign tumors are follic- ular adenomas driven by oncogenic RAS mutation with un- certain malignant potential (208). Nodules lacking all of these mutations still have a substantial cancer risk, which is due to the presence of a subset of tumors that lack any of the mutations tested by this seven-gene panel (162). Expansion of the current panels to include additional mutations and gene rearrangements with this next-generation sequencing assay was associated with a sensitivity of 90% [95% CI 80%–99%], specificity of 93%
[95% CI 88%–98%], PPV of 83% [95% CI 72%–95%], and
NPV of 96% [95% CI 92%–100%] in a recent single-center study of 143 consecutive FN/SFN FNA specimens with known surgical outcomes. In this study (170), retrospective (n = 91) and prospective data (n = 52) were combined. A limitation was that the pathologists reviewing the surgical specimens were aware of earlier generation molecular marker seven-gene panel test results, although they were blinded to results of the next- generation mutation panel (162,170). Given the overlap of some of the markers detected in earlier and later generation assays, there is a potential for bias (170), and the results need to be replicated in other studies.
Molecular testing using the GEC was reported to have a 94% NPV [95% CI 79%–99%], and a 37% PPV [95% CI
23%–52%] in the FN/SFN/Hu¨rthle cell neoplasm Bethesda subgroup (163). A recent unblinded study from the Mayo Clinic utilizing a prospective patient registry reported the following diagnostic accuracy estimates in a subgroup of 31 indeterminate nodules from the same Bethesda subgroup that were subject to histopathologic confirmation: sensitivity 80% (four of five nodules), specificity 12% (3 of 26 nodules), PPV | 243 |
15% (4 of 27), and NPV 75% (three of four) (173). In a single- center retrospective study including 64 nodules subjected to 167 GEC testing and a cytology read as FN/FN with oncocytic features, the PPV for a suspicious GEC result was 37% (11 of 30), although the PPV was significantly higher in the FN group (53%) compared with the FN with oncocytic features group (15%) (200). There were insufficient numbers of pa- tients with benign 167 GEC results who underwent surgery to draw conclusions on NPV; moreover, no long-term follow-up data were reported (200). The relatively small number of in- determinate nodules is an important limitation. Furthermore, precision estimates (95% CIs) for the diagnostic accuracy measures were not reported but could be assumed to be rel- atively wide given the small sample size (173,200). | 245 |
[A19] Suspicious for malignancy cytology | 246 |
RECOMMENDATION 17
If the cytology is reported as suspicious for papillary carcinoma (SUSP), surgical management should be simi- lar to that of malignant cytology, depending on clinical risk factors, sonographic features, patient preference, and possibly results of mutational testing (if performed).
(Strong recommendation, Low-quality evidence)
After consideration of clinical and sonographic features, mutational testing for BRAF or the seven-gene mutation marker panel (BRAF, RAS, RET/PTC, PAX8/ PPARc) may be considered in nodules with SUSP cytology if such data would be expected to alter surgical decision- making.
(Weak recommendation, Moderate-quality evidence) | 247 |
This diagnostic category of the Bethesda System is reserved for aspirates with cytologic features that raise a strong suspi- cion for malignancy (mainly for PTC) but are not sufficient for a conclusive diagnosis (99,209). This is the highest risk cate- gory of indeterminate cytology in the Bethesda System, with an estimated cancer risk of 60%–75% (209). Studies that uti- lize the Bethesda System have reported this cytologic diag- nosis in 1%–6% of thyroid FNAs and found malignancy after surgery in 53%–87% (mean, 75%) of these nodules (191). Due to the high risk of cancer, the diagnosis of suspicious for papillary carcinoma is an indication for surgery.
Mutational testing has been proposed to refine risk prior to surgery, assuming that surgical management would change based on a positive test result. BRAF mutations have been reported to confer close to 100% probability of malignancy (162,198,210,211). Testing for BRAF mutations in nodules suspicious for malignancy has been reported to have a sen- sitivity of 36% (10 of 28) and specificity of 100% (24 of 24) in a single-center retrospective study (162). In another single center retrospective study in which FNA slides deemed to be suspicious for PTC were tested after surgery, the sensitivity of BRAF testing for PTC was 45.5% (15 of 33), and specificity was 85.7% (12 of 14) (212). Testing for a seven-gene panel of mutations (including BRAF, RAS, RET/PTC, with or without PAX8/PPARc) in nodules with cytology suspicious for ma- lignancy is associated with a sensitivity of 50%–68%, speci- ficity of 86%–96%, PPV of 80%–95%, and NPV of 72%–75%
in respective single-center studies (162,165,168). Molecular testing using the 167 GEC has a PPV that is similar to cytol- ogy alone (76%) and a NPV of 85% (163), and it is therefore not indicated in patients with this cytologic diagnosis. | 248 |
[A20] What is the utility of 18FDG-PET scanning to predict malignant or benign disease when FNA cytology is indeterminate (AUS/FLUS, FN, SUSP)? | 249 |
RECOMMENDATION 18
18FDG-PET imaging is not routinely recommended for the evaluation of thyroid nodules with indeterminate cytology.
(Weak recommendation, Moderate-quality evidence) | 250 |
Eight studies have been performed and are the subject of two meta-analyses (213–222). While early data suggested a high NPV for 18FDG-PET in this setting, most studies failed to use the Bethesda System for Reporting Thyroid Cyto- pathology and included numerous small nodules <1 cm in diameter (221). A recent meta-analysis included seven studies, of which five were prospective (222). The cancer prevalence was 26% inclusive of all combined data, con- firming a typical study cohort. Sensitivity and specificity of 18FDG-PET were 89% and 55%, respectively, resulting in a 41% PPV and 93% NPV, which is similar to the perfor- mance of the 167 GEC. Vriens et al. (223) performed a cost- effectiveness analysis using 18FDG-PET performance data from their own meta-analysis and 2012 reimbursement rates of the Dutch system. They showed that 18FDG-PET was more cost effective than surgery, the 167 GEC, or mutational testing. A recent prospective analysis of 56 nodules with indetermi- nate FNA cytology used both 18FDG-PET and thyroid US to further evaluate the nodules (220). In a multivariate analy- sis, the authors demonstrated no additional diagnostic benefit or improved risk assessment when adding 18FDG-PET to that already obtained from thyroid US, bringing into question the incremental benefit of PET imaging in patients with cytolog- ically indeterminate thyroid nodules. | 251 |
[A21] What is the appropriate operation
for cytologically indeterminate thyroid nodules? | 252 |
RECOMMENDATION 19
When surgery is considered for patients with a solitary, cytologically indeterminate nodule, thyroid lobectomy is the recommended initial surgical approach. This approach
may be modified based on clinical or sonographic char- acteristics, patient preference, and/or molecular testing when performed (see Recommendations 13–16).
(Strong recommendation, Moderate-quality evidence) | 253 |
RECOMMENDATION 20
Because of increased risk for malignancy, total
thyroidectomy may be preferred in patients with indeter-
minate nodules that are cytologically suspicious for ma- lignancy, positive for known mutations specific for carcinoma, sonographically suspicious, or large (>4 cm), or in patients with familial thyroid carcinoma or history of radiation exposure, if completion thyroidectomy would be | 254 |
recommended based on the indeterminate nodule being malignant following lobectomy.
(Strong recommendation, Moderate-quality evidence)
Patients with indeterminate nodules who have bilateral nodular disease, those with significant medical co- morbidities, or those who prefer to undergo bilateral thy- roidectomy to avoid the possibility of requiring a future surgery on the contralateral lobe, may undergo total or near-total thyroidectomy, assuming completion thyroid- ectomy would be recommended if the indeterminate nodule proved malignant following lobectomy.
(Weak recommendation, Low-quality evidence) | 256 |
The primary goal of thyroid surgery for a thyroid nodule that is cytologically indeterminate (AUS/FLUS or FN or SUSP) is to establish a histological diagnosis and definitive removal, while reducing the risks associated with remedial surgery in the previously operated field if the nodule proves to be malignant. Surgical options to address the nodule should be limited to lobectomy (hemithyroidectomy) with or without isthmusectomy, near-total thyroidectomy (removal of all grossly visible thyroid tissue, leaving only a small amount [<1 g] of tissue adjacent to the recurrent laryngeal nerve near the ligament of Berry), or total thyroidectomy (removal of all grossly visible thyroid tissue). Removal of the nodule alone, partial lobectomy, and subtotal thyroidectomy, leaving >1g of tissue with the posterior capsule on the uninvolved side, are inappropriate operations for possible thyroid cancer (224).
Decisions regarding the extent of surgery for indeterminate thyroid nodules are influenced by several factors (225), in- cluding the estimated presurgical likelihood of malignancy based upon clinical risk factors (>4 cm, family history, and/or history of radiation) (226–229), sonographic pattern (Table 6, Fig. 2) (202–203), cytologic category (Table 8), and ancillary test findings (see molecular testing section [A15–19]). These risk factors, as well as patient preference, presence of con- tralateral nodularity or coexistent hyperthyroidism, and medical comorbidities, impact decisions regarding thyroid lobectomy with the possible need for subsequent completion thyroidectomy versus total thyroidectomy up front.
Nodules that are cytologically classified as AUS/FLUS or FN and benign using the 167 GEC, or AUS/FLUS and neg- ative using the seven-gene mutation panel have an estimated 5%–6% risk of malignancy (162,163). Nodules that are cyto- logically classified as FN and negative using the seven-gene mutation panel have an estimated 14% risk of malignancy, which is slightly lower than the risk based upon the Bethesda classification alone (162).
Nodules that are cytologically classified as AUS/FLUS or FN and as suspicious using the 167 GEC have an estimated 37%–44% risk of malignancy, which is slightly higher than the risk based upon the Bethesda classification alone (Table 8) (163,171).
Nodules that are cytologically classified as SUSP cytology and benign using the 167 GEC or negative using the mutation seven-gene panel, also have an estimated 15%–28% risk of malignancy.
In contrast, nodules that are cytologically classified as AUS/FLUS or FN and that are positive for known RAS mu- tations associated with thyroid carcinoma have an estimated
84% risk of malignancy and should be considered in a similar risk category to cytologically suspicious for malignancy (Table 8) (103,230). Nodules that are cytologically classified as AUS/FLUS or FN or SUSP and that are positive for known BRAFV600E, RET/PTC, or PAX8/PPARc mutations have an estimated risk of malignancy of >95% and should be con- sidered in a similar category to cytologically diagnosed thyroid carcinoma (198,210,211).
Sonographic pattern of nodules with AUS/FLUS cytology may aid in risk stratification and management. In one study, sonographically benign or seemingly very low risk nodules with AUS/FLUS cytology were noted to be malignant in only 8% of cases, compared to 58% when sonographic sus- picious was low or intermediate, and 100% when sono- graphic suspicion of malignancy was high (203). Another study supported this finding with sonographic findings highly suspicious for malignancy (taller than wide shape, irregular borders, and/or marked hypoechogenicity) having >90% specificity and PPV for malignancy (202).
The risks of total thyroidectomy are significantly greater than that for thyroid lobectomy, with a recent meta-analysis suggesting a pooled relative risk (RR) significantly greater for all complications, including recurrent laryngeal nerve in- jury (transient RR = 1.7, permanent RR = 1.9), hypocalcemia (transient RR = 10.7, permanent RR = 3.2), and hemorrhage/ hematoma (RR = 2.6) (231). Further, total thyroidectomy is associated with the rare but potential risk of bilateral recurrent laryngeal nerve injury necessitating tracheostomy. Surgeon
experience likely influences the risks of thyroidectomy, with
higher volume surgeons having lower complication rates (232,233).
Total thyroidectomy necessitates thyroid hormone re- placement, while lobectomy is associated with an incidence of postoperative biochemical hypothyroidism estimated at 22%, with clinical or overt hypothyroidism estimated at 4% (234). A significantly increased risk of hypothyroidism fol- lowing lobectomy has been reported in the presence of auto- immune thyroid disease (e.g., as reflected by the presence of thyroid antibodies) or high normal/elevated TSH (231,234). Hypothyroidism is not an indication for thyroidectomy, and its use as justification for total thyroidectomy over lobectomy should be weighed against the higher risks associated with total thyroidectomy. In contrast, coexistent hyperthyroidism may be an indication for total thyroidectomy depending upon the etiology.
Thyroid lobectomy (hemithyroidectomy) provides defini- tive histological diagnosis and complete tumor removal for cytologically indeterminate nodules with a lower risk of complications compared to total thyroidectomy and may be sufficient for smaller, solitary intrathyroidal nodules that ul- timately prove malignant. As the likelihood of malignancy increases, the potential need for a second operation also in- creases, if the cytologically indeterminate nodule ultimately proves malignant and if completion thyroidectomy would be recommended. Intraoperative evaluation, with or without frozen section, can occasionally confirm malignancy at the time of lobectomy allowing for conversion to total thyroid- ectomy if indicated. Frozen section is most helpful if the histopathologic diagnosis is classic PTC, whereas its impact is low in follicular variant of PTC and FTC. The individual patient must weigh the relative advantages and disadvantages of thyroid lobectomy with possible total thyroidectomy or | 257 |
subsequent completion thyroidectomy versus initial total thyroidectomy. | 259 |
[A22] How should multinodular thyroid glands (i.e., two or more clinically relevant nodules) be evaluated for malignancy? | 260 |
RECOMMENDATION 21
Patients with multiple thyroid nodules ‡1 cm should be evaluated in the same fashion as patients with a solitary nodule ‡1 cm, excepting that each nodule that is >1 cm carries an independent risk of malignancy and therefore multiple nodules may require FNA.
(Strong recommendation, Moderate-quality evidence)
When multiple nodules ‡1 cm are present, FNA should be performed preferentially based upon nodule sono- graphic pattern and respective size cutoff (Table 6, Fig. 2).
(Strong recommendation, Moderate-quality evidence)
If none of the nodules has a high or moderate suspicion sonographic pattern, and multiple sonographically similar very low or low suspicion pattern nodules coalesce with no intervening normal parenchyma, the likelihood of malig- nancy is low and it is reasonable to aspirate the largest nodules (‡2 cm) or continue surveillance without FNA.
(Weak recommendation, Low-quality evidence) | 261 |
RECOMMENDATION 22
A low or low-normal serum TSH concentration in patients with multiple nodules may suggest that some nodule(s) may be autonomous. In such cases, a radionuclide (pref- erably 123I) thyroid scan should be considered and directly compared to the US images to determine functionality of each nodule ‡1 cm. FNA should then be considered only for those isofunctioning or nonfunctioning nodules, among which those with high suspicion sonographic pattern should be aspirated preferentially.
(Weak recommendation, Low-quality evidence) | 262 |
Patients with multiple thyroid nodules have the same risk of malignancy as those with solitary nodules (32,74). How- ever, when evaluating the risk of cancer per individual nod- ule, one large study found that a solitary nodule had a higher likelihood of malignancy than did a nonsolitary nodule ( p < 0.01), although in agreement with the other studies the risk of malignancy per patient was the same and independent of the number of nodules (77). A recent systematic review and meta-analysis confirmed the slightly higher risk of ma- lignancy in a solitary nodule compared with an individual nodule in a MNG. However, this appeared to hold true mostly outside of the United States and in iodine-deficient popula- tions (235). A diagnostic US should be performed to evaluate the sonographic risk pattern of each nodule, but if only the ‘‘dominant’’ or largest nodule is aspirated, the thyroid cancer may be missed (74). Therefore, multiple thyroid nodules
‡1 cm may require aspiration, based on sonographic pattern
(Recommendation 8, Table 6, Fig. 2) to fully exclude clini- cally relevant thyroid cancer. Radionuclide scanning may also be considered in patients with multiple thyroid nodules
with the goal of identifying and aspirating appropriate hy- pofunctioning nodules. Such imaging may prove especially useful when the serum TSH is below or in the low-normal range. Similarly, sonographic risk assessment of each nodule can assist in identifying those nodules with the highest like- lihood of cancer (see section [A10]). | 263 |
[A23] What are the best methods for long-term follow-up of patients with thyroid nodules?
[A24] Recommendations for initial follow-up of nodules with benign FNA cytology | 264 |
RECOMMENDATION 23
Given the low false-negative rate of US-guided FNA cy- tology and the higher yield of missed malignancies based upon nodule sonographic pattern rather than growth, the
follow-up of thyroid nodules with benign cytology diag- noses should be determined by risk stratification based
upon US pattern.
Nodules with high suspicion US pattern: repeat US and
US-guided FNA within 12 months.
(Strong recommendation, Moderate-quality evidence)
Nodules with low to intermediate suspicion US pat- tern: repeat US at 12–24 months. If sonographic evidence
of growth (20% increase in at least two nodule dimensions
with a minimal increase of 2 mm or more than a 50%
change in volume) or development of new suspicious so- nographic features, the FNA could be repeated or obser- vation continued with repeat US, with repeat FNA in case of continued growth.
(Weak recommendation, Low-quality evidence)
Nodules with very low suspicion US pattern (including spongiform nodules): the utility of surveillance US and assessment of nodule growth as an indicator for repeat FNA to detect a missed malignancy is limited. If US is repeated, it should be done at ‡24 months.
(Weak recommendation, Low-quality evidence) | 265 |
[A25] Recommendation for follow-up of nodules with two benign FNA cytology results
If a nodule has undergone repeat US-guided FNA with a second benign cytology result, US surveillance for this nod- ule for continued risk of malignancy is no longer indicated.
(Strong recommendation, Moderate-quality evidence)
Given that there is a low but discrete false-negative rate for nodules with benign FNA cytology results, is there an opti- mal way to identify these missed malignancies? Although the risk of malignancy after two benign cytology results is vir- tually zero (129–133,236), routine rebiopsy is not a viable or cost-effective option because of the low false-negative rate of an US-guided FNA benign cytology result. Prior guidelines have recommended repeat FNA for nodules that grow during serial sonographic observation. However, nodule growth can be variably defined. Because of interobserver variation, Brauer et al. (237) reported a 50% increase in nodule volume as the minimally significant reproducibly recorded change in | 266 |
nodule size, which is equivalent to a 20% increase in two of the three nodule dimensions. If a 50% volume increase cutoff is applied, only 4%–10% of nodules were reported to be larger at a mean of 18 months (133,238). However, using cutoffs of a 15% volume increase based upon inter- nally assessed interobserver coefficients of variation, pub- lished series report that 32%–50% of nodules increase in size over a 4–5 year period (239,240). Because of the stringent methodology of these studies, adoption of a 15% volume increase as statistically significant is not practically applicable.
A recent 5-year prospective multicenter study evaluated outcomes of 1597 nodules from 992 patients with either cy- tologically or sonographically benign nodules (241). Nodules 1 cm or larger underwent US-guided FNA and subcentimeter nodules were defined as sonographically benign based upon imaging characteristics equivalent to the ATA low or very low suspicion US patterns. All nodules were followed by annual US exams. The false-negative rate of a benign cy- tology diagnosis was 1.1%. Of the four missed cancers, on baseline US imaging three were hypoechoic and solid and one was isoechoic with microcalcifications; none was spon- giform or mixed cystic solid and noncalcified (ATA very low suspicion pattern). During sonographic surveillance, repeat FNA was prompted by either growth (two nodules) or de- velopment of a new suspicious sonographic feature (two nodules). In addition, the shortest time interval to detect change and repeat the FNA was 2 years. Another critical observation from this study was that only one cancer was detected in 5 years among the 852 subcentimeter nodules classified as sonographicially benign at baseline. This cancer was identified on the 5-year follow-up US, when its com- position changed from mixed cystic/solid to hypoechoic solid with irregular margins prompting FNA (241). Currently, there are no follow-up studies of nodule growth that extend observation beyond 5 years to help inform decision-making about long-term surveillance. Additional research would be valuable because indefinite follow-up of nodules with benign cytology is costly and may be unnecessary.
Recent investigations of repeat US-guided FNA in nodules with initial benign cytology show higher detection rates for missed malignancy for those nodules with a high suspicion so- nographic pattern rather than size increase (236,242). Kwak et al.
(236) reported a significantly higher malignancy rate of 20.4% in nodules with benign cytology that exhibited either marked hy- poechogenicity, irregular borders, microcalcifications, or a taller than wide shape versus a 1.4% risk in those that exhibited a 15% volume increase but lacked these US features. Importantly, the low risk of malignancy did not differ between US negative nodules that grew and those that demonstrated no interval size change (1.4% vs. 0.5%, p = 0.18). Similarly, Rosario et al. (242) detected cancer in 17.4% of nodules with benign cytologic di- agnoses and suspicious US features versus 1.3% of those without suspicious characteristic that grew, using criteria of a 50% vol- ume increase. These studies indicate that the use of suspi- cious US characteristics rather than nodule growth should be the indication for repeat FNA despite an initial benign cytology diagnosis. Repeat US and FNA should be repeated within 12 months as guided by clinical judgement. Given the low false-negative rate of US-guided FNA cytology and the higher yield of missed malignancies based upon nodule sonographic pattern rather than growth, the follow-up of
thyroid nodules with benign cytology diagnoses should be determined by risk stratification based upon US pattern as defined in Recommendation 8. If follow-up US for surveil- lance is performed and the nodule size is stable, the utility of subsequent US imaging for detection of potential malignancy by nodule growth assessment is very low and if performed, the time interval for any additional US exam should be at least as long that between the initial benign FNA cytology result and first follow-up. However, even if a repeat US is not indicated based on a benign cytology, US pattern, or stability in nodule size, larger nodules may require monitoring for growth that could result in symptoms and thus prompt surgical interven- tion despite benign cytology.
One recent study evaluated the long-term consequences of a false-negative benign cytology (140). A total of 1369 pa- tients with 2010 cytologically benign thyroid nodules were followed for a mean of 8.5 years. Eighteen false-negative cases were identified, although only a subset of patients un- derwent repeat FNA or thyroid surgery. Thirty deaths were documented in the entire cohort over this time period and none were attributable to thyroid cancer. These data sup- port that an initial benign cytology conveys an overall ex- cellent prognosis and a conservative follow-up strategy is reasonable. | 268 |
[A26] Follow-up for nodules that do not meet FNA criteria | 269 |
RECOMMENDATION 24
Nodules may be detected on US that do not meet criteria for FNA at initial imaging (Recommendation 8). The strategy for sonographic follow-up of these nodules should be based upon the nodule’s sonographic pattern.
Nodules with high suspicion US pattern: repeat US in 6–12 months.
(Weak recommendation, Low-quality evidence)
Nodules with low to intermediate suspicion US pat- tern: consider repeat US at 12–24 months.
(Weak recommendation, Low-quality evidence)
Nodules >1 cm with very low suspicion US pattern (including spongiform nodules) and pure cyst: the utility and time interval of surveillance US for risk of malignancy is not known. If US is repeated, it should be at ‡24 months.
(No recommendation, Insufficient evidence)
Nodules £1 cm with very low suspicion US pattern (including spongiform nodules) and pure cysts do not re- quire routine sonographic follow-up.
(Weak recommendation, Low-quality evidence) | 270 |
Ultrasound studies demonstrate that up to 50% of adults have thyroid nodules. The vast majority of these are subcentimeter, and FNA evaluation is generally not indicated. In addition, based upon both sonographic pattern and size cutoffs (Re- commendation 8), many nodules >1 cm may also be followed without FNA. Although no prospective studies address the optimal cost-effective surveillance strategy for these nodules that have not undergone FNA, a recent study by Durante et al.
(241) confirms that subcentimeter thyroid nodules corre- sponding to the ATA very low suspicion risk pattern are highly | 271 |
unlikely to change during 5-year sonographic follow-up, and the risk of malignancy is exceedingly low. The findings from studies correlating sonographic features and malignancy risk in aspirated nodules can be extrapolated to inform a follow-up strategy for this group of nodules that do not meet FNA criteria at the time of their initial detection. For example, the interval for follow-up sonography for a nodule that is hypoechoic and taller than wide should be shorter than that for an isoechoic solid nodule with smooth borders.
US-guided FNA should be performed based upon follow- up US imaging if the nodule subsequently meets criteria based upon Recommendation 8. | 273 |
[A27] What is the role of medical or surgical therapy for benign thyroid nodules? | 274 |
RECOMMENDATION 25
Routine TSH suppression therapy for benign thyroid nod- ules in iodine sufficient populations is not recommended. Though modest responses to therapy can be detected, the potential harm outweighs benefit for most patients.
(Strong recommendation, High-quality evidence) | 275 |
RECOMMENDATION 26
Individual patients with benign, solid, or mostly solid nodules should have adequate iodine intake. If inadequate dietary intake is found or suspected, a daily supplement (containing 150 lg iodine) is recommended.
(Strong recommendation, Moderate-quality evidence) | 276 |
RECOMMENDATION 27
Surgery may be considered for growing nodules that are benign after repeat FNA if they are large (>4 cm), causing compressive or structural symptoms, or based upon clinical concern.
(Weak recommendation, Low-quality evidence)
Patients with growing nodules that are benign after FNA should be regularly monitored. Most asymptomatic nodules demonstrating modest growth should be followed without intervention.
(Strong recommendation, Low-quality evidence) | 277 |
RECOMMENDATION 28
Recurrent cystic thyroid nodules with benign cytology should be considered for surgical removal or percutaneous ethanol injection (PEI) based on compressive symptoms and cosmetic concerns. Asymptomatic cystic nodules may be followed conservatively.
(Weak recommendation, Low-quality evidence) | 278 |
RECOMMENDATION 29
There are no data to guide recommendations on the use of thyroid hormone therapy in patients with growing nodules that are benign on cytology.
(No recommendation, Insufficient evidence)
Evidence from multiple prospective, RCTs, and from three meta-analyses suggest that thyroid hormone supplementation
in doses that suppress the serum TSH to subnormal levels may result in a decrease in nodule size and may prevent the appearance of new nodules in regions of the world with borderline low iodine intake (239,243–245). However, the effect is modest, with most studies suggesting an average 5%–15% reduction in nodule volume when treated with suppressive levothyroxine (LT4) therapy for 6–18 months. Two high-quality meta-analyses confirm that six to eight pa- tients will require suppressive LT4 therapy to achieve one successful treatment response (246,247). The extent of TSH suppression achieved in high-quality studies is variable, though the majority suppressed TSH to <0.2 mIU/L, with many to <0.1 mIU/L. Hyperthyroidism to this degree has been significantly associated with an increased risk of cardiac arrhythmias and osteoporosis, as well as adverse symptomatology. Together, these data confirm that LT4 suppressive therapy demonstrates modest (though usually clinically insignificant) efficacy in nodule volume reduction, but increases the risk of adverse consequences related to iatrogenic thyrotoxicosis. One large prospective, randomized trial demonstrated that sufficient die- tary iodine intake (150 lg daily) also reduced nodule size slightly more than placebo (248). The consumption of adequate dietary iodine is recommended for all adults and is without harm when not excessive. Data supporting LT4 therapy in non–TSH- suppressive doses for prevention of thyroid nodule growth are incomplete. One recent cohort analysis suggested non- suppressive doses of LT4 therapy conferred protection from nodule growth over time (249). However, the nonblinded, non- randomized nature of the trial precludes broad translation of the data, and the efficacy of nonsuppressive LT4 remains unproven. Cystic nodules that are cytologically benign can be monitored for recurrence (fluid reaccumulation), which can be seen in 60%–90% of patients (250,251). For those patients with subse- quent recurrent symptomatic cystic fluid accumulation, surgical removal, generally by hemithyroidectomy, or PEI are both rea- sonable strategies. Four controlled studies demonstrated a 75%– 85% success rate after PEI compared with a 7%–38% success rate in controls treated by simple cyst evacuation or saline in- jection. Success was achieved after an average of two PEI treatments. Complications included mild to moderate local pain, flushing, dizziness, and dysphonia (250–253). Surgery may be considered for growing solid nodules that are benign on repeat cytology if they are large (>4 cm), causing compressive or structural symptoms, or based upon clinical concern (254,255). | 279 |
[A28] How should thyroid nodules in pregnant
women be managed?
[A29] FNA for thyroid nodules discovered during preg- nancy | 280 |
RECOMMENDATION 30
FNA of clinically relevant thyroid nodules (refer to section [A10]) should be performed in euthyroid and hy-
pothyroid pregnant women.
(Strong recommendation, Moderate-quality evidence)
For women with suppressed serum TSH levels that persist beyond 16 weeks gestation, FNA may be deferred until after pregnancy and cessation of lactation. At that | 281 |
time, a radionuclide scan can be performed to evaluate nodule function if the serum TSH remains suppressed.
(Strong recommendation, Moderate-quality evidence) | 283 |
It is uncertain if thyroid nodules discovered in pregnant
women are more likely to be malignant than those found in
nonpregnant women, since there are no population-based studies to address this question. Pregnancy does not appear to
modify microscopic cellular appearance, thus standard diag- nostic criteria should be applied for cytologic evaluation (256). Serial evaluation of nodules throughout pregnancy has dem- onstrated that thyroid nodules will enlarge slightly throughout gestation, though this does not imply malignant transformation (257). The recommended evaluation of a clinically relevant
nodule in a pregnant patient is thus the same as for a non-
pregnant patient, with the exception that a radionuclide scan is
contraindicated. In addition, for patients with nodules diag- nosed as DTC by FNA during pregnancy, delaying surgery until after delivery does not affect outcome (258). Surgery performed during pregnancy is associated with greater risk of complications, longer hospital stays, and higher costs (259). | 284 |
[A30] Approaches to pregnant patients with malignant or indeterminate cytology | 285 |
RECOMMENDATION 31
PTC discovered by cytology in early pregnancy should be monitored sonographically. If it grows substantially (as defined in section [A24]) before 24–26 weeks ges-
tation, or if US reveals cervical lymph nodes that are suspicious for metastatic disease, surgery should be
considered during pregnancy. However, if the disease remains stable by midgestation, or if it is diagnosed in the second half of pregnancy, surgery may be de- ferred until after delivery.
(Weak recommendation, Low-quality evidence) | 286 |
If FNA cytology is consistent with PTC, surgery is gener- ally recommended. However, the decision to perform such surgery either during pregnancy or after delivery must be individualized. If surgery is not performed, the utility of
thyroid hormone therapy targeted to lower serum TSH levels to improve the prognosis of thyroid cancer diagnosed during gestation is not known. Because higher serum TSH levels may
be correlated with a more advanced stage of cancer at surgery (260), if the patient’s serum TSH is >2 mU/L, it may be rea- sonable to initiate thyroid hormone therapy to maintain the TSH between 0.3 to 2.0 mU/L for the remainder of gestation. Most data confirm that the prognosis of women with well- differentiated thyroid cancer identified but not treated during pregnancy is similar to that of nonpregnant patients. Because of this, surgery in most pregnant patients is deferred until postpartum (258,261), and no further testing is required. However, some studies differ from these findings. Two Ital- ian cohort studies have investigated women diagnosed with DTC in relation to the timing of pregnancy. Messuti et al.
(262) noted a statistically higher rate of persistence/recur- rence when DTC was diagnosed during pregnancy or within 2 years postpartum. However, the stimulated Tg was found to be >10 ng/mL during 131I ablation in many cases, suggesting the extent of thyroidectomy and/or tumor resection may have
been limited in this cohort and therefore contributed to bio- chemical persistence of disease. Vannucchi et al. (263) fol- lowed a small cohort of 10 patients with DTC during pregnancy or within 1 year post partum, again noting a large rate of persistent disease (60%) compared to nonpregnant controls (4.2%–13.1%). Similarly, the majority of cases with persistent disease were attributable to biochemical elevations in Tg or anti-Tg antibodies, again raising the question of whether the extent of initial resection was limited in com- parison to nonpregnant controls. Given the likelihood that biochemical persistence could be attributable to an increased size of remnant tissue or incomplete surgical resection in both studies, these data should not refute previous, larger analyses showing no increased recurrence rates when DTC is diag- nosed during pregnancy.
Theoretically, molecular marker analysis could be helpful in the evaluation of DTC or clinically relevant, cytologically indeterminate thyroid nodules detected during pregnancy. However, the application of molecular testing in pregnant women with indeterminate cytology remains uncertain. There are no published data validating the performance of any molecular marker in this population. Therefore, the committee cannot recommend for or against their use in pregnant women. However, it is theoretically possible that changes in a nodule’s RNA expression may occur during gestation altering performance of the 167 GEC while the seven-gene mutational panel (BRAF, RAS, PAX8/PPARc, RET/PTC) would be more likely to demonstrate similar performance to that of a nonpregnant population.
When surgery is advised during pregnancy, it is most often because of high-risk clinical or sonographic findings, nodule growth, or change over short duration follow-up or it is based upon physician judgement. To minimize the risk of
miscarriage, surgery during pregnancy should be done in the
second trimester before 24 weeks gestation (264). However, PTC discovered during pregnancy does not behave more aggressively than that diagnosed in a similar-aged group of nonpregnant women (258,265). A retrospective study of pregnant women with DTC found no difference in either recurrence or survival rates between women operated dur- ing or after pregnancy (258). Further, retrospective data suggest that treatment delays of <1 year from the time of thyroid cancer discovery do not adversely affect patient outcome (266). A separate study reported a higher rate of complications in pregnant women undergoing thyroid sur- gery compared with nonpregnant women (267). If FNA cytology is indeterminate, monitoring may be considered with further evaluation may be delayed until after delivery. Some experts recommend thyroid hormone suppression therapy for pregnant women with FNA suspicious for or diagnostic of PTC, if surgery is deferred until the postpar- tum period (259). | 287 |
[B1] DIFFERENTIATED THYROID CANCER: INITIAL MANAGEMENT GUIDELINES
Differentiated thyroid cancer, arising from thyroid follic- ular epithelial cells, accounts for the vast majority of thyroid cancers. Of the differentiated cancers, papillary cancer comprises about 85% of cases compared to about 12% that have follicular histology, including conventional and onco- cytic (Hu¨rthle cell) carcinomas, and <3% that are poorly | 288 |
differentiated tumors (268). In general, stage for stage, the prognoses of PTC and follicular cancer are similar (266,269). | 290 |
[B2] Goals of initial therapy of DTC
The basic goals of initial therapy for patients with DTC are to improve overall and disease-specific survival, reduce the
risk of persistent/recurrent disease and associated morbidity,
and permit accurate disease staging and risk stratification,
while minimizing treatment-related morbidity and unneces-
sary therapy. The specific goals of initial therapy are to
Remove the primary tumor, disease that has extended beyond the thyroid capsule, and clinically significant lymph node metastases. Completeness of surgical re- section is an important determinant of outcome, while residual metastatic lymph nodes represent the most com- mon site of disease persistence/recurrence (270–272).
Minimize the risk of disease recurrence and metastatic spread. Adequate surgery is the most important treat- ment variable influencing prognosis, while RAI treat- ment, TSH suppression, and other treatments each play adjunctive roles in at least some patients (273–275).
Facilitate postoperative treatment with RAI, where appropriate. For patients undergoing RAI remnant ablation, or RAI treatment of presumed (adjuvant therapy) or known (therapy) residual or metastatic disease, removal of all normal thyroid tissue is an important element of initial surgery (276).
Permit accurate staging and risk stratification of the
disease. Because disease staging and risk stratification should be used to guide initial prognostication, disease management, and follow-up strategies, accurate post- operative risk assessment is a crucial element in the management of patients with DTC (277,278).
Permit accurate long-term surveillance for disease re- currence.
Minimize treatment-related morbidity. The extent of surgery and the experience of the surgeon both play important roles in determining the risk of surgical complications (232,233,279,280). | 291 |
[B3] What is the role of preoperative staging with diagnostic imaging and laboratory tests?
[B4] Neck imaging—ultrasound | 292 |
RECOMMENDATION 32
Preoperative neck US for cervical (central and especially lateral neck compartments) lymph nodes is recommended for all patients undergoing thyroidectomy for malignant or suspicious for malignancy cytologic or molecular findings.
(Strong recommendation, Moderate-quality evidence)
US-guided FNA of sonographically suspicious lymph
nodes ‡8–10 mm in the smallest diameter should be per-
formed to confirm malignancy if this would change man- agement.
(Strong recommendation, Moderate-quality evidence)
The addition of FNA-Tg washout in the evaluation of suspicious cervical lymph nodes is appropriate in select
patients, but interpretation may be difficult in patients with an intact thyroid gland.
(Weak recommendation, Low-quality evidence) | 293 |
Differentiated thyroid carcinoma (particularly papillary car- cinoma) involves cervical lymph node metastases in 20%–50% of patients in most series using standard pathologic techniques (84,145,281–283), and it may be present even when the primary tumor is small and intrathyroidal (284). The frequency of mi- crometastases (<2 mm) may approach 90%, depending on the sensitivity of the detection method (285,286). However, the clinical implications of micrometastases are likely less signifi- cant compared to macrometastases. Preoperative US identifies suspicious cervical adenopathy in 20%–31% of cases, poten- tially altering the surgical approach (287,288) in as many as 20% of patients (289–291). However, preoperative US identifies only half of the lymph nodes found at surgery, due to the presence of the overlying thyroid gland (292).
Sonographic features suggestive of abnormal metastatic
lymph nodes include enlargement, loss of the fatty hilum, a
rounded rather than oval shape, hyperechogenicity, cystic
change, calcifications, and peripheral vascularity (Table 7). No single sonographic feature is adequately sensitive for detection of lymph nodes with metastatic thyroid cancer. One study correlated the sonographic features acquired 4 days preoperatively directly with the histology of 56 cervical lymph nodes identified in 19 patients. Some of the most specific criteria were short axis >5 mm (96%), presence of cystic areas (100%), presence of hyperechogenic punctua- tions representing either colloid or microcalcifications (100%), and peripheral vascularity (82%). Of these, the only one with sufficient sensitivity was peripheral vascularity (86%). The others had sensitivities of <60% and would not be adequate to use as a single criterion for identification of malignant involvement (292). As shown by earlier studies (293,294), the ultrasonographic feature with the highest sensitivity is absence of a hilum (100%), but this has a low specificity of 29%. Microcalcifications have the highest specificity; any lymph nodes with microcalcifications should be considered abnormal (292) (Table 7).
The location of the lymph nodes may also be useful for decision-making. Malignant lymph nodes are much more likely to occur in levels III, IV, and VI than in level II (Fig. 3) (292,294), although this may not be true for PTC tumors arising in the upper pole of the thyroid, which have a higher propensity to demonstrate skip metastases to levels III and II (295). Figure 3 illustrates the delineation of cervical lymph node levels I through VI.
Confirmation of malignancy in lymph nodes with a sus- picious sonographic appearance is achieved by US-guided FNA aspiration for cytology and/or measurement of Tg in the needle washout. A Tg concentration
and the probability of N1 disease increases with higher Tg
levels (296). This FNA measurement of Tg is likely valid even in patients with circulating anti-Tg autoantibodies (297,298), although one study challenges the validity of this measurement in patients with anti-Tg autoantibodies (299). Tg washout may be helpful, particularly in cases in which the lymph nodes are cystic, cytologic evaluation of the lymph node is inadequate, or the cytologic and sonographic evaluations are divergent (i.e., normal cytologic biopsy of | 294 |
FIG. 3. Lymph node compartments separated into levels and sublevels. Level VI contains the thyroid gland, and the adjacent nodes bordered superiorly by the hyoid bone, inferiorly by the innominate (brachiocephalic) artery, and laterally on each side by the carotid sheaths. The level II, III, and IV nodes are arrayed along the jugular veins on each side, bordered anteromedially by level VI and laterally by the posterior border of the sternocleidomastoid muscle. The level III nodes are bounded superiorly by the level of the hyoid bone and inferiorly by the cricoid cartilage; levels II and IV are above and below level III, respectively. The level I node compartment includes the submental and submandibular nodes, above the hyoid bone, and anterior to the posterior edge of the submandibular gland. Finally, the level V nodes are in the posterior triangle, lateral to the lateral edge of the sternocleidomastoid muscle. Levels I, II, and V can be further subdivided as noted in the figure. The inferior extent of level VI is defined as the suprasternal notch. Many authors also include the pretracheal and paratracheal superior mediastinal lymph nodes above the level of the innominate artery (sometimes referred to as level VII) in central neck dissection (341). | 296 |
a large lymph node with microcalcifications) (300). In a retrospective study of 241 lymph nodes in 220 patients who underwent US-guided FNA with Tg in FNA (FNA-Tg) washout fluid measurements for suspicious lymph nodes, additional FNA-Tg helped to diagnose a metastatic lymph node with one or two suspicious US features but did not offer incremental benefit for those lymph nodes with highly sus- picious US features in which FNA alone was sufficient for diagnosis. Two recent systematic reviews showed that false- positive Tg washout may occur, particularly in lymph nodes in the central compartment when the thyroid gland is still present (301,302). The review by Pak et al. (302) suggests that an FNA-Tg cutoff of 32 ng/mL has the best sensitivity and specificity in patients with an intact thyroid gland. Others have suggested interpreting the FNA-Tg in context of the serum Tg and TSH in these patients (303,304). There is no
standardization of FNA-Tg procedures or assays to date,
which makes this additional diagnostic tool sometimes dif-
ficult to interpret (305). Future standardization including matrix type (phosphate-buffered saline, Tg-free serum, etc.) and volume of diluent matrix would help with interpretation of a Tg washout.
Accurate staging is important in determining the prognosis and tailoring treatment for patients with DTC. However, unlike many tumor types, the presence of metastatic disease does not obviate the need for surgical excision of the primary
tumor in DTC (306). Because metastatic disease may respond to RAI therapy, removal of the thyroid as well as the primary tumor and accessible loco-regional disease remains an im- portant component of initial treatment even in most patients with metastatic disease. | 297 |
[B5] Neck imaging—CT/MRI/PET | 298 |
RECOMMENDATION 33
Preoperative use of cross-sectional imaging studies (CT, MRI) with intravenous (IV) contrast is recommended as an adjunct to US for patients with clinical suspicion for
advanced disease, including invasive primary tumor, or clinically apparent multiple or bulky lymph node involve- ment.
(Strong recommendation, Low-quality evidence)
Routine preoperative 18FDG-PET scanning is not rec- ommended.
(Strong recommendation, Low-quality evidence) | 299 |
Since US evaluation is operator dependent and cannot al- ways adequately image deep anatomic structures and those acoustically shadowed by bone or air, alternative imaging procedures may be preferable or useful as an adjunct in | 300 |
some clinical settings. Patients displaying bulky or widely distributed nodal disease on initial US examination may present with involvement of nodal regions beyond typical cervical regions, some of which maybe difficult to visualize on routine preoperative US, including the mediastinum, infra- clavicular, retropharyngeal, and parapharyngeal regions. In a study of 37 consecutive patients who had preoperative CT
and US and subsequently underwent total thyroidectomy and neck dissection, the sensitivity of CT was better than US for
the evaluation central and lateral compartment lymph nodes
examined together (77% vs 62%, p = 0.002), but there were
no differences between the two imaging modalities when the
central and lateral compartments were examined separately (307). In a series of 299 consecutively registered patients with pathologically proven PTC who underwent preoperative CT and US, US was more accurate than CT in predicting extrathyroidal tumor extension and multifocal bilobar dis- ease ( p < 0.05). The accuracy of staging was better overall
with US ( p < 0.01), and US had greater sensitivity than CT at predicting lateral compartment metastases ( p = 0.041) (308). However, another study showed that combined preopera- tive mapping with US and CT was superior to US alone in
the preoperative detection of nodal disease, especially in the central neck (309). The sensitivities of MRI and PET for the detection of cervical lymph node metastases are relatively low (30%–40%) (310). PET can also detect inflammatory
lymph nodes, which reduces the specificity of this test in many patients with DTC. MRI can be used in the detection of cervical nodal metastasis. MRI is affected by respiratory artifacts and may be more difficult to interpret than CT scanning by surgeons in the operating room for low-volume nodal disease (311).
Invasive DTC has been reported to occur in 10%–15% of patients at the time of diagnosis (312). For this group of patients, cross-sectional imaging can also be a useful sup- plement for preoperative planning to accurately delineate the extent of laryngeal, tracheal, esophageal, or vascular in- volvement (309,313). Endoscopy of the trachea and or esophagus, with or without ultrasonography, at the beginning of the initial operation looking for evidence of intraluminal extension can also be helpful in cases of suspected areodi- gestive tract invasion.
Locally invasive primary tumors may be associated with characteristic signs and symptoms including progressive dysphagia, respiratory compromise, hemoptysis, rapid tumor enlargement, significant voice change or the finding of vocal cord paralysis, and mass fixation to the airway or neck structures. Certain sonographic features of the primary tu- mor, including extrathyroidal extension especially with posterior capsular extension and extension into the medias- tinum, may also prompt axial imaging (307). Chest CT is
useful in defining the inferior border of disease and in de-
termining the extent to which mediastinal structures are in- volved in cases with significant caudal spread. CT findings may influence management by indicating the need for ster-
notomy and/or tracheal or laryngeal resection/reconstruction, which would likely require assembling additional resources and personnel in preparation for surgery. Neck CT with
contrast can therefore be useful in delineating the extent of laryngeal, tracheal, and/or esophageal involvement in tumors displaying aggressive local invasion, as well as delineating bulky nodal disease, which may harbor significant extranodal
extension that involves muscle and/or blood vessels. Pre- operative knowledge of these features of the primary tumor or metastases could significantly influence the surgical plan (314). 18FDG-PET scanning may be sensitive in some pa-
tients for neck or mediastinal involvement and may reveal
distant metastases as well.
When cross-sectional imaging is performed, use of IV contrast is an important adjunct because it helps to delineate the anatomic relationship between the primary tumor or metastatic disease and these other structures. Iodine is gen-
erally cleared within 4–8 weeks in most patients, so concern
about iodine burden from IV contrast causing a clinically
significant delay in subsequent whole-body scans (WBSs) or
RAI treatment after the imaging followed by surgery is gen-
erally unfounded (315). The benefit gained from improved
anatomic imaging generally outweighs any potential risk of a
several week delay in RAI imaging or therapy. When there is
concern, a urinary iodine to creatinine ratio can be measured. | 302 |
[B6] Measurement of serum Tg and anti-Tg antibodies | 303 |
RECOMMENDATION 34
Routine preoperative measurement of serum Tg or anti-Tg antibodies is not recommended.
(Weak recommendation, Low-quality evidence) | 304 |
Data from a systematic review and meta-analysis sug- gested that high preoperative concentrations of serum Tg may predict a higher sensitivity for postoperative surveil- lance with serum Tg (316). Preoperative anti-Tg antibodies do not appear to be an independent preoperative predictor of stage in patients with DTC, but the evidence is limited. In a cross-sectional analysis of 1770 patients with perioperative anti-Tg antibodies status data in the National Thyroid Cancer Treatment Cooperative Study (a large thyroid cancer registry that included 11 North American centers and enrolled pa- tients between 1987 and 2011), serum anti-Tg antibody status was not significantly associated with stage of disease on multivariate analysis, or with disease-free or overall survival on univariate or multivariate analyses (317). Evidence that preoperative measurement of serum Tg impacts patient management or outcomes is not yet available. | 305 |
[B7] Operative approach for a biopsy diagnostic for follic- ular cell–derived malignancy | 306 |
RECOMMENDATION 35
For patients with thyroid cancer >4 cm, or with gross extrathyroidal extension (clinical T4), or clinically appar- ent metastatic disease to nodes (clinical N1) or distant sites (clinical M1), the initial surgical procedure should include a near-total or total thyroidectomy and gross removal of all primary tumor unless there are contraindications to this procedure.
(Strong recommendation, Moderate-quality evidence)
For patients with thyroid cancer >1 cm and <4 cm without extrathyroidal extension, and without clinical evi- dence of any lymph node metastases (cN0), the initial surgical procedure can be either a bilateral procedure (near- total or total thyroidectomy) or a unilateral procedure | 307 |
(lobectomy). Thyroid lobectomy alone may be sufficient initial treatment for low-risk papillary and follicular car- cinomas; however, the treatment team may choose total thyroidectomy to enable RAI therapy or to enhance follow- up based upon disease features and/or patient preferences.
(Strong recommendation, Moderate-quality evidence)
If surgery is chosen for patients with thyroid cancer
<1 cm without extrathyroidal extension and cN0, the initial surgical procedure should be a thyroid lobectomy unless there are clear indications to remove the contralateral lobe. Thyroid lobectomy alone is sufficient treatment for small, unifocal, intrathyroidal carcinomas in the absence of prior head and neck radiation, familial thyroid carcinoma, or clinically detectable cervical nodal metastases.
(Strong recommendation, Moderate-quality evidence) | 309 |
Surgery for thyroid cancer is an important element of a multifaceted treatment approach. The operation must be compatible with the overall treatment strategy and follow-up plan recommended by the managing team. Consideration should be given to referring patients with high-risk features (clinical N1 disease, concern for recurrent laryngeal nerve [RLN] involvement, or grossly invasive disease) to experi- enced surgeons, as both completeness of surgery and expe- rience of the surgeon can have a significant impact on clinical outcomes and complication rates (232,233,279,280). Pre- vious guidelines have endorsed total thyroidectomy as the primary initial surgical treatment option for nearly all DTCs
>1 cm with or without evidence of loco-regional or distant metastases (25). This was based on retrospective data sug- gesting that a bilateral surgical procedure would improve survival (318), decrease recurrence rates (319–321), allow for routine use of RAI remnant ablation, and facilitate de- tection of recurrent/persistent disease during follow-up. However, recent data have demonstrated that in properly selected patients, clinical outcomes are very similar follow- ing unilateral or bilateral thyroid surgery (322–326). Fur- thermore, since the requirement for routine use of RAI ablation was one of the major reasons given in support of total thyroidectomy in low to intermediate risk patients, our cur- rent more selective approach to RAI ablation in these patients requires a critical reassessment of this indication. In some patients, the presence of the remaining lobe of the gland may obviate the lifelong need for exogenous thyroid hormone therapy. Finally, as our follow-up management paradigm has moved away from diagnostic whole body RAI scanning and toward a greater reliance on neck ultrasonography and serial serum Tg measurements (even in patients that did not receive RAI remnant ablation), we must also question whether total thyroidectomy and RAI remnant ablation is required to fa- cilitate follow-up in low to intermediate risk patients.
In an analysis of 52,173 PTC patients diagnosed between 1985 and 1998 from the National Cancer Data Base (43,227 receiving total thyroidectomy, 8946 undergoing lobectomy), Bilimoria et al. (318) demonstrated a slightly higher 10-year relative overall survival for total thyroidectomy as opposed to thyroid lobectomy (98.4% vs. 97.1%, respectively, p < 0.05) and a slightly lower 10-year recurrence rate (7.7% vs. 9.8%, respectively, p < 0.05). When analyzed by size of the primary tumor, statistically significant differences in survival and
recurrence were seen for all sizes >1 cm based on the extent of initial surgery. However, data on extrathyroidal extension, completeness of resection, and other comorbid conditions, which could have had a major impact on survival and re- currence risk, were not available. Therefore, it is unclear how often lobectomy was done based on proper selection of low to intermediate risk patients versus how often lobectomy was done in high-risk patients because of comorbid conditions, inability to obtain a complete resection, or status of the contralateral RLN. This is an important distinction because thyroid lobectomy patients were found to have extrathyroidal extension in 7% of cases (325), underwent external beam radiation therapy (EBRT) in 1%–2% (324), and RAI therapy in 12%–18% (318,325), and high-risk features were present in 8% (325). Given the small magnitude of differences re- ported for survival and recurrence between the total thyroid- ectomy and the lobectomy patients, it is quite possible that the slightly poorer outcomes seen in the lobectomy group could have been influenced by lobectomy patients with concurrent high risk features. Adam et al. (327) performed an updated analysis of 61,775 patients in the National Cancer Database who underwent thyroid surgery between 1998 and 2006. The researchers demonstrated that the overall survival advantage seen for patients with 1–4 cm PTC who underwent thyroid- ectomy in the study by Bilimoria et al. (318) disappeared when further adjustment was made for additional variables related to complexity and severity of illness. This lack of overall survival advantage was also seen when the group was subdivided into patients with 1–2 cm and 2–4 cm PTC.
Previously, Haigh et al. (325) had analyzed 5432 PTC patients from the SEER database (4612 receiving total thy- roidectomy and 820 undergoing lobectomy) and found no difference in 10-year overall survival between total thyroid- ectomy and thyroid lobectomy when risk stratified by the AMES classification system. Interestingly, patients selected for thyroid lobectomy included 7% with extrathyroidal ex- tension, 1% with distant metastases, and 5% with primary tumors >5 cm, and 8% were classified as having high risk based on AMES.
More recently, two additional studies have analyzed the SEER database, and both have failed to demonstrate a sig- nificant difference in survival when comparing total thy- roidectomy with thyroid lobectomy (323,324). Barney et al.
(323) included 23,605 DTC patients diagnosed between 1983 and 2002 (12,598 with total thyroidectomy, 3266 with lo- bectomy) and found no difference in 10-year overall survival (90.4% for total thyroidectomy vs. 90.8% for lobectomy) or 10-year cause-specific survival (96.8% for total thyroidec- tomy vs. 98.6% for lobectomy). Furthermore, in a multivar- iate analysis that included age, T, N, M, sex, year of diagnosis, extent of surgery, and RAI use, no difference in overall survival or cause specific survival was seen with re- spect to the extent of initial surgery. Mendelsohn et al. (324) analyzed 22,724 PTC patients diagnosed between 1998 and 2001 (16,760 with total thyroidectomy, 5964 with lobecto- my) and found no differences in overall survival or disease- specific survival in a comparison of total thyroidectomy with lobectomy. Interestingly, of the patients that had lobectomy, 1.6% received external beam radiation therapy, 16% had extrathryoidal extension, 9% of tumors were >4 cm, and 20% received RAI ablation (once again indicating that lobectomy was done in some high-risk patients). | 310 |
Consistent with the SEER data analyses, two single-center studies also confirmed that lobectomy is associated with excellent survival in properly selected patients (322,326). After a median follow-up of 8 years, only one disease- specific death was seen in a cohort of 889 PTC patients with T1–T2 tumors treated with either total thyroidectomy (n = 528) or lobectomy (n = 361) (326). Furthermore, Mat- suzu et al. (322) reported a cause-specific survival rate of 98% after a median of 17 years of follow-up in properly select PTC patients treated with lobectomy and ipsilateral neck dissection.
Given the propensity for PTC to be multifocal (often in- volving both lobes), it is not surprising that some studies have demonstrated a lower risk of loco-regional disease recurrence following total thyroidectomy as compared to thyroid lo- bectomy (319–321). However, with proper patient selection, loco-regional recurrence rates of less than 1%–4% and completion thyroidectomy rates of <10% can be achieved following thyroid lobectomy (326,328). Furthermore, the few recurrences that develop during long-term follow-up are readily detected and appropriately treated with no impact on survival (322,326,328).
Therefore, we conclude that in properly selected low- to intermediate-risk patients (patients with unifocal tumors
<4 cm, and no evidence of extrathyroidal extension or lymph node metastases by examination or imaging), the extent of initial thyroid surgery probably has little impact on disease-specific survival. While recurrence rates can be quite low in these patients, it is likely that the lowest rates of recurrence during long-term follow-up would be associated with a total thyroidectomy. But since salvage therapy is quite effective in the few patients that recur after thyroid lobectomy, a conservative management approach to com- pletion surgery, accepting a slightly higher risk of loco- regional recurrence, is a reasonable management strategy. Finally, a more selective use of RAI coupled with a greater reliance on neck US and serial serum Tg measurements for detection of recurrent disease is likely to significantly de- crease the mandate for total thyroidectomies in low- and intermediate-risk patients done solely to facilitate RAI remnant ablation and follow-up.
Near-total or total thyroidectomy is necessary if the overall strategy is to include RAI therapy postoperatively, and thus is recommended if the primary thyroid carcinoma is >4 cm, if there is gross extrathyroidal extension, or if regional or dis- tant metastases are clinically present. For tumors that are between 1 and 4 cm in size, either a bilateral thyroidectomy (total or near-total) or a unilateral procedure (thyroid lobec- tomy) may be suitable as treatment plan. Older age (>45 years), contralateral thyroid nodules, a personal history of radiation therapy to the head and neck, and familial DTC may be criteria for recommending a bilateral procedure because of plans for RAI therapy or to facilitate follow-up strategies or address suspicions of bilateral disease (270,278,322,326).
The relationship between surgeon volume and patient outcomes has been studied extensively over the last 20 years. Institutional studies examining outcomes following thy- roidectomy by high-volume surgeons have been published demonstrating overall safety. In one of the first studies ex- amining the relationship between surgeon volume and thy- roidectomy outcomes at a state level, Sosa et al. (232) found a strong association between higher surgeon volume and
favorable patient outcomes, especially with regard to RLN injury and wound complications. This was especially pro- nounced for patients undergoing total thyroidectomy for thyroid cancer. Others have made similar observations (233,329,330). In a recent study of patients undergoing thyroidectomy in the Health Care Utilization Project Nationwide Inpatient Sample (HCUP-NIS), surgeons were divided into categories of low (<10 cases/year; encompassing 6072 surgeons), intermediate (10–
100 cases/year; 11,544 surgeons), and high volume (>100 cases/
year; 4009 surgeons) (331). Over 80% of thyroid resections were performed by low- and intermediate-volume surgeons. On average, high-volume surgeons had the lowest complication rates for patients who underwent total thyroidectomy for cancer at 7.5%; intermediate-volume surgeons had a rate of 13.4%, and low-volume surgeons, 18.9% ( p < 0.001).
From robust population-level data such as these, it can be concluded that referral of patients to high-volume thyroid surgeons is associated, on average, with superior outcomes. However, such referral is not always possible, given the relative scarcity of high-volume surgeons and their geo- graphic distribution. In addition, there are some data sug- gesting that other factors, such as surgeon age, should be considered (332). Therefore, conclusions at a population le- vel cannot always be applied to individual surgeons and pa- tient circumstances. It may, however, be reasonable to consider sending patients with more extensive disease and concern for grossly invasive disease to a high-volume sur- geon experienced in the management of advanced thyroid cancer.
It is worth noting that even high-volume surgeons have a higher overall postoperative complication rate when per- forming total thyroidectomy compared with lobectomy (333). Using the HCUP-NIS, these authors found that high- volume thyroid surgeons had a complication rate of 7.6% following thyroid lobectomy but a rate of 14.5% following total thyroidectomy. For low-volume surgeons, the compli- cation rates were 11.8% and 24.1%, respectively. Therefore, patients should carefully weigh the relative benefits and risks of total thyroidectomy versus thyroid lobectomy, even when surgery is performed by high-volume surgeons. | 312 |
[B8] Lymph node dissection | 313 |
RECOMMENDATION 36
Therapeutic central-compartment (level VI) neck dissection for patients with clinically involved central nodes should accompany total thyroidectomy to provide clearance of disease from the central neck.
(Strong recommendation, Moderate-quality evidence)
Prophylactic central-compartment neck dissection (ipsilateral or bilateral) should be considered in patients with papillary thyroid carcinoma with clinically unin- volved central neck lymph nodes (cN0) who have ad- vanced primary tumors (T3 or T4) or clinically involved lateral neck nodes (cN1b), or if the information will be used to plan further steps in therapy.
(Weak recommendation, Low-quality evidence)
Thyroidectomy without prophylactic central neck dissection is appropriate for small (T1 or T2), noninvasive, | 314 |
Subsets and Splits