An International Survey on Utilization of Five Thyroid Nodule Risk Stratification Systems: A Needs Assessment with Future Implications

Introduction

Over the past two decades, many organizations have developed ultrasound-based risk stratification systems and management guidelines for thyroid nodules, hereinafter termed RSSs. These include the American College of Radiology Thyroid Imaging Reporting and Data System (ACR TI-RADS), the American Thyroid Association (ATA) guidelines, the Chinese Medical Association (C-TIRADS), the European Thyroid Association (ETA, EU-TIRADS), the Korean Society of Thyroid Radiology/Korean Thyroid Association (KSThR/KTA, K-TIRADS), the Society of Radiologists in Ultrasound (SRU Consensus Conference Statement), and the American Association of Clinical Endocrinology (AACE), the American College of Endocrinology (ACE), and the Associazione Medici Endocrinologi (AME), the latter three societies having collaborated to produce a unified document (AACE/ACE/AME Medical Guidelines for Clinical Practice for the Diagnosis and Management of Thyroid Nodules) (1 –7). As well, several groups of investigators have developed RSSs that are not associated with a specific professional organization (8,9).

Multiple studies have compared RSSs to determine if some perform better than others, but the results have been inconsistent, probably reflecting differing patient populations, inclusion criteria, analytic methods, and other factors. For example, a meta-analysis of 8 studies that evaluated the performance of 5 RSSs found that they showed that the ACR TI-RADS resulted in fewer unnecessary biopsies than the ATA guidelines, K-TIRADS, EU-TIRADS, or the Kwak TIRADS (TIRADS proposed by Kwak) (10). However, other research has produced different results. For example, in one recent study, the ATA guidelines outperformed the ACR TI-RADS (11). Reported variations in performance make it difficult for individual practitioners and groups to know which RSS to employ for their patients.

In addition to performance, the choice of which RSS to use is influenced by other considerations (Table 1). Notably, the RSSs take somewhat different approaches to assigning nodules to risk categories based on ultrasound features. For example, the ATA guidelines include pattern matching, while the ACR TI-RADS awards points in five categories, which are then summed to produce a cancer risk level. Other systems, such as the EU-TIRADS and K-TIRADS, use an intermediate approach in which classification is based on the presence of key suspicious features in combination with others. The malignancy likelihood associated with each risk category also differ considerably between RSSs. Disparities between RSSs are further characterized by differing size cutoffs for fine-needle aspiration biopsy (FNAB) and recommendations for follow-up. Additionally, practitioners may be more likely to adopt an RSS that is endorsed by a professional society to which they belong, which may in turn relate to their geographic location.

Not surprisingly, these differences have led to confusion for medical professionals and patients. To help resolve this conundrum, a grassroots initiative managed by the steering committee of the International Thyroid Nodule Ultrasound Working Group (ITNUWG) is striving to develop an international RSS, tentatively termed I-TIRADS, that harmonizes the leading RSSs (12). To help guide this effort, which at this writing is nearing the end of its first phase, the ITNUWG conducted an international survey on RSS use to determine the factors that govern RSS choice and investigate why some physicians do not use an RSS at all. In this article, we summarize the results of this survey with the hope of fostering support from practitioners who deal with thyroid nodules.

Materials and Methods

Results

Discussion

In this international survey of self-selected practitioners with an interest in thyroid nodule risk stratification, 91% of respondents agreed or strongly agreed that there was value in RSS use. Not unexpectedly, we found that the choice of RSS was closely associated with medical specialty and geographic region. For example, the K-TIRADS is likely to be favored in the Republic of Korea, while the EU-TIRADS is bound to be more familiar to and adopted by European physicians. However, a surprising finding was that almost 31% of respondents reported use of more than one RSS in their practice. This suggests that in addition to variability in the choice of RSS related to location and specialty, there is also inconsistency within practices.

We speculate that this may occur when the physicians who comprise the practice prefer a particular RSS but referring physicians request that a different system be used for their patients. It is also possible an individual's RSS use relates to their training experience, and their choice may be different from that of their colleagues. Of the respondents using multiple RSSs, the most common combination was the ACR TI-RADS and ATA, particularly in American institutions. For example, radiologists in the United States may categorize nodules with the ACR TI-RADS as recommended by the ACR while simultaneously reporting them using the ATA guidelines at the request of endocrinologists and surgeons, who understandably want to avoid re-classifying them based on reported ultrasound findings.

More than half of the respondents expressed a desire for a universal lexicon to reduce interobserver variability, an issue that has also been highlighted by several investigators. A multicenter study found that interobserver variability for six features ranged from poor to substantial, with most characterized as moderate (13). Similarly, Hoang et al. found poor interobserver agreement for margin and punctate echogenic foci, leading them to recommend targeted education (14). Respondents also indicated support for an online atlas of ultrasound images and/or video clips to illustrate features.

The reasons given by the 271 respondents who indicated that they did not personally use an RSS were particularly instructive. Addressing their concerns will be critical to encourage adoption of a universal RSS. Notably, more than 60% preferred to use their own description of the ultrasound features they consider most relevant, rather than apply an RSS. However, this further contributes to confusion for patients and physicians caused by the use of multiple systems and guidelines, as well as leading to reduced efficiency and errors caused by re-classification.

It also complicates research that seeks to refine risk assessment based on grayscale ultrasound and other techniques, such as elastography. This is supported by a study by Solymosi et al. that compared FNAB recommendations based on personal expertise with a composite score derived from the five RSSs in our study (15). These investigators found that personal experience provided a more accurate diagnosis for thyroid malignancies overall, missing a lower number of small thyroid cancers, while TIRADS exhibited similar accuracy for clinically relevant or potentially aggressive lesions.

Even more concerning are variable recommendations for biopsy and follow-up imaging between RSSs. This issue has become particularly acute in recent years as patients increasingly access their own reports and wonder why a nodule might receive divergent management advice. This will be increasingly important as active surveillance and minimally invasive therapy for thyroid cancer management gain acceptance (16 –18).

Although we do not believe that a universal RSS will eliminate variations in management, which will always be guided by local practice and other considerations, we believe that it could serve as a foundation upon which such decisions can be made. Even more importantly, a widely endorsed system produced by an international panel of experts could substantially reduce the collective effort currently expended by multiple professional organizations to develop or revise their own guidelines. Additionally, a universal RSS could be revised more frequently to better keep pace with techniques such as ultrasound elastography that are not widely used now but could be adopted in the future.

Our study had several limitations. First, although we attempted to canvas a diverse group of physicians, these results may not be generalizable because most respondents were academic practitioners and/or endocrinologists. Second, we were unable to determine the response rate for the relevant population of practitioners since most of the participating societies offered the survey to all their members, many of whom may not see patients with thyroid nodules. As well, many organizations whose members are interested in thyroid nodules, such as the AACE and the American Association of Endocrine Surgeons, were not surveyed.

It was not possible to determine the response rate, as some respondents indicated membership in more than one society, and we did not ask to which organization's solicitation they were replying. Physicians with a greater interest in RSSs were more likely to reply to a survey titled “Risk Stratification Systems: Usage and Needs,” potentially biasing the results. However, our purpose was to explore the practice patterns and opinions of a group of stakeholders, rather than conduct formal hypothesis-driven research.

Conclusions

Despite these shortcomings, we believe that the results of this survey, particularly the concerns expressed by practitioners who do not use an RSS, will be valuable to inform the ongoing work of the ITNUWG. Respondents indicated strong support for a universal lexicon and image atlas, the development of which is part of the I-TIRADS initiative. As well, we hope the finding that more than one RSS is used in nearly one third of practices will foster interest in adopting a unified system that has the potential to reduce diversity in thyroid nodule classification, benefiting practitioners and patients alike.