European Perspective on 2015 American Thyroid Association Management Guidelines for Adult Patients with Thyroid Nodules and Differentiated Thyroid Cancer: Proceedings of an Interactive International Symposium

ATA 2015: Overview

Authored by 14 North American and two European experts, ATA 2015 is considerably expanded compared to ATA 2009. The total number of separate recommendations in ATA 2015 rose nearly 80% relative to the number in ATA 2009 (from 102 to 180). Moreover, according to ATA 2015, approximately 40% of the document's numbered recommendations are new (n = 21) or substantially changed (n = 19). The expansion is even greater when one considers that unlike ATA 2009, ATA 2015 pertains only to adults: the ATA's first management guidelines for children with thyroid nodules and DTC appeared earlier in 2015 (3).

Another notable change in ATA 2015 is separate grading of the strength of recommendations and the quality of the supporting evidence. Recommendations can be strong, weak, or absent. With one exception, the “no recommendation” classification is used only when data regarding the topic are believed to be “insufficient.” Strong or weak recommendations, by contrast, can be based on low-, moderate-, or high-quality evidence. This system led to ATA 2015 containing 28 “strong recommendations” based on “low-quality” evidence. This combination arguably affects most expert opinion regarding thyroid nodules and DTC, including that expressed here, and implies that more informative studies are needed regarding key issues in these settings.

Preoperative assessment of thyroid nodules

European experts addressed eight areas of divergent viewpoints regarding preoperative assessment of patients with thyroid nodules. First, ATA 2015 recommendations 2B and 2C were questioned. These recommendations respectively advocate for a radionuclide thyroid scan if serum thyrotropin (TSH) is subnormal and against such a scan if TSH is normal or elevated. European panelists noted that many of their countries have—or had—longstanding iodine deficiency and a relatively high prevalence of thyroid autonomy. In this situation, TSH values within the reference range may not always suffice to rule out thyroid autonomy in a substantial percentage of patients (4 –8). Scintigraphy, however, may be able to rule out hyperfunctioning nodules, thus ensuring exclusion of “hot” nodules for fine-needle aspiration (FNA). Hence, false-positive or indeterminate cytological results may be markedly diminished in iodine-deficient regions (9). Therefore, as a conservative approach to avoid overlooking thyroid autonomy, use of a TSH cutoff within the lower reference range might be considered to select patients for scintigraphy to rule out nodule autonomy, especially in iodine-deficient areas. This strategy would result in scans being routinely performed in patients with low-normal as well as abnormally low TSH; whether this approach would miss some autonomous adenomas with indeterminate biopsies is unknown. Scanning using technetium, which is inexpensive and widely available, would facilitate this strategy. Ultimately, a large study assessing the relationship (if any) between TSH levels and scintigraphic results would be very useful in elucidating the optimal TSH cutoff for routine scanning of thyroid nodules to rule out autonomy.

Second, European panelists favored routine serum calcitonin measurement in patients with thyroid nodules. Calcitonin screening permits earlier detection of medullary thyroid cancer (MTC), which may alter patient care, and European and other studies have suggested the strategy's cost-effectiveness (10 –14). ATA 2015 (in recommendation 4), as well as other European experts (15), take neutral positions on routinely measuring this analyte. The latter perspective may be influenced by the issue of indeterminate calcitonin measurements, which may lead to unnecessary anxiety or surgery, and which are particularly concerning in areas such as the United States, where calcitonin stimulation has long been unavailable in routine practice. An alternative, pragmatic approach to calcitonin testing, particularly in areas with high prevalence of MTC, could include routine calcitonin determination whenever patients are to undergo surgery for thyroid nodules.

In the text after recommendation 4, ATA 2015 alludes to a subgroup in whom elevated calcitonin could change management and in whom measurement of this analyte thus “may be considered.” This subgroup should perhaps be more clearly defined, for example as patients with apical nodules or other MTC risk factors.

To elucidate indications for calcitonin measurement, it will be of interest for centers that more widely apply the modality to publish their experience with the diagnostic performance of calcitonin testing, particularly relative to that of FNA biopsy and ultrasonography. Such reports would supplement and update older literature (13,16). Other useful publications might focus on the diagnostic performance of calcitonin testing in patients with benign nodular goiters and those with Bethesda III–V (indeterminate) biopsies. Additional cost-effectiveness studies are needed regarding calcitonin testing, as well as more outcome studies, particularly in the setting of medullary microcarcinomas.

A third divergence with ATA 2015 content on thyroid nodules involves recommendation 6 and the subsequent text. This content endorses routine ultrasonography of the thyroid and cervical lymph nodes in this setting, and suggests evaluation and reporting standards. European experts strongly agreed with these concepts, going beyond the ATA 2015 recommendation to emphasize that neck ultrasonography should be used in planning all thyroid nodule–related procedures. However, in our opinion, ATA 2015 might have contained more methodological detail regarding presurgical neck ultrasonography, for example the need for a high-resolution ultrasonographic system, a high-frequency (≥10 MHz) linear probe, and an experienced operator. Moreover, the guidelines might have highlighted the difficulty of analyzing the central compartment, and might have suggested including a detailed diagram of all lateral neck compartments in procedure reports. Such a diagram could perhaps be based on the reference diagram in the European Thyroid Association (ETA) guidelines for ultrasonography in patients with thyroid cancer (17). ATA 2015 also might have described in greater detail criteria for benign, indeterminate, or malignant neck lymph nodes (17,18). For example, beyond nodule size, important criteria for malignancy include loss of a central hilus, microcalcifications, irregular shape, and cystic composition.

European panelists concurred with ATA 2015 regarding the importance of ultrasonography for nodule risk stratification. However, ATA 2015 might have recommended for this purpose the Thyroid Imaging Reporting and Database System (TIRADS), due to its extensive validation (19,20), albeit recent literature (21 –26) suggests comparable diagnostic performance for the system introduced in ATA 2015 versus assessed TIRADS variants. TIRADS has high sensitivity and negative predictive value for malignancy: 95.7% and 99.7%, respectively, in a 4550-nodule study (27) using cytological results as the gold standard. Employing a TIRADS-based algorithm may reduce unneeded FNAs by an estimated 34% or more (27).

As alluded to, TIRADS comes in a number of variants (e.g., Tessler et al.) (28), including emerging and promising web-based versions. We would suggest use of the ETA “EU-TIRADS” system (20), which was published after the appearance of ATA 2015. EU-TIRADS includes an illustrated thyroid ultrasonography atlas and lexicon, a standardized report definition of each category of nodules, risk stratification of malignancy, and indications for FNA based on a clinical algorithm flow diagram. Together with the simplicity of the system, these user-friendly features should promote inter-observer reproducibility of descriptions and facilitate the presentation of results (20). Going one step further, an International Thyroid Nodule Ultrasound Working Group, representing major thyroid associations including the ATA, is developing the “U-TIRADS” universal guidelines on sonographic stratification of thyroid nodules, which will provide a unique stratification score.

A fourth area of divergence concerns indications for FNA (ATA 2015 recommendation 8). We recognize that few if any outcome data are available to provide guidance regarding this question. However, compared to North America, Europe has a much higher prevalence of iodine deficiency and nodular thyroid disease. Given the logistical and economic demands of managing a relatively prevalent condition, we favor less intense surveillance of patients with thyroid nodules, particularly more sparing application of FNA (Supplementary Table S1; Supplementary Data are available online at www.liebertpub.com/thy).

ATA 2015's advocacy (in recommendation 9) of a standard system to communicate thyroid nodule cytology results was welcomed. The recent update (29) of The Bethesda System for Reporting Thyroid Cytopathology is, in our opinion, an improvement over the previous version (30). To increase accessibility of cytology findings to clinicians, particularly nonspecialists, ATA 2015 might have recommended briefly describing clinical and imaging findings along with the Bethesda classification in cytology reports. This information would supplement the traditional description and interpretation of cytological findings, facilitating integration of cytological, clinical, and imaging findings into decision making regarding patient management.

A sixth area where European panelists believe that ATA 2015 nodule-related content merits fine-tuning involves recommendation 16, which addresses management of nodules with the Bethesda classification “follicular neoplasm/suspicious for follicular neoplasm (FN/SFN).” The recommendation might suggest that this classification prompt Doppler ultrasonography and review of previous sonograms before patients are sent for surgery (31). The text also might suggest that referral for operation generally occur only for patients with nodules with TIRADS 4C or 5 sonographic classification, with suspicious clinical findings, or with both, subject to the caregiving physician's clinical judgment.

European panelists generally agreed with recommendations 13–18 regarding molecular testing of FNA samples. However, to provide greater heuristic value, those recommendations might have been organized along the following lines:

For cytology samples considered “inadequate” by qualitative criteria, mutational testing may be informative (32), although the prevalence of malignancy (i.e., the pretest probability) is low. Thus, low-sensitivity tests may only marginally increase diagnostic yield. However, with “inadequate” samples, the current best strategy would appear to be to repeat FNA, to reevaluate sonographic features, or to do both (33,34).

Surgery and role of pathology

An important area of divergence regarding ATA 2015's recommendations on surgery and the role of pathology (Table 1) is recommendation 33A. This recommendation advocates preoperative cross-sectional imaging with intravenous contrast in patients with clinical suspicion of an invasive primary tumor or other advanced disease, or with clinically evident extensive lymph node involvement. Use of contrast agent in patients who are among the most likely to require RAIT was questioned. One example would be individuals with aggressive disease invading the aerodigestive tract, who often have low RAI uptake, even absent iodine excess (42 –44). The high iodine content of contrast agents would at least delay, and could impair the efficacy of, iodine-131 (¹³¹I) treatment (45). Although urinary iodine values may normalize within one month after iodinated contrast administration, it is unclear whether these values accurately reflect iodine accumulation by thyroid tissue, which could interfere with ¹³¹I uptake. For this reason, we believe that RAIT needs to be withheld for at least three months after contrast administration, a delay that might negatively affect outcomes, albeit no published evidence has yet addressed this issue. Evaluation of tissue planes and detection of local invasion may be improved by the use of computed tomography (CT) contrast. However, European panelists believe that magnetic resonance imaging (MRI) or CT without contrast can provide sufficient information regarding tumor and vascular and aerodigestive structures for effective surgical planning. Since these modalities definitely would not affect RAIT, they should, whenever possible, be preferred. For the same reason, MRI with non-iodinated contrast would also be preferable to CT with iodinated contrast.

Additionally, differences were expressed over the less extensive surgical approach favored by ATA 2015 regarding thyroid nodules <4 cm. The new guidelines permit withholding surgery or performing only lobectomy for most patients with thyroid cancer <1 cm without extrathyroidal extension, nodal involvement, or unfavorable history/characteristics (recommendation 35C). Also, the new guidelines deem lobectomy sufficient to treat low-risk papillary or follicular carcinomas >1 cm but <4 cm (recommendation 35B). The perspectives of continental European panelists regarding these recommendations were generally unfavorable, especially for tumors >2 cm but <4 cm. The experts stated that in countries with current or recent longstanding iodine deficiency, multinodular disease and bilateral involvement are frequent. This situation limits the applicability of ATA 2015, since such cases might not be sufficiently addressed by lobectomy. Also, patients undergoing lobectomy may run an appreciable risk of needing completion total thyroidectomy (46,47) due to the discovery of high-risk features only detectable postoperatively (48). A large retrospective analysis (N = 287) from an American tertiary referral center (49) suggested that as many as 43% of lobectomized patients may need completion thyroidectomy; the “real-world” incidence of cases requiring such surgery should be elucidated in a prospective study.

Additionally, although avoiding postsurgical thyroid hormone therapy is sometimes given as a rationale for less extensive resection, in Europe, up to 75% of lobectomized patients may ultimately be candidates for such therapy (50 –52). However, lobectomy does have the advantage of more frequently requiring only partial hormone substitution. This characteristic may decrease health risks of nonadherence to thyroid hormone therapy relative to risks encountered when complete substitution is required.

Another argument against lobectomy for low-risk papillary or follicular carcinomas >1 cm but <4 cm is the inability to apply subsequent RAIT. Omitting RAIT would eliminate that intervention's possible outcome benefits, and would eliminate RAIT's potential diagnostic contributions, for example, highly sensitive post-therapy scintigraphy and more sensitive follow-up thyroglobulin (Tg) determination and “diagnostic” scintigraphy (53 –55).

Further, important evidence cited by lobectomy advocates (and their opponents) comprises data from large databases. Such databases may be subject to material inaccuracies (56), and hence may be challenging to use to draw definitive conclusions. For example, the study by Adam et al. (57)—major evidence cited by ATA 2015 to support less extensive thyroid surgery—can be questioned on several counts. Adam et al. investigated 61,775 adults from the U.S. National Cancer Database who, between 1998 and 2006, underwent total thyroidectomy or lobectomy to treat 1–4 cm papillary thyroid tumors. After statistical adjustment, the researchers found no significant differences in overall survival between patients given the less extensive versus the more extensive procedure (recurrence rates for the subgroups were not reported). However, notwithstanding the statistical adjustments, the study was potentially biased by the significantly greater prevalence of more aggressive disease (multifocality, capsular invasion, lymph node involvement) in its total thyroidectomy subgroup. Additionally, 33.2% of lobectomized patients reportedly received RAIT despite substantial thyroid remnant, raising the possibility of procedure miscoding, nonadherence to guidelines, or unusual RAIT application. It should be acknowledged that large database studies favoring total thyroidectomy, for example that of Bilimoria et al. (58), have similar limitations to those of Adam et al.

In sum, the European panelists were reluctant to adopt the ATA 2015 approach regarding hemi-thyroidectomy unless (i) multinodular disease and bilateral involvement are clearly absent after well executed comprehensive examination and (ii) there is clearly no chance that RAIT might be indicated.

A third area where European panelists suggested modification of ATA 2015 was the guidelines' recommendations on lymph node dissection. Regarding part A of recommendation 36, which endorses therapeutic central compartment neck dissection, some European experts felt that the central compartment should be defined as level VI with/without level VII rather than as level VI alone. Also, some European surgeons favor central neck dissection as prophylaxis in ≥T1b papillary thyroid cancer, especially when the surgeon is experienced with the procedure. Lastly, European experts suggested that recommendation 37 might state that not only histopathological confirmation but also clinical or imaging-related suspicion of metastatic lateral cervical lymphadenopathy might warrant therapeutic lateral neck compartmental node dissection (59). However, no outcome data have yet been published favoring this strategy. Nonetheless, one reason for this perspective is that contemporary imaging of highly enlarged lymph nodes may reveal malignancy accurately enough to obviate the need for FNA and biopsy.

European panelists welcomed ATA 2015 recommendations regarding perioperative management of the voice and parathyroid glands. However, ATA 2015 indications for laryngeal examination were felt to be too conservative. According to ATA 2015 recommendations 41 and 44, such examination should be reserved for certain subgroups: preoperatively, patients with voice abnormalities, a history of cervical or upper chest surgery placing the recurrent laryngeal nerve or vagus nerve at risk, or known posterior extrathyroidal extension or extensive central nodal metastases; and postoperatively, those with voice abnormalities. European experts believe that a formal laryngeal exam should occur both pre- and postoperatively in every patient. Otherwise, up to 50% of vocal cord injuries may be missed (60,61). Early identification of vocal cord damage provides surgeons with feedback regarding short-term outcomes. Additionally, such diagnosis provides valuable input for decision making regarding re-operation, or regarding any needed remedial interventions, which may be more efficacious if undertaken sooner (62,63). Similarly, European surgeons tend to favor more frequent application of neuromonitoring during thyroid surgery than that suggested in ATA 2015 recommendation 42B, which weakly advocates that such monitoring “may be considered” to aid nerve identification and confirm neural function. As noted in ATA 2015's explanatory text, neuromonitoring may help preventing bilateral paresis, since the modality's detection of nerve paresis on the first side alerts the surgeon to avoid contralateral resection. However, the effects on outcome of intraoperative neuromonitoring versus no such monitoring when skilled surgeons are performing the thyroid surgery require further investigation (64).

European panelists welcomed ATA 2015's inclusion (on pages 32–33) of text noting the association between greater experience in thyroid surgery and lower complication rates (65 –69). European experts suggested investigating the association of surgeon experience with lymphadenectomy and outcomes of that procedure.

ATA 2015 was praised for more strongly emphasizing nodule and tumor morphology and topography and for showing less enthusiasm for molecular biomarkers for risk stratification than did ATA 2009. Emphasizing these points in clinical decision making gives even greater importance to clear, comprehensive ultrasonographic and pathology reports.

Also, there was agreement with ATA 2015 repeatedly noting the emerging utility of TERT (70 –73) and TP53 (74) mutations as potential unfavorable prognostic markers. Testing for such genetic alterations may be most appropriate in advanced disease, and findings regarding these mutations should be interpreted in the context of the BRAF mutation status. However, ATA 2015 arguably ascribes greater importance to tumor size and, regarding follicular thyroid cancer, perhaps lesser importance to capsular and vascular invasiveness as prognostic determinants than do European experts (75).

RAIT

European panelists differed with five main elements of ATA 2015's RAIT-related content (Table 2). The first element was ATA 2015's somewhat limited indications for high-activity postoperative RAIT of DTC (recommendation 51; Supplementary Table S2). European presenters discussed these indications based on ATA 2015's definitions of types of high-activity RAI administration following primary surgery (page 55). The new guidelines classify such administration by goal, as any of remnant ablation, meant to facilitate staging and detection of recurrent disease by Tg measurement or scintigraphy; adjuvant therapy, “intended to improve disease-free survival by … destroying suspected, but unproven residual disease, especially in patients at increased risk of … recurrence”; or therapy, “intended to improve disease-specific and disease-free survival by treating persistent disease in higher-risk patients.” In our opinion, it is questionable whether postoperative ¹³¹I should be administered for remnant ablation alone, especially given recent data from RCTs (76,77) and given improved ultrasound systems and Tg assays. In the absence of any other benefit, solely facilitating follow-up may not always warrant the risk of unwanted effects of RAI. These effects, although infrequent and mild relative to those of many oncologic therapies, can include an increased risk of a second primary malignancy (78,79).

However, as adjuvant therapy, postoperative high-activity ¹³¹I administration may offer one or more of recurrence-free survival, disease-specific survival, or overall survival benefits. The literature suggests such benefits in intermediate-risk patients as well as high-risk patients (80,81). As of yet, no formal head-to-head comparison of adjuvant RAIT versus “no postoperative RAI” has been published. However, two prospective, multicenter, RCTs—IoN from the United Kingdom (82) and ESTIMABL 2 from France (clinicaltrials.gov identifiers NCT01398085 and NCT01837745, respectively)—are comparing ¹³¹I, 1.11 GBq (30 mCi), versus no ¹³¹I in low-risk and, in the case of IoN, also intermediate-risk patients. We would favor retaining the longstanding, widely applied practice (83,84) of postoperative RAIT in low-risk and intermediate-risk DTC patients until these studies reach conclusions suggesting that practice should be changed.

The second main area of divergence was ATA 2015's suggestions of ¹³¹I activities for postoperative RAIT. For adjuvant therapy, ATA 2015 “generally favors” >1.11–5.55 GBq (>30–150 mCi; recommendations 55B and 56). For RAIT of known disease, 3.7–7.4 GBq (100–200 mCi) activities are mentioned in recommendations 77B, 78, and 79B to treat patients <70 years old who have iodine-avid pulmonary micrometastases, pulmonary macronodular metastases, or bone metastases, respectively; 5.55 GBq (150 mCi) is advocated as an upper limit when such patients are ≥70 years old (recommendations 73B and 77B).

However, we are influenced by substantial data, albeit single-center and somewhat controversial (85,86). These data favor higher versus lower initial activities and fewer higher-activity administrations versus more numerous lower-activity administrations, regardless of the patients' initial risk status. A large long-term analysis of the Würzburg DTC Database (N = 1298; follow-up ≥5 years per patient) (85) showed significantly higher DTC-specific survival (p = 0.002) in ETA low-risk patients ≥45 years old who received >2.0 GBq (>54 mCi) versus ≤2.0 GBq as the first postoperative activity. The analysis also found significantly higher recurrence-free survival rates and DTC-specific survival rates (p ≤ 0.001) when ETA high-risk M0 patients ≥45 years old (n = 265) received first postoperative activities >2.0 GBq (>54 mCi).

Another large analysis using the same database (N = 1229) (86), with a median follow-up of 9.0 years (min–max 0.1–31.8 years), showed that recurrence rates rose significantly (p = 0.001) together with the number of RAIT courses required, irrespective of the cumulative RAI activity, up to 22.2 GBq (600 mCi). The DTC mortality rate also increased significantly in parallel with the number of RAIT courses needed (p < 0.001). This finding was attributed to the high mortality rate after four RAIT courses. The need for many activities might be a surrogate for poorly RAI-responsive or RAI-resistant disease. Additionally, if complete response was attained with one course, recurrence rates and disease-specific mortality rates did not differ between initially ETA low-risk versus high-risk patients.

The divergent perspectives of European panelists from those of ATA 2015 regarding indications for and regimens of RAIT may be influenced by different case mixes and definitions of risk in Europe versus North America. RCTs of course represent the best way of resolving these issues.

A third European comment on ATA 2015's guidance on RAIT was that the language on rhTSH (Thyrogen, TSHα; Genzyme, Cambridge, MA) to stimulate postoperative RAIT for M0 disease seems overcautious. We refer to terminology such as “acceptable alternative” (recommendation 54A), “may be considered as an alternative” (recommendation 54B), “should be considered” (recommendation 54D), and “may be indicated in selected patients” (recommendation 75). Use of such terminology may be influenced by the regulatory situation and reimbursement models in the United States, where rhTSH is not approved for use in therapy of DTC, and by a desire for “more controlled data from long-term outcome studies” (recommendation 54C).

Nonetheless, European panelists felt that a stronger endorsement of rhTSH preparation is warranted in most settings, as is found in the EANM guidelines for RAIT (87), European Society for Molecular Oncology thyroid cancer guidelines (88), or a European expert consensus statement (89) (Supplementary Table S3). Among the rationales for this position was the superior quality of life and decreased hypothyroid morbidity associated with rhTSH versus thyroid hormone withdrawal (THW). These findings have been widely documented in RCTs and other clinical studies (90 –94) as well as patient surveys (84,95,96). It also was noted that a long-term European study (10-year follow-up) observed no outcome differences in a mixed group of ATA 2009 low-, intermediate-, and high-risk patients given rhTSH-stimulated or THW-mediated postoperative RAIT, or in the respective subgroups (97). Similar results were seen in a medium-term U.S. study of ATA 2009 intermediate- and high-risk patients (98).

A fourth European difference with ATA 2015's section on RAIT was the guidelines' lack of a stance favoring dosimetry (99 –103) versus “empiric” fixed activity use in ¹³¹I treatment of advanced locoregional or distantly metastatic disease. ATA 2015 recommendation 73 acknowledges “theoretical advantages” of dosimetric approaches versus empiric activity use. However, likely because no RCT yet has been published comparing the strategies, ATA 2015 terms the evidence insufficient to recommend one or the other. Nonetheless, we find a retrospective analysis involving 87 patients treated for locoregional or distant metastatic, RAI-avid DTC, or both, with a 51 ± 35-month mean follow-up convincing (104). In this study, dosimetrically treated patients (n = 43) were significantly more likely (p = 0.029) to attain a complete response according to Response Criteria in Solid Tumors 1.1 then were their “empirically treated” counterparts (n = 44). Additionally, the dosimetrically treated group showed a trend (p = 0.052) toward a lower likelihood of progression. At the same time, the frequency of RAIT side effects did not differ between groups. Additionally, a more recent study by Deandreis et al. (105) arguably may be viewed as additional evidence favoring dosimetric approaches as well as higher cumulative activities. In the study by Deandreis et al. (104), a dosimetric approach achieved similar outcomes in a U.S. series with more advanced disease compared to lower cumulative “fixed” activities used in a French series with less advanced DTC (106).

Lastly, modification regarding the text on “blind” RAIT in patients with substantially elevated or rapidly rising Tg, or increasing anti-Tg antibody levels but negative “diagnostic” imaging (recommendations 80 and 81) in ATA 2015 was suggested. Recommendation 81 notes “the risk of high cumulative administered activities” and “uncertain benefits” of RAIT in this setting. However, dosimetric data suggest that lesions detectable only with high ¹³¹I activities will mostly receive an insufficient therapeutic dose (107,108). Therefore, in the majority of cases, the benefits of blind high-activity administration seem to be mainly diagnostic. ATA 2015 might have alluded to this situation, which we believe argues for highly selective application of “blind” RAIT.

Assessment of initial therapy and dynamic risk stratification

European participants largely agreed with ATA 2015 content regarding assessment of initial therapy and dynamic risk stratification—concepts co-pioneered by European investigators (109). The guidelines note (on page 47, in explanatory text regarding recommendation 49) the challenge of obtaining adequate information for risk stratification from biochemical testing and imaging. Nonetheless, European panelists would have preferred greater acknowledgement in ATA 2015 of the frequent lack of availability in “real-world” clinical practice of data necessary for complete initial postoperative classification of persistent/recurrent disease risk according to the modified ATA 2009 system, routine use of which is advocated in ATA 2015 recommendation 48. For example, a recent two-center quality assurance survey (110) not infrequently found uncertain cervical lymph node status due to lack of surgical exploration or missing data on capsular infiltration. At the symposium, the need for complete pathological reporting of the number and size of involved lymph nodes was emphasized. This need has increased, since these variables have become major delineators of low- versus intermediate-risk status in ATA 2015. ATA 2015 might have offered guidance on managing patients with significant data gaps. It would have been helpful to include the point made at the symposium by one of the U.S. experts that risk stratification should not be based on guesswork. In patients with important unknown characteristics, clinical judgment must play an especially prominent role in treatment decisions.

European investigators noted that ATA risk stratification schemes derive largely from data from patients treated more intensively than is advocated in ATA 2015, albeit new literature is appearing regarding such schemes in less intensively treated patients (111). It also was suggested that ATA risk categories more precisely align with specific values for recurrence/persistent risk: <5% for low risk, 5–30% for intermediate risk, and >30% for high risk.

Additionally, we propose that future ATA guidelines address the risk stratification of follicular thyroid carcinoma in greater detail. There is a particular need for evidence and consensus regarding two major points: (i) whether a useful staging system is possible according to the number of veins invaded, and (ii) whether there is a significant prognostic difference between “encapsulated with angioinvasion” and “widely invasive” classifications.

Regarding follicular carcinomas, one approach could be to align risk categorization with these tumors' classification in the fourth edition of the World Health Organization handbook (112), albeit the outcome benefits of using this scheme have not yet been established. According to this classification, minimally invasive follicular carcinomas without angioinvasion are low risk, and those with angioinvasion are intermediate risk. Currently, some authorities also risk stratify follicular carcinomas according to the degree of angioinvasion, with higher risk assigned to patients with more extensive angioinvasion (typically more than four vessels invaded).

Further, ATA 2015 Table 11, describing disease characteristics associated with ATA risk classifications, does not mention poorly differentiated thyroid carcinoma (PDTC). Because this entity sometimes evolves from DTC rather than presenting separately, PDTC perhaps could be added to the high-risk section of Table 11.

Treatment of persistent disease, recurrences, and advanced thyroid cancer

European panelists proposed a number of refinements in wording or content of ATA 2015 recommendations regarding treatment of persistent disease, recurrences, and advanced DTC. Several changes were suggested for recommendation 71, which addresses the optimal directed approach to patients with suspected structural neck recurrence. Panelists felt that it might be preferable to use size ranges plus other criteria rather than single absolute sizes (i.e., ≥8 mm for involved central neck nodes and ≥10 mm for involved lateral lymph nodes) to indicate compartmental dissection. This preference is based on the only moderate-quality evidence supporting single absolute values as cutoffs, on the inability of current imaging techniques to determine exact tumor dimensions on a millimeter scale accurately, and on the favorable long-term outcomes reported for such surgery, albeit in highly skilled hands (113). In our opinion, the number of recurrent neck nodes should also factor into surgery-related decision making, as should the ratio of involved to excised nodes during the initial lymph node dissection, if performed, and perhaps patient age. Further, disease trajectory and surgical comorbidity risks should be considered.

Additionally, given the frequently heavy pretreatment of the patients to whom it pertains, recommendation 71 itself, rather than the explanatory text, might mention focal modalities such as percutaneous alcohol ablation as experimental alternatives to surgery in selected patients (114 –118).

Several suggestions were made regarding ATA 2015's guidance on RAI-refractory disease, recommendation 91. First, European experts believe that the concept of RAI-refractory disease is often relative rather than absolute. For example, one of the recommendation's definitions of such disease is “radioiodine … concentrated in some lesions but not in others.” However, ¹³¹I may have a palliative role in this situation, for example when RAI-avid lesions cause symptoms. Hence, ATA 2015 might best avoid blanket injunctions against further RAIT in the “RAI-refractory” setting.

Indeed, ATA 2015 might even portray RAI-refractory disease as a “moving target” in light of the promising albeit preliminary results of clinical trials of tumor re-sensitization to RAI, for example using selumetinib (119). However, this option should be considered, and requires further research, in adequately sized multicenter, controlled studies. European panelists deemed re-sensitization to be keenly interesting because, in their opinion, RAI was the first, and continues to be the best, targeted therapy of inoperable DTC.

Additionally, some European experts suggested that the recommendation regarding RAI-refractory DTC might benefit from alluding to 22.2 GBq (600 mCi) as a possible cumulative activity threshold beyond which RAIT may be unlikely to cure DTC distant metastases (86,120). However, other panelists disagreed, believing that ¹³¹I thresholds reflect retrospective epidemiological analyses rather than tumor biology.

ATA 2015 recommendation 92A was greeted enthusiastically by the European presenters, and indeed should perhaps be a strong rather than a weak recommendation. This guidance identifies RAI-refractory metastatic DTC patients who can receive protracted follow-up without systemic intervention beyond thyroid hormone suppressive therapy, so long as symptoms and evident progression are absent. Recent work on prognostic factors in RAI-refractory DTC could support identification of such patients (121). Describing RAI-refractory patients with a more favorable prognosis could help curb haste to give sometimes poorly tolerated, relatively expensive kinase inhibitors at any sign of disease progression. Such haste could arise in physicians unaccustomed to the sometimes indolent pace of even advanced DTC, or in patients not counseled to view advanced DTC as a manageable chronic condition.

European experts agreed with ATA 2015 recommendation 92B, which advocates against routine mutational testing for prognostic purposes in advanced DTC patients. Little evidence supports the prognostic value of mutational analysis, which tends to be expensive (immunostaining for BRAF being one exception). Nonetheless, some modifications to this recommendation can be proposed. For one, the text might suggest Tg rate of change in well differentiated tumors, for example doubling time (121 –124), or fluorodeoxyglucose avidity (125,126) as potentially useful prognostic indicators.

Regarding recommendation 93, which offers guidance on directed therapy of advanced DTC, based on their clinical experience, the panelists suggested that part A, describing radiofrequency ablation, cryoablation, and stereotactic radiation as “valid alternatives” to surgery, could perhaps be strong. Panelists also suggested that bronchial stenting or bronchial laser therapy and other palliative focal approaches could be mentioned.

European presenters proposed an addition to recommendation 94, which addresses treatment of brain and other central nervous system (CNS) metastases. Namely, the recommendation could mention rhTSH preparation of RAIT as another (although “off-label”) option possibly to lessen the risk or magnitude of TSH-induced tumor expansion or RAI-induced inflammatory response. Compared with THW, rhTSH sometimes (127 –130), if not always (131), mitigates these phenomena in patients with distant DTC lesions, presumably because rhTSH is associated with much shorter-lived TSH elevations (132). However, when rhTSH is used, care regarding tumor expansion must still be taken, and corticosteroid premedication should be applied (131). rhTSH injection also allows earlier RAIT, since it raises TSH concentration much faster than does THW.

As a minor point regarding recommendation 96, which provides guidance on kinase inhibitor treatment of advanced DTC, one may question the strength of recommendation/quality of evidence for its part A. This recommendation states that kinase inhibitors should be considered in RAI-refractory DTC patients with metastatic, rapidly progressive, symptomatic, and/or imminently symptomatic or imminently life-threatening disease. Although kinase inhibitors were approved in some countries based on Phase II trials (133), the benefits of these drugs have been established in RCTs (134,135). The evidence therefore would appear to be of high quality, and the recommendation might perhaps be strong rather than moderate.

Sequential administration of two kinase inhibitors is the topic of recommendation 97. That text identifies “patients who incur disease progression in response to initial kinase inhibitor therapy without prohibitive adverse effects” as candidates for a second-line kinase inhibitor. This recommendation might provide advice/direction regarding timing of kinase inhibitor discontinuation and regarding mixed responses to kinase inhibitors, for example disease control at all sites but bone, albeit such information would currently be based only on expert opinion.

European presenters welcomed the inclusion of recommendation 98, strongly advocating “active surveillance and timely intervention in response to emergent [kinase inhibitor] toxicities.” They felt, however, that the recommendation might advise inclusion of medical oncologists and specialist nurses in coordinated teams to manage kinase inhibitor therapy, particularly toxicities and adverse events. It would also be beneficial to offer guidance by teams experienced with these drugs in the setting of DTC; such recommendations would facilitate prescribing/managing kinase inhibitor therapy.

Regarding recommendation 100, which advocates “consideration” of cytotoxic chemotherapy as salvage therapy for very advanced, refractory DTC, the text seems unclear as to whether traditional or new agents are meant. Clarity might be improved by using the phrase “newer cytotoxic chemotherapy agents or newer combinations of chemotherapy agents” (e.g., gemcitabine plus oxaliplatin) (136) and adding text stating that “conventional” cytotoxic chemotherapy is not recommended in DTC.

European panelists also suggested adding content to recommendation 101 on bone-directed agents. The recommendation itself might give guidance on the regimens and duration of therapy for the agents mentioned therein.

Future directions

European panelists suggested that ATA 2015 might have placed more emphasis on the further investigation of two modalities. First, ultrasound elastography might perhaps have been noted as an adjunctive method worthy of study in nodules with an FN/SFN Bethesda classification, as some but not all reports (137 –142) suggest promising results.

Second, based on preliminary data, modalities for molecular imaging of nodules (143 –149) beyond fluorodeoxyglucose-positron emission tomography may represent comparably accurate and less costly alternatives to DTC-associated mutation analysis, and would appear to warrant further exploration.