Trends in Diagnosis of Noninvasive Follicular Thyroid Neoplasm With Papillarylike Nuclear Features and Total Thyroidectomies for Patients With Papillary Thyroid Neoplasms

Key Points

Question  What has been the result of 2 recent efforts to prevent overtreatment of papillary thyroid carcinomas (PTCs): the new diagnostic category, noninvasive follicular thyroid neoplasm with papillarylike nuclear features (NIFTP), and the change in guideline support for hemithyroidectomy for selected PTCs up to 4 cm in size?

Findings  In this cohort study of 3368 pathology records from 18 hospitals in 6 countries during 2015 and 2019, 4.8% of papillary thyroid neoplasms were diagnosed as NIFTP in 2019. The proportion of eligible PTCs treated with hemithyroidectomy increased between 2015 and 2019.

Meaning  Overtreatment prevention strategies have had mixed success: NIFTP diagnosis has only been applied to a small proportion of cancers, but more patients received hemithyroidectomy when eligible.

Importance  Increasing detection of early-stage papillary thyroid neoplasms without improvements in mortality has prompted development of strategies to prevent or mitigate overtreatment.

Objective  To determine adoption rates of 2 recent strategies developed to limit overtreatment of low-risk thyroid cancers: (1) a new classification, noninvasive follicular thyroid neoplasm with papillarylike nuclear features (NIFTP), and (2) hemithyroidectomy for selected papillary thyroid carcinomas (PTCs) up to 4 cm in size.

Design, Setting, and Participants  This is a cross-sectional analysis of 3368 pathology records of 2 cohorts of patients from 18 hospitals in 6 countries during 2 time periods (2015 and 2019). Participating hospitals were included from the US (n = 12), Canada (n = 2), Denmark (n = 1), South Korea (n = 1), South Africa (n = 1), and India (n = 1). The records of the first 100 patients per institution for each year who underwent thyroid-directed surgery (hemithyroidectomy, total thyroidectomy, or completion thyroidectomy) were reviewed.

Main Outcomes and Measures  Frequency of diagnosis of NIFTP, PTCs, and thyroidectomies during the study period.

Results  Of the 790 papillary thyroid neoplasms captured in the 2019 cohort, 38 (4.8%) were diagnosed as NIFTP. Diagnosis of NIFTP was observed in the US, South Africa, and India. There was minimal difference in the total proportion of PTCs in the 2015 cohort compared with the 2019 cohort (778 [47.1%] vs 752 [44.5%]; difference, 2.6% [95% CI, −16.9% to 22.1%]). The proportion of PTCs eligible for hemithyroidectomy but treated with total thyroidectomy in the 2 cohorts demonstrated a decreasing trend from 2015 to 2019 (341 of 453 [75.3%] vs 253 of 434 [58.3%]; difference, 17.0% [95% CI, −1.2% to 35.2%]).

Conclusions and Relevance  Results of this cohort study showed that the 2 mitigation strategies for preventing overtreatment of early-stage thyroid cancer have had mixed success. The diagnosis of NIFTP has only been applied to a small proportion of thyroid neoplasms compared with expected rates. However, more patients eligible for hemithyroidectomy received it in 2019 compared with 2015, showing some success with this deescalation strategy.

The incidence of thyroid cancer has increased steadily over the past 3 decades, mainly attributable to the detection of small, localized papillary thyroid carcinomas (PTCs).^1-8 As the detection of early-stage PTCs has increased without improvements in thyroid cancer–related morbidity or mortality rates, strategies have been developed to try to mitigate the problem, such as participation in the Choosing Wisely campaign, studying active surveillance, and increasing thresholds for thyroid nodule biopsy.^9-13

For this cohort study, we were specifically interested in 2 strategies that have been introduced in recent years. Beginning in 2014, more conservative surgery for low-risk patients with well-differentiated thyroid cancer was supported in several specialty guidelines to decrease overtreatment, with the American Thyroid Association (ATA) 2015 guidelines in particular recommending consideration of lobectomy for qualifying low-risk thyroid neoplasms up to 4 cm in size.^14-18 Previously, the ATA recommended total thyroidectomy for differentiated thyroid cancers 1 cm or larger in size.¹⁹ Then, in 2016, noninvasive encapsulated follicular variant PTC was reclassified as noninvasive follicular thyroid neoplasm with papillarylike nuclear features (NIFTP) to reflect key histopathologic features of PTC but a lack of aggressiveness. These tumors were estimated to affect approximately 45 000 patients worldwide annually, or 18.6% of patients diagnosed with papillary thyroid neoplasms.²⁰

Appropriate deescalation of treatment can have important positive implications for patient and health care system burden related to thyroid cancer management.^17,21-24 At this time, little is known about how the introduction of NIFTP and the support of hemithyroidectomy for selected early-stage cancers has affected real-world practice. We sought to determine how the new NIFTP classification has changed diagnostic rates of PTC, whether there is variation in use of the new diagnostic category across different locations, and whether extent of surgery for PTC has changed with the new ATA guideline recommendations.

We completed a multicenter, retrospective analysis of patients who underwent thyroid-directed surgery through the American Head & Neck Society Endocrine Surgery Section Collaborative, a global network of surgeons who perform neck endocrine surgery. Each center obtained local ethics approval from their respective institutional review boards, and data were shared and combined for analysis through data use agreements. Individual informed consent was waived at all centers owing to use of a limited, deidentified data set.

All surgeons were invited to contribute their data as long as they were able to provide consecutive cases for the designated years of thyroid surgical pathology, reviewed by pathologists. Surgeons contributing to the data set were from 18 centers in 6 countries (US, Canada, Denmark, South Korea, South Africa, and India).

The number of thyroidectomies per year in each primary sampling unit (ie, each institution) was estimated to vary by institution and country within the range of 100 to 500.^24,25 To ensure that an adequate and balanced sample size was achieved for the study questions, each institution submitted the first consecutive 100 or highest number available (if <100) of pathology records of all patients who underwent thyroidectomies for the 2 time periods of the study (2015 and 2019). If available, centers were also invited to submit the first 100 or highest number available of cases for 2016, 2017, and 2018.

Eligible cases were the first 100 consecutive adult patients (18 years and older at the time of surgery) who had undergone thyroid-directed surgery, either thyroid lobectomy, completion thyroidectomy, or total thyroidectomy for any thyroid pathology in calendar years 2015 and 2019. Exclusion criteria were as follows: (1) patients aged 17 years or younger, (2) patients who underwent thyroidectomy as part of a nonthyroid-directed surgery (eg, laryngeal cancer), and (3) surgery was not performed at the participating institution (eg, the pathology was a consultation case). Patients with missing data on final pathologic diagnosis were excluded. Participating centers that were able to provide the identical data for the years 2016 through 2018 contributed to a secondary analysis of trends.

A standardized case report form was developed combining the standards and definitions of the College of American Pathologists synoptic reporting guidelines and the North American Association of Cancer Registries forms.^26-28 The form was piloted, refined, and distributed to all participating institutions for data capture. Data elements extracted included age, sex, year of surgery, fine-needle aspiration (FNA) cytology result, whether molecular testing was performed, type of surgery performed, greatest dimension of the nodule or tumor, and final pathology. Data sources were the individual patient clinical records and operative and pathology reports. Pathologic diagnoses were categorized according to the major categories offered by the College of American Pathologists system. Incidental pathologic diagnoses such as small tumors found in surgeries completed for a large goiter or for Graves disease were noted as secondary diagnoses and were not included in the analysis of the primary outcome. Case definitions for benign and other are presented in eTable 1 in the Supplement.

The primary outcome was the frequency of NIFTP diagnosed in patients who underwent thyroidectomy in 2019. Secondary outcomes for 2015 and 2019 data were: (1) proportion of PTC diagnosed among all thyroidectomies and (2) proportion of total thyroidectomies performed in patients diagnosed with PTC. Factors contributing to extent of primary surgery over the 2 study periods, including age at time of surgery, sex, molecular testing, FNA cytology, and country where surgery was performed, were also quantified. For the selected sites that provided data for 2016 through 2018, (1) the proportion of NIFTP diagnosed among all thyroidectomies, (2) the proportion of PTC diagnosed among all thyroidectomies, and (3) total thyroidectomies performed in patients diagnosed with PTC from the study population were calculated.

For the primary analysis, demographics for the NIFTP and PTC populations were summarized by year. Continuous variables were expressed as medians and quartiles. Crude comparisons among patients were performed using the Mann-Whitney U test. Categorical variables were expressed as frequencies with 95% CIs and were compared using the χ² test. Similar methods were applied to analyze the secondary outcomes. Appropriate effect-size metric and 95% CI around the effect size was used to express the magnitude of observed difference or strength of association.

To quantify the association of selected factors with extent of primary surgery over the 2 study periods, we fitted a logistic regression analysis to evaluate patients who differed with respect to the year surgery was performed but were similar with respect to other measured characteristics. This secondary analysis was performed on patients diagnosed with NIFTP or PTCs that were smaller than 4 cm in size. The logistic regression model included baseline predictor variables that had been selected on the basis of their a priori possibility of confounding the relationship between suspected pathologic diagnosis and extent of surgery performed (the dependent variable).^17,29-32 Baseline variables considered were age at time of surgery, sex, molecular testing, and FNA cytologic results. The country where the surgery was performed was included in the model to quantify and account for differences in surgical practices globally. The outcome variable indicated whether management of the PTC involved a hemithyroidectomy or total thyroidectomy. Presence of multicollinearity in predictor variables was assessed by measuring the variance inflation factor. A variance inflation factor value that exceeded 5 was considered a problematic amount of collinearity.³³ All statistical analyses were performed using RStudio (RStudio Inc).

The total pathology records in the data set at the time of this analysis was 3945. The primary analysis used the records for 2015 and 2019 (n = 3368) from the 18 participating institutions. Six countries contributed, including 12 US institutions (n = 2201), 2 institutions in Canada (n = 398), 1 institution in Denmark (n = 200), 1 institution in South Korea (n = 200), 1 institution in South Africa (n = 142), and 1 institution in India (n = 199).

Demographic and clinical characteristics of the study population analyzed for the primary outcome are summarized in Tables 1 and 2. A total of 3340 pathology records were analyzed for the primary outcome. Excluded records included 14 (0.4%) that reported no final pathology, 9 (0.3%) that included participants who were younger than 18 years, 1 each (0.1%) in the 2015 and 2019 data sets beyond 100 cases; and 3 (0.1%) that included patients who underwent nonthyroid-directed surgery (1 total laryngectomy and 2 parathyroidectomies).

The 2019 cohort included 1688 thyroid surgeries. Just 38 (2.3%) patients were diagnosed with NIFTP, which was 38 of 790 (4.8%) papillary thyroid neoplasms. The majority of patients with NIFTP (29 of 38 [80.6%]) were diagnosed following hemithyroidectomy, while 9 (23.7%) patients underwent total thyroidectomy. A higher proportion of patients diagnosed with NIFTP underwent molecular testing prior to surgery (16 of 38 [43.2%]) compared with patients who were diagnosed with PTC (165 of 752 [22.8%]; difference, 20.4% [95% CI, 2.4%-38.4%]).

There was a minimal difference in the total number of PTCs diagnosed in the 2015 cohort compared with the 2019 cohort (778 [47.1%] vs 752 [44.5%]; difference, 2.6% [95% CI, −16.9% to 22.1%]). Interestingly, 48 of 790 (6.1%) PTCs had been identified as benign on preoperative FNA cytologic results, higher than the published estimates.³⁴ Rates of molecular testing did not change substantially over the study period, decreasing slightly from 26.1% (197 of 778) in 2015 to 22.8% (165 of 752) in 2019 (difference, 3.3% [95% CI, −13.5% to 20.1%]).

The proportion of PTCs treated with total thyroidectomy decreased substantially from 2015 to 2019 (517 of 778 [66.9%] vs 386 of 752 [52.2%]; difference, 14.7% [95% CI, −4.3% to 33.7%]). In particular, the proportion of PTCs treated with total thyroidectomy that would have been eligible for hemithyroidectomy (suspicious for or confirming malignancy on cytologic testing or molecular testing and ≤4 cm) showed a decreasing trend from 2015 to 2019 (341 of 453 [75.3%] vs 253 of 434 [58.3%]; difference, 17.0% [95% CI, −1.2% to 35.2%]).

The majority of total thyroidectomies were performed for PTC in both 2015 and 2019 in institutions in Canada (92 of 103 [89.3%] vs 42 of 55 [76.4%]), India (48 of 79 [60.8%] vs 45 of 81 [55.6%]), South Korea (41 of 44 [93.2%] vs 32 of 34 [94.1%]), and the US (328 of 591 [55.5%] vs 262 of 496 [52.8%]). In contrast, total thyroidectomies were performed commonly for benign disease in 2015 and 2019 in the sample from Denmark (4 of 7 [57.1%] vs 10 of 10 [100%]). The proportion of total thyroidectomies for PTC in South Africa were comparable in 2015 (6 of 16 [37.5%]) and 2019 (5 of 12 [41.7%]).

A total of 1217 patients with complete data were included in the logistic regression model for extent of primary surgery (ie, total thyroidectomy vs hemithyroidectomy) for PTC (Table 3). Compared with 2015, the odds ratio (OR) of receiving a total thyroidectomy when eligible for a hemithyroidectomy in 2019 was 0.40 (95% CI, 0.30-0.52). Thyroid cancers with FNA cytology classified as nondiagnostic (Bethesda I; OR, 0.15 [95% CI, 0.06-0.33]), benign (Bethesda II; OR, 0.17 [95% CI, 0.10-0.29]), follicular lesion of undetermined significance (Bethesda III; OR, 0.14 [95% CI, 0.09-0.21]), or follicular neoplasm (Bethesda IV; OR, 0.19 [95% CI, 0.11-0.32]) were associated with decreased odds of total thyroidectomy compared with those classified as malignant (Bethesda VI). The OR for thyroid cancers with FNA cytology classified as suspicious for malignancy (Bethesda V) was 0.78 (95% CI, 0.54-1.14) compared with those classified as malignant. Compared with institutions in the US, patients treated for PTC in institutions in Canada, Denmark, and South Korea had an OR of receiving a total thyroidectomy of 0.49 (95% CI, 0.34-0.73), 0.02 (95% CI, 0.003-0.08), and 0.21 (95% CI, 0.14-0.31), respectively. The OR of total thyroidectomy observed at the institution in South Africa compared with the US was 0.45 (95% CI, 0.18-1.12). By contrast, the OR of management of PTC with a total thyroidectomy compared with a hemithyroidectomy was higher in the 1 participating institution in India compared with the US institutions (12.68 [95% CI, 5.31-37.71]). There was no association of age, molecular testing, and sex with extent of surgery. Multicollinearity was not observed in the regression analysis.

The trends in NIFTP diagnosis and total thyroidectomies from 2015 to 2019 were compared (eTables 2 and 3 in the Supplement). One institution in the US and 1 institution in Canada contributed 921 pathology records to this analysis. The diagnosis of NIFTP among papillary thyroid neoplasms did not change substantially by year since its introduction (2016: 7 of 94 [7.4%]; 2017: 9 of 95 [9.5%]; 2018: 8 of 113 [7.1%]; 2019: 6 of 125 [4.8%]). Total thyroidectomies for the primary treatment of PTC demonstrated a steady decline from 65.1% (69 of 106) in 2015 to 41.1% (39 of 95) in 2019 (difference, 24.0% [95% CI, 10.2%-36.6%]) (Figure).

In this multi-institutional analysis of a cohort of PTC specimens, in 2019 the diagnosis of NIFTP was observed in only 4.7% of papillary thyroid neoplasms. Among patients diagnosed with papillary thyroid neoplasms in 2019, the introduction of NIFTP accounted for 7.1% of cases in the US, 11.1% in South Africa, and 4.0% in India. In contrast, introduction of the NIFTP nomenclature was anticipated to affect an estimated 18.6% of papillary thyroid neoplasms globally.²⁰ Analysis of trends in NIFTP diagnosis from 2016 to 2019 suggest that the application of the NIFTP diagnosis did not considerably change over time following its introduction. We observed a trend toward more conservative surgery when appropriate for eligible PTCs; 41.7% of eligible PTCs were treated with hemithyroidectomy in 2019.

These data are consistent with a recent evaluation from the US Surveillance, Epidemiology, and End Results (SEER) cancer registry that showed a 1.3% diagnostic rate of NIFTP in 2017, much lower than estimated by Nikiforov et al.^20,35 The observed low frequency of diagnosis of NIFTP compared with expectations could be owing to a lower prevalence of NIFTP than originally anticipated, slow incorporation of the NIFTP nomenclature by pathologists globally, or some other unexplained reason requiring further investigation.^1,36-40 The low interobserver reliability in thyroid histopathology could be one factor.⁴¹ Pathologists have also outlined their reservations surrounding the diagnostic criteria, namely, the challenges associated with assessing the entire tumor capsule for large tumors, the histological assessment of the entire tumor to exclude focal necrosis or vascular invasion, and the lack of molecular testing specific for this entity.⁴²

Whether molecular testing has potential as an adjunct tool for NIFTP diagnosis in the future remains to be seen. Among established molecular tests for PTC, it is known that mutations to BRAF V600E,^43,44 p53, or to the TERT promotor²⁰ are typically absent in NIFTP, underscoring their indolent biology. Immunohistochemical markers including programmed cell death 1 ligand 1⁴⁵ have also been studied. In the present study population, only 40% of patients diagnosed with NIFTP had undergone molecular testing. The role of molecular testing in the diagnosis of NIFTP requires further external validation and cost-effectiveness analyses prior to becoming standard of care for these lesions.^46,47

The findings showing a trend toward hemithyroidectomy for appropriately selected cancers are in alignment with both a single-institution study and an analysis of surgical trends in the US SEER cancer registry that extended just through 2016.^48,49 The SEER analysis demonstrated a decrease in the use of total thyroidectomy and also a decrease in the use of radioactive iodine treatment for PTCs measuring 2 cm or less.⁴⁹ Nevertheless, the present study findings suggest more than 50% of PTCs eligible for deescalated management with hemithyroidectomy continued to be treated with total thyroidectomy in 2019. The continued high volume of total thyroidectomies performed for early-stage PTC deserves future investigation in later studies. It could reflect additional factors that are contemplated in surgical decision-making, such as surgeon and patient preferences, and the anticipated need for adjuvant therapy.

The strengths of this study are its contemporary information on trends, inclusion of international sites, straightforward design, and use of standardized, broadly used registry definitions.^26-28 The study did not use centralized pathology reinterpretation, and readers may wonder about the effect of inter-rater and regional variability, but this was intentional in the design. The goal was to observe the use of the NIFTP diagnosis in a real-world setting to examine how NIFTP application varied across settings. We observed marked uniformity in the low rates of application of the NIFTP diagnosis. Inferences about reasons for total or hemithyroidectomy for hemithyroidectomy-eligible cases are not possible beyond what is analyzed here; we did not collect detailed data on surgical decision-making. Despite this, there is no reason to believe that there is systematic error in the data that is likely to result in confounding of the results. Furthermore, fewer than 10% of cancer case data extracted by institutions was missing patient information. Of note, the present sample included patients who underwent surgical resection and, therefore, is not representative of all PTC diagnoses. Some patients diagnosed with malignant thyroid neoplasms on FNA choose active surveillance, or decline, are not offered, or are unable to undergo surgery.

This cohort study demonstrated that the 2 approaches we studied to assess progress in efforts to prevent or mitigate overtreatment of early-stage PTC show mixed success so far. Three years after the introduction of the NIFTP diagnostic criteria, reclassification of selected PTCs to NIFTP has only been applied to a small patient population. Further research is needed to determine whether fewer tumors qualify for this classification than initially expected, whether limited use of the NIFTP nomenclature represents an opportunity for education about or refinement of the NIFTP diagnostic category, or other yet unknown reasons. A better understanding of these causes would be valuable as we continue to improve risk stratification in thyroid oncology and for those considering nomenclature and classification changes in other fields.

Fewer patients eligible for hemithyroidectomy underwent total thyroidectomy, in keeping with current recommendations and showing some adoption of this guideline. Further studies may help elucidate the fundamental factors that delay or promote progress in adoption of this guideline.

This is the first study of the American Head & Neck Society Endocrine Surgery Section Collaborative, a global network of endocrine surgeons. The success of this study demonstrates the feasibility of an international collaboration with data transfer and opens the way for additional studies in the future.

Accepted for Publication: September 4, 2021.

Published Online: November 24, 2021. doi:10.1001/jamaoto.2021.3277

Corresponding Author: Lisa Caulley, MD, MPH, Department of Otolaryngology–Head and Neck Surgery, The Ottawa Hospital, 501 Smyth Rd, Ottawa, ON K1H 8L6, Canada (lic955@mail.harvard.edu).

Author Contributions: Drs Caulley and Davies had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Caulley, Eskander, Stack, Randolph, Davies.

Acquisition, analysis, or interpretation of data: Caulley, Eskander, Yang, Auh, Cairncross, Cho, Golbon, Iyer, Liu, Lee, Lindeman, Meltzer, Molin, Moore, Noel, Nozolino, Pasternak, Price, Ramsay, Rolighed, Sajisevi, Sharma, Sinclair, Sorensen, Tae, Tang, Tsao, Williams, Wrenn, Xing, Zafereo.

Drafting of the manuscript: Caulley, Eskander, Davies.

Critical revision of the manuscript for important intellectual content: Caulley, Eskander, Cairncross, Cho, Golbon, Iyer, Liu, Lee, Lindeman, Moore, Noel, Nozolino, Pasternak, Price, Ramsay, Rolighed, Sajisevi, Sharma, Sinclair, Sorensen, Tae, Tang, Tsao, Williams, Wrenn, Xing, Zafereo, Stack, Davies.

Statistical analysis: Caulley, Yang, Auh, Moore, Nozolino, Ramsay.

Administrative, technical, or material support: Caulley, Eskander, Auh, Cho, Golbon, Iyer, Liu, Lee, Meltzer, Molin, Moore, Noel, Nozolino, Sharma, Sinclair, Tae, Wrenn, Xing, Randolph, Davies.

Supervision: Caulley, Cho, Iyer, Noel, Sinclair, Zafereo, Stack, Randolph, Davies.

Conflict of Interest Disclosures: Dr Caulley reported grants from the Canadian Institutes of Health Research and the PSI Foundation outside the submitted work. Dr Eskander reported grants from Merck and personal fees from Bristol Myers Squibb outside the submitted work. Dr Williams reported serving as a member of the Bayer scientific advisory board. Dr Zafereo reported grants from Eli Lilly, Merck, and GenePro Diagnostics outside the submitted work. No other disclosures were reported.

Disclaimer: Dr Davies is an Associate Editor of JAMA Otolaryngology–Head & Neck Surgery, but she was not involved in any of the decisions regarding review of the manuscript or its acceptance.

Additional Information: This multinational, multi-institutional analysis was a project of the American Head & Neck Society Endocrine Surgery Section Collaborative.