Abstract

Clinical guidelines are recommendations for clinical practice produced by experts in the field based on their interpretation of the best available evidence. If all clinical questions were appropriately addressed by adequately powered, randomized controlled trials, we would not need guidelines. A list of references would suffice.
Guidelines are not gospel. Unlike the Gospels, they are updated periodically and modified more frequently as new information becomes available. However, like the Gospels, guidelines are also (to some extent) faith based. We are all influenced by our mentors and tend to have faith in their teachings. We often prefer (or have faith in) research studies that support our point of view and practice habits, particularly when we have contributed to that research.
Our (primarily) Europeans colleagues (EC) (1) have formulated a critique of the 2015 American Thyroid Association (ATA) guidelines for adult patients with thyroid nodules and differentiated thyroid cancer (DTC; henceforth the ATA guidelines) (2). In 2015, the European Association of Nuclear Medicine convened a symposium to explore the differences in opinion between European experts and the ATA Guidelines Task Force. Of the 17 European co-authors of this perspective, six are nuclear medicine specialists. In contrast only 1 of the 16 ATA guidelines co-authors is a nuclear medicine physician (a European who is also trained in endocrinology and oncology). Unlike the United States, where almost all patients with thyroid cancer are under the care of endocrinologists, in parts of Europe, either an endocrinologist or a nuclear medicine physician directs the care of thyroid cancer patients. Although four ATA guideline co-authors are also co-authors of this critique, their role was to ensure appropriate interpretation of the ATA guidelines, not to endorse the critique.
As noted by the authors of this critique, of the 180 ATA guideline recommendations, 28 are considered “strong recommendations” based on “low-quality” evidence, and many other recommendations are also based on “low-quality” evidence. Therefore, it is not a surprise that the authors of this critique do not agree with all of the ATA recommendations. Yet, some of their views suffer from this very same limitation. We assume that it is also quite likely that the co-authors who represented the ATA do not uniformly agree with the final recommendations offered in the ATA guidelines or this critique. That our EC disagree with almost one third of the recommendations may seem surprising until some of the areas of disagreement are analyzed more closely.
Some of the disagreements are relatively minor or even qualitative.
For example, while agreeing with the central role of ultrasound (US) in thyroid nodule evaluation, our EC would prefer a more rigorous description of US requirements and of the nodes in the central neck; they agree that US is useful for individual nodule risk stratification but prefer the European Thyroid Imaging Reporting and Data System (TI-RADS) system to the ATA US criteria. This seems a matter of personal preference, and direct comparisons of various TI-RADS systems have shown relatively good agreement (3). It should also be mentioned that there is an ongoing international effort to align the various systems. Moreover, they would like more emphasis on the utility of US characteristics as an adjunct to the Bethesda classification.
They also agree with the use of US for preoperative thyroid cancer node staging but are comfortable diagnosing malignancy in some nodes, based on US criteria, without necessarily requiring a biopsy.
Considering the limited space devoted to follicular thyroid carcinoma (FTC) in the ATA guidelines, they recommend greater detail in addressing the risk stratification for FTC.
They also caution against over-reliance on large national databases for therapeutic decision making due to the inability to verify information about individual patients in those databases.
They agree with the thyroid cancer risk assessment approach but are concerned that some of the information necessary for risk stratification may not be available “in the real world.” This seems more like a criticism of the “real world” than the guidelines.
They strongly recommend routine vocal cord examination before and after thyroid surgery to provide individual surgeons with precise data about the risk of recurrent laryngeal nerve injury from their own operations. The ATA guidelines recommend selective vocal cord examination. It is uncertain whether individual patient outcomes would benefit from routine vocal cord examination.
Some of the disagreements are based on regional differences. There is a higher prevalence of thyroid nodules, multinodular thyroids, and thyroid nodular autonomy in Europe compared to the United States because of the persistence of mild iodine deficiency in some countries.
Given the higher prevalence of thyroid nodules in Europe, less intensive use of fine-needle aspiration (FNA) and less intensive follow-up for patients with nodular thyroids is recommended in certain countries.
The ATA guidelines recommend a radionuclide scan (preferably utilizing 123I) for patients with subnormal serum thyrotropin (TSH) to exclude “hot” nodules because these nodules do not require an FNA. Based on the higher prevalence of thyroid autonomy, our EC argue for more liberal use of thyroid scanning (including utilizing Tc-99m pertechnetate) in patients with low normal rather than frankly low serum TSH concentrations. They are concerned that some indeterminate (Bethesda III or IV) thyroid nodules are autonomous (“hot”) nodules without a suppressed serum TSH. This is a readily testable hypothesis.
The authors caution against unilateral lobectomy for thyroid cancer for two reasons: one is the high prevalence of thyroid multi-nodularity and multifocality of papillary thyroid cancer (PTC), and the other relates to radioactive iodine (RAI) therapy (see below).
The authors continue to advocate for universal calcitonin (Ct) screening in patients with nodular thyroid disease to promote early detection of medullary thyroid carcinoma (MTC). In contrast, other European experts consider Ct screening of uncertain benefit (4), whereas the ATA guidelines remained agnostic about the benefits of this approach (2). The specificity of an elevated serum Ct <100 pg/mL to diagnose MTC is improved with pentagastrin stimulation. However, pentagastrin is no longer available in the United States. Given the more problematic current supply of pentagastrin in Europe, it will be important to monitor the continued level of enthusiasm for this approach. The authors propose a potentially important, readily testable alternative approach: measure Ct on all patients with nodular thyroids who require surgery. The prevalence of a change in surgical approach based this protocol will be of great interest.
Next, it is not a surprise that many of the disagreements relate to nuances in the use of RAI for DTC.
The ATA guidelines distinguish between RAI remnant ablation, RAI adjuvant therapy, and RAI therapy for known residual/metastatic disease; RAI remnant ablation is generally not favored. Our EC argue strongly against RAI remnant ablation, a position championed by many in the United States as well (5).
With respect to adjuvant therapy, our EC favor higher rather than lower activities.
While acknowledging that “blind” high-dose RAI therapy might serve a diagnostic purpose in the setting of an elevated serum thyroglobulin (Tg) with a negative low-dose RAI scan, the EC caution that it is unlikely to deliver sufficient RAI activity to be therapeutic. The ATA guidelines consider the question of blind therapy in a similar context but do not specifically address the role of a low-dose RAI scan to help decide whether a therapeutic dose of RAI might be beneficial.
Our EC and others (6) advise that the term “RAI refractory” is rarely absolute; palliation may occur in patients who are “RAI refractory.”
The EC are concerned about preoperative computed tomography (CT) scans with contrast prior to thyroidectomy for DTC, particularly when treatment with RAI is considered urgent. They prefer magnetic resonance imaging with contrast or CT scans without contrast. They are appropriately concerned that a low urine iodide one month after intravenous contrast may not translate into successful RAI therapy, particularly if excess iodine accumulates in thyroid cancer cells.
Our EC express a strong preference for recombinant human TSH (rhTSH) preparation for RAI therapy. They acknowledge that rhTSH is licensed in the United States for RAI remnant ablation but not specifically for thyroid cancer therapy.
Our EC strongly favor dosimetry over empirical dosing of RAI. Based on the dosimetry experience at the Memorial Sloane Kettering Cancer Center (MSKCC) and the fixed-dose RAI (generally 100 mCi/3.7 GBq) at the Institute Gustav Roussy, Deandreis et al. consider these therapeutic alternatives to be equivalent (7). Our EC reinterpret these data as favoring the dosimetry group, given the similar outcomes, despite more advanced disease in the MSKCC cohort. In fact, we stipulate that only a randomized controlled trial comparing these two modalities will provide a definitive answer.
Our EC are greatly concerned about the ATA guideline endorsement of lobectomy for many patients with DTC because this surgical approach precludes immediate RAI therapy. They reference the armchair exercise of Kluijthout et al. who predict that RAI therapy would have been recommended for 26% of patients, requiring completion thyroidectomy in 20% of patients “eligible” for lobectomy (8). It is uncertain what percentage of these patients would actually benefit from RAI. It is also unknown what percentage would require completion thyroidectomy in a prospective study, as acknowledged in the critique. Excellent outcomes in patients undergoing lobectomy alone were previously reported, although those data predated our current ability to evaluate recurrences by Tg and US monitoring (9,10). It is undeniably inconvenient for patients to undergo completion thyroidectomy (or lobe ablation), but this approach is still available after lobectomy if RAI is deemed necessary. It is difficult to weigh the inconvenience of completion thyroidectomy versus the complication rate of more universal total thyroidectomy, particularly when less experienced surgeons perform the surgery. In this context, it must be emphasized that only about 20% of all thyroidectomies in the United States are performed by high-volume surgeons and that the rate of complications is significantly higher among low-volume surgeons (11). Presumably, these numbers are similar in Europe, although we are not aware of formal data on this topic.
Although not mentioned in either the ATA guidelines or the critique, RAI for thyroid cancer is generally administered to outpatients in the United States, whereas hospitalization is required for most patients in Western Europe. Although formal studies addressing the impact of differences in regulatory requirements are lacking, these differences seem to influence certain practice patterns.
However, the fundamental and central area of disagreement between the two groups concerns RAI therapy for DTC. The ATA favors a more limited approach, whereas the EC authoring this critique favor a more permissive/less restrictive approach. The centrality of this disagreement is reflected in other commentaries by some of these authors who express surprise, shock, and even dismay (12), at times without rising to the level and tact of constructive professional criticism, about the (apparently) new position recommended by the ATA (13). The long-standing refractory nature of this debate is exemplified by the disparate opinions of two legendary thyroid experts reviewing the same patient database at the University of Michigan. Beierwaltes noted that “there is no question today that we should ablate normal thyroid tissue as part of the treatment of well-differentiated thyroid carcinoma” (14), whereas Sisson counter-argued that “the aggregate of evidence does not convincingly demonstrate that ablation of small remnants—and especially those remote from the primary tumor—lowers the rate of recurrent tumor” (15).
A brief overview of the evolution of RAI use in the United States might be helpful. In the early 1940s, RAI was initially utilized for physiological studies and therapy of hyperthyroidism. Based on its exquisite targeting ability, apparent efficacy and generally low toxicity, RAI rapidly became the treatment of choice for metastatic DTC (16). However, RAI was never subjected to the extensive scrutiny that would currently be required for approval by the Food and Drug Administration or other regulatory agencies. It may surprise some younger readers to learn how infrequently RAI was used for DTC in the United States in the 1960s and 1970s. For example, at the Mayo Clinic, after bilateral apparently curative PTC surgery, only 3–4% received RAI between 1966 and 1971, and 8–9% received RAI between 1972 and 1975 (17). Nodal metastases were common in this patient population, whereas papillary thyroid microcarcinomas (≤1 cm) were less prevalent in surgical series than today. The infrequent use of RAI was supported by the 20-year disease-specific survival of 99% for PTC patients with low MACIS scores (85% of all PTC) who did not receive RAI (18).
In 1977, Mazzaferri et al. published his landmark PTC study (19), which reported a dramatically decreased recurrence rate for patients treated with RAI compared to other therapeutic approaches. The study was a retrospective analysis of 576 PTC patients treated at a number of U.S. Air Force Hospitals by a large number of (primarily general) surgeons using a variety of surgical approaches. There was no attempt to control for disease stage in the various therapy groups. It is presumably not coincidental that 33% of Mayo Clinic PTC patients received RAI between 1976 and 1979, although data from the Mayo Clinic did not necessarily support that approach. In 1981, Mazzaferri et al. reported the 10-year follow-up on the same PTC cohort, noting a beneficial effect of RAI on recurrences and mortality for tumors >1.5 cm (20). By 1980–1981, 52% of Mayo Clinic PTC patients received RAI, and by 1984, this reached 69%. In 1981, Mazzaferri et al. recommended total or near-total thyroidectomy followed by RAI for all PTC >1.5 cm and for those that were bilateral, metastatic, or locally invasive, as well as for all FTC (21). For the next several decades, the Mazzaferri approach was practiced in many centers throughout the United States for most DTC patients.
Over time, it became apparent that many DTC patients with an excellent prognosis received RAI with little evidence of benefit and with evidence for potential harm. For example, based on the Surveillance, Epidemiology, and End Results database, 3.3% of patients with T1N0 (tumor <2 cm, node negative) PTC received RAI in 1973, whereas 38.1% of these patients received RAI in 2007 (22). The mortality from T1N0 PTC is essentially nil. Between 1973 and 2007, a dramatic increase in salivary gland malignancies and leukemia was noted in the RAI treatment group compared to no RAI, particularly in patients <45 years of age (22). Given the known, albeit small, increased risk of second primary malignancies after RAI therapy (23) and the possible increased risk of clonal hematopoiesis (a potential leukemia precursor) after RAI (24), it is essential that we treat patients who might benefit from RAI based on the principle of primum non nocere. Although most patients with DTC who receive RAI do well, the pertinent question is whether the outcome is due to or independent of the therapy.
Like a pendulum, practice patterns swing widely in opposite directions as new information appears or older data are reanalyzed. The ATA clearly believes that the pendulum had swung too far in the direction of RAI overuse. The authors of the critique strongly disagree and would prefer to wait for the results of ongoing randomized controlled trials before altering their practice patterns. There is evidence that the pendulum is swinging toward more limited RAI use in the United States (25). Whether it will swing too far in the opposite direction remains to be determined.
The ATA guidelines do not provide specific RAI recommendations for individual patients. They recommend giving RAI to most patients in the ATA high risk of recurrence group, generally not giving RAI to those in the ATA low risk of recurrence group (which includes low-volume nodal metastatic disease), and considering RAI for the patients in the ATA intermediate-risk group (which includes patients with intermediate-size nodal metastases). The authors of the critique express general disagreement with the ATA approach, but it is uncertain whether the disagreement is global (i.e., favoring treating almost all DTC patients with RAI) or more specific, related to individual patient characteristics. Moreover, it is by no means certain that there is unanimous consensus with the views offered in this critique within Europe. It would be of great interest to compare how ATA guideline members and our EC would treat specific DTC patients. However, there is a strong suspicion that the disagreement is of a fundamental nature (see below).
The European Thyroid Cancer Taskforce (ETCT) (26) described three DTC risk groups. The very low-risk group is comprised primarily of papillary thyroid microcarcinomas for which RAI is not indicated. In many centers, some of these patients might currently forgo surgery in favor of active surveillance (27). RAI is considered definitely indicated for the ETCT high-risk group: documented persistent disease or at high risk of recurrent disease, including any lymph node metastases and/or microscopic or gross extrathyroidal extension. The low-risk group includes patients with tumors >1 cm, which do not fall into the high-risk group. Considering the use of RAI for the low-risk group, the text notes that there is “no consensus: benefits are controversial and there are still uncertainties about whether it should be administered to all patients or only to selected patients” (26). However, in the accompanying table, it is noted that most low-risk patients will receive RAI. Some of the co-authors of this critique (13) strongly support this consensus report, implying that most DTC patients with tumors >1 cm should receive RAI.
We believe that appropriate analyses of the benefits of RAI require a clear distinction between initial risk category, recurrent/persistent disease, and distant metastatic disease/disease-specific mortality (DSM). For PTC, the ratio of recurrent/persistent disease to mortality is high, with a low risk of DSM. Most PTC recurrences are nodal. As also noted by some of our EC (28,29), it is possibly that surgical skill is inversely proportional to the risk of recurrent/persistent disease and possibly the chance of a beneficial response to RAI. The significance of some recurrent/persistent disease is uncertain, as ever smaller, potentially biopsiable nodal metastases are discovered on US and as we follow a large reservoir of DTC patients with biochemical incomplete responses (based on serum Tg) with apparently excellent outcomes (30). On the other hand, patients with distant metastatic disease have a high risk of DSM and are of greater concern. We believe that almost all deaths from DTC will come from those in the ATA high and intermediate risk of recurrence groups (31). We also strongly support separate outcome analyses for PTC, FTC, and Hürthle cell thyroid carcinomas.
What does the future hold? We all eagerly await the results of ongoing appropriately powered randomized controlled trials of RAI for DTC. However, even these trials may not provide the final answers. We need additional studies analyzing the molecular landscape of DTC to help predict the outcome of RAI therapy. For example, it is already evident that BRAFV600E -mutant PTC are less likely to be RAI avid than BRAF wild-type tumors (32). Redifferentiation therapy holds the promise of improved RAI efficacy by targeted therapy designed to increase sodium–iodide symporter (NIS) expression and membrane insertion, and hence RAI avidity, but requires additional study (33 –35). We also hope to refine our ability to determine which patients actually need RAI by predicting persistent/recurrent disease that is likely to be progressive. In this regard, the current use of postoperative serum Tg and the future use of liquid biopsies (including circulating tumor RNA, DNA, or tumor cells) may be informative (36). It is also important to try to predict which patients have an increased risk of second primary malignancies after RAI so as potentially to avoid RAI in those patients (24). The therapeutic environment is also changing. As we develop newer targeted DTC therapies with improved efficacy and diminished toxicity, the central therapeutic role of RAI may, in part, be challenged.
Practice guidelines provide practical advice for clinicians, but also serve additional important secondary functions. They identify and focus attention on gaps in our knowledge and promote research to close those gaps. They stimulate appropriate criticism and encourage collaborative efforts to reconcile differences of opinion. By all criteria, the ATA guidelines have been highly successful. This current critique (1) and the ongoing International collaboration of endocrinologists and nuclear medicine physicians initiated in Martinique (37) provide additional evidence of that success. In this context, we are reminded of the words (likely inaccurately) attributed to Mark Twain: “It ain't what you don't know that gets you into trouble. It's what you know for sure that just ain't so.” With the benefit of enlightened criticism and additional research, we hope to learn “what we know for sure” that is so. Finally, we are convinced that there are indeed many nuances and shades of gray, that there is a need for an ongoing dialectic process to discuss controversies in a professional way, and that all practice recommendations should be subject to continuous reflection and refinement.
