eTable 1. Genetic Alterations Identified in Test-Positive Nodules
eTable 2. Follow-up Information of DTCs, Including Patients With Other Significant Foci of Thyroid Cancer
eTable 3. Pairwise Comparisons for Size Distribution by Histologic Diagnosis
Customize your JAMA Network experience by selecting one or more topics from the list below.
Valderrabano P, Khazai L, Thompson ZJ, et al. Association of Tumor Size With Histologic and Clinical Outcomes Among Patients With Cytologically Indeterminate Thyroid Nodules. JAMA Otolaryngol Head Neck Surg. 2018;144(9):788–795. doi:10.1001/jamaoto.2018.1070
Should tumor size determine the extent of surgery for cytologically indeterminate thyroid nodules?
In this cohort study of 652 indeterminate thyroid nodules, more than 90% of indeterminate thyroid nodules were benign or low-risk malignant tumors for which a lobectomy would be sufficient initial treatment regardless of tumor size. Tumor size was not associated with cancer rates, cancer aggressiveness, or response to therapy.
In the absence of other indications for total thyroidectomy, this study suggests that the preferred surgical approach for cytologically indeterminate thyroid nodules is a thyroid lobectomy regardless of tumor size.
Tens of thousands of unnecessary operations are performed each year for diagnostic purposes among patients with cytologically indeterminate thyroid nodules. Whereas a diagnostic lobectomy is recommended for most patients with solitary indeterminate thyroid nodules, a total thyroidectomy is preferred for nodules larger than 4 cm.
To determine whether histologic or clinical outcomes of indeterminate thyroid nodules 4 cm or larger are worse than those for nodules smaller than 4 cm, thus justifying a more aggressive initial surgical approach.
Design, Setting, and Participants
In this retrospective cohort study, 652 indeterminate thyroid nodules (546 nodules <4 cm and 106 nodules ≥4 cm) with surgical follow-up were consecutively evaluated at an academic cancer center from October 1, 2008, through April 30, 2016.
Main Outcomes and Measures
Differences in cancer rates, rates of invasive features, cancer aggressiveness, and response to therapy between indeterminate thyroid nodules smaller than 4 cm and 4 cm or larger.
A total of 652 indeterminate thyroid nodules (546 nodules <4 cm and 106 nodules ≥4 cm) from 589 patients (mean [SD] age, 53.1 [13.8] years; 453 [76.9%] female) were studied. No differences were found in the baseline characteristics of patients or nodules between the 2 size groups. Tumor size was not associated with the cancer rate as a categorical (140 of 546 [25.6%] for nodules <4 cm and 33 of 106 [31.1%] for nodules ≥4 cm; effect size, 0.05; 95% CI, 0.002-0.12) or continuous (odds ratio [OR], 1.03; 95% CI, 0.92-1.15) variable. No association was found between nodule size and prevalence of extrathyroidal extension, positive margins, lymphovascular invasion, lymph node metastasis, or distant metastasis. Most malignant tumors were low risk in both size groups (70% in the nodules <4 cm and 72% in the nodules ≥4 cm), and tumor size was not associated with tumor aggressiveness as a categorical (effect size, 0.10; 95% CI, 0.03-0.31) or continuous variable (OR for intermediate-risk cancer, 0.91; 95% CI, 0.72-1.14; OR for high-risk cancer, 1.43; 95% CI, 0.96-2.15). At the last follow-up visit, 88 of 105 patients (83.8%) with malignant tumors in the smaller than 4 cm group and 21 of 25 (84.0%) in the 4 cm or greater group had no evidence of disease, and tumor size was not associated with response to therapy (effect size, 0.13; 95% CI, 0.07-0.33).
Conclusions and Relevance
Most indeterminate thyroid nodules are benign or low-risk malignant tumors regardless of tumor size. In the absence of other indications for total thyroidectomy, this study suggests that a thyroid lobectomy is sufficient initial treatment for most solitary cytologically indeterminate thyroid nodules independent of the tumor size.
Each year more than 60 000 operations for histologically benign thyroid nodules are unnecessarily performed for diagnostic purposes among patients with cytologically indeterminate thyroid nodules (ITNs) (eg, atypia or follicular lesion of undetermined significance or follicular or Hürthle cell neoplasm of the Bethesda System for Reporting Thyroid Cytopathology [hereafter referred to as the Bethesda system]).1,2 When surgery is indicated, a diagnostic lobectomy is generally the recommended approach for patients with solitary ITNs.3 The 2015 American Thyroid Association (ATA) guidelines for the management of patients with thyroid nodules and differentiated thyroid cancer (DTC) suggest, however, that a total thyroidectomy may be preferred in certain situations in which risk of cancer is increased, particularly if, in the event of being malignant, a completion thyroidectomy would be indicated (recommendation 20.a).3 Large tumor size (ie, >4 cm) has been traditionally associated with an increased risk of cancer even after a benign or indeterminate cytologic result,4,5 and a total thyroidectomy is currently the preferred treatment option for DTCs larger than 4 cm. For this reason, the ATA guidelines suggest that a total thyroidectomy may be the preferred surgical approach for ITNs larger than 4 cm.3
Several recent developments prompted us to reevaluate this recommendation. First, the recognition that, because of the indolent behavior of most thyroid cancers, less initial treatment is often better because more extensive resections in low-risk cancers are associated with an increased complication rate without improving survival.6,7 This concept has been the driving force for many of the changes presented in the latest version of the ATA guidelines.3 Second, increasing evidence suggests that most malignant tumors diagnosed after indeterminate cytologic results are low risk. These cancers have significantly better prognosis than those found in lesions with cytologic results diagnostic of or suspicious for cancer.8-11 Third, a large paradigm shift occurred in the diagnostic criteria for thyroid cancer, with almost 40% of ITNs formerly called cancer now being reclassified as premalignant, for which a diagnostic lobectomy is considered to be sufficient treatment.12-14 This study aimed to determine whether histologic or follow-up outcomes of ITNs 4 cm or larger are worse than for nodules smaller than 4 cm, thus justifying a more aggressive initial surgical approach.
Electronic medical records of patients with ITNs evaluated from October 1, 2008, through April 30, 2016, at the H. Lee Moffitt Cancer Center and Research Institute, Tampa, Florida, were retrospectively reviewed. Of 1053 ITNs, 24 (2.3%) had presurgical evidence of cancer and were excluded from the study: 5 had vocal cord paralysis, 14 had proven cervical lymph node metastasis, 1 had vocal cord paralysis and proven cervical lymph node metastasis, 1 had proven distant metastasis, 2 had elevated basal serum calcitonin level (>100 pg/mL [to convert to picomoles per liter, multiply by 0.292]), and 1 had explosive growth. In addition, 371 ITNs (35.2%) without histologic correlation (no surgical follow-up or unknown histologic type of the biopsied nodule) and 6 (0.6%) without presurgical size available were excluded. Finally, 652 ITNs (61.9%) in 589 patients were included in the study. The University of South Florida Institutional Review Board provided a waiver of consent for this study.
All cytologic diagnoses were issued by board-certified cytopathologists at our institution according to the Bethesda system and were abstracted from the reports. For nodules with multiple biopsy specimens, the most recent was used for the analyses. Histologic diagnoses were retrieved from reports issued before February 2017 at the pathology department at H. Lee Moffitt Cancer Center and Research Institute (634 ITNs [97.2%]) or elsewhere (18 ITNs [2.8%]). Tumor staging was based on the seventh edition of the American Joint Committee on Cancer classification for DTCs.15 The histopathologic diagnoses and features of the biopsied nodules, matched by size and location, were used for the analyses. Histologic findings were reviewed for 91 of 95 follicular variant of papillary thyroid carcinomas (FVPTCs) (95.8%) diagnosed before April 2016; these were reclassified as noninvasive follicular thyroid neoplasms with papillary-like nuclear features (NIFTPs) or as papillary thyroid carcinomas (PTCs) according to the current standards.12 The DTCs were classified into 3 groups according to clinicopathologic features. Low-risk DTCs had all the following features: no extrathyroidal extension, negative resection margins, no lymphovascular invasion (<4 foci for minimally invasive follicular thyroid carcinomas), clinical N0 or 5 or fewer lymph node metastases all 2 mm or smaller, and no distant metastasis. High-risk DTCs had any of the following: gross extrathyroidal extension (T4), N1b or any lymph node metastasis larger than 1 cm, and/or distant metastasis. Intermediate-risk DTCs were all others. Response to therapy was assessed as recommended in the ATA guidelines for malignant tumors treated with radioactive iodine3 and as suggested in a recent publication for the remaining tumors.16
Oncogene panels were used selectively since the middle of 2012 and consistently since 2014 for the evaluation of ITNs at our institution. A 7-gene oncogene panel (miRInform, Asuragen, in all but 1 patient) was used until September 2014, and results were considered to be positive if any of the 14 point mutations or 3 gene rearrangements were identified by the test.17 ThyroSeq version 2 (University of Pittsburgh Medical Center and CBLPath Inc) was used thereafter, and results were considered to be positive if a genetic alteration (>1000 hotspots in 14 genes and 42 gene rearrangements) was identified at an allelic frequency greater than 5% or 10%, depending on the mutation as described in the original validation studies.18-20 Although resection rates were lower for test-negative nodules than for test-positive nodules, tests results were used primarily to determine the extent of surgery rather than to avoid resection, in accordance with the guidelines’ recommendations at the time.21,22
Comparisons were made between tumors that measured less than 4 cm and 4 cm or more on preoperative ultrasound images using χ2 tests, and Fisher exact tests were used for categorical variables. Effect sizes are reported with 95% bootstrap CIs for the Cramer V (ranging from 0, indicating no relationship, to 1, indicating perfect relationship). The association between 2 variables is not generally acceptable for values below 0.15, strong for values above 0.3, and redundant for values greater than 0.5. Differences in recurrence rate and response to therapy at the last visit were assessed for patients with DTCs who had at least 6 months of follow-up and did not have any other thyroid cancer foci except for PTCs 1 cm or smaller. Several analyses were also performed considering size as a continuous variable. Logistic regression tested the association of size with cancer, multinomial logistic regression evaluated the association of size with cancer aggressiveness (low risk, intermediate risk, and high risk, as defined above), and Cohen d effect sizes with 95% CIs for multiple comparisons evaluated differences in size by histologic analysis. The Cohen d effect sizes were considered as small (≤0.2), medium (0.2-0.8), or large (≥0.8).23 A Cohen d effect size of 0.0 represents complete overlap of the means of 2 groups. All analyses were performed with R, version 3.4.3 (R Foundation for Statistical Computing) and SAS, version 9.4 (SAS Institute Inc).
A total of 652 ITNs among 589 patients (mean [SD] age, 53.1 [13.8] years; 453 [76.9%] female) were included in the study. Of these, 546 ITNs (83.7%) in 485 patients were smaller than 4 cm (mean, 2.6 cm), and 106 ITNs (16.2%) in 104 patients were 4 cm or larger (mean, 5.4 cm). Baseline characteristics, including age, sex, thyroid function, multinodularity, presence of clinically significant (>1 cm) contralateral nodules, medical history of other malignant tumors, radiation exposure to the neck, and family history of thyroid cancer or genetic syndromes associated with thyroid cancer, had weak associations with tumor size (Table 1). Cytologic types were classified as atypia or follicular lesion of undetermined significance in 272 nodules (41.7%) and as follicular or Hürthle cell neoplasm in 380 nodules (58.3%) without differences between the groups. A total of 190 nodules were evaluated with oncogene panels before surgery: 62 (32.6%) with a 7-gene oncogene panel and 128 (67.3%) with ThyroSeq version 2. The proportion of nodules evaluated with oncogene panels was similar between nodules larger than 4 cm (163 of 546 [29.9%]) and 4 cm or smaller (27 of 106 [25.5%]).
Oncogene panels had a positive result in 51 of 190 nodules (26.8%), with a higher proportion of mutations identified among the smaller than 4 cm group (48 of 163 [29.4%] vs 3 of 26 [11.5%]) but a weak effect size (0.14; 95% CI, 0.03-0.24). In both groups, two-thirds of the identified mutations were in the RAS (eg, NRAS [OMIM 164790], HRAS [OMIM 190020], and KRAS [OMIM 190070]) genes (effect size, 0.10; 95% CI, 0.01-0.19). eTable 1 in the Supplement provides a complete list of genetic alterations in the test-positive nodules of each size group.
A total of 173 of 652 ITNs (26.5%) were determined to be malignant by histologic examination (114 [17.5%] if NIFTPs are excluded). No differences were found in the prevalence of cancer between nodules smaller than 4 cm and 4 cm or larger with NIFTPs (140 of 546 [25.6%] for nodules <4 cm and 33 of 106 [31.1%] for nodules ≥4 cm; effect size, 0.05; 95% CI, 0.002-0.12) or without NIFTPs (93 of 546 [17.0%] for nodules <4 cm and 21 of 106 [19.8%] for nodules ≥4 cm; effect size, 0.03; 95% CI, 0.002-0.11) (Table 2). Similar results were observed between nodules evaluated with and without oncogene panels. A weak association was found between size and the distribution of the histologic diagnoses (effect size, 0.15; 95% CI, 0.11-0.26), with a higher proportion of nonneoplastic benign nodules (153 of 546 [28.0%] vs 20 of 106 [18.9%]) and nonfollicular pattern PTCs (40 of 546 [7.3%] vs 2 of 106 [1.9%]) and a lower proportion of follicular thyroid carcinomas (18 of 546 [3.3%] vs 9 of 106 [8.5%]) in ITNs smaller than 4 cm compared with ITNs 4 cm or larger.
Locoregional or distant involvement was infrequent among malignant tumors (Table 3). No differences (weak associations) were found in the proportion of nodules with extrathyroidal extension, positive surgical margins, lymph node metastasis, or distant metastasis between the 2 groups. A weak association was found between the presence of distant metastasis and larger malignant tumors (effect size, 0.16; 95% CI, 0.01-0.30). A lobectomy may have been appropriate for more than 90% of all ITNs (with benign histologic findings or low-risk malignant tumors, which included NIFTPs) if size was not taken into account to decide the extent of surgery. High-risk features were found in 5 cases (0.8%), including the 2 medullary thyroid carcinomas.
There were 164 patients with 167 foci of DTC in this cohort. Seventeen malignant tumors with another thyroid cancer focus (which could act as a confounding factor) and 17 malignant tumors with less than 6 months of follow-up were excluded. Follow-up outcomes were finally retrieved from 105 patients with DTCs smaller than 4 cm and 25 patients with DTCs 4 cm or larger (Table 4). The median follow-up time was 36 months, with no differences between groups (37 months for patients with DTCs <4 cm and 33 months for patients with DTCs ≥4 cm; effect size, 4.0; 95% CI, −5.0 to 17.5). No recurrences occurred in any of the groups during follow-up, and no differences were found in the response to therapy between the groups (effect size, 0.13; 95% CI, 0.07-0.33). Most patients had excellent response (no evidence of disease) at the last follow-up visit (88 of 105 patients [83.8%] in the <4-cm group and 21 of 25 [84.0%] in the ≥4-cm group). Similar outcomes were observed if patients with other concurrent thyroid cancer foci were included (eTable 2 in the Supplement).
Size as a continuous variable was not associated with the risk of cancer (OR, 1.03; 95% CI, 0.92-1.15) or with the aggressiveness of cancers. Compared with the odds of the nodule being low-risk cancer, for each unit increase in size, there was a non–statistically significant 43% increase in the odds of the nodule being high-risk cancer (OR, 1.43; 95% CI, 0.96-2.15) and a non–statistically significant 9% decrease in the odds of the nodule being intermediate-risk cancer (OR, 0.91; 95% CI, 0.72-1.14). The size metrics of each histologic diagnosis are summarized in Table 2. Among all possible pairwise comparisons in size (eTable 3 in the Supplement), the following Cohen d effect sizes were large: non-FVPTC PTCs (other PTCs) were smaller than adenomas (effect size, 1.01; 95% CI, 0.68-1.34), NIFTPs (effect size, 0.89; 95% CI, 0.48-1.30), FVPTCs (effect size, 1.04; 95% CI, 0.58-1.50), follicular thyroid carcinomas (effect size, 1.67; 95% CI, 0.68-1.34), and nonfollicular cell-derived cancers (other) (effect size, 1.63; 95% CI, 0.71-2.55), and hyperplastic and adenomatoid nodules were smaller than follicular thyroid carcinomas (effect size, 1.03; 95% CI, 0.61-1.45) and nonfollicular cell-derived cancers (other) (effect size, 1.07; 95% CI, 0.25-1.89). The effect sizes of all other possible pairwise comparisons were small or medium.
In this series, ITN size of 4 cm or larger was not associated with an increased risk of cancer, cancer aggressiveness, or worse outcomes. In fact, most ITNs, regardless of nodule size, were benign or low-risk malignant tumors for which a lobectomy could be a sufficient initial treatment.
The rate of cancer for indeterminate categories at our institution, considering NIFTPs as benign or malignant, are within the expected and previously reported values,25 and the histologic diagnoses and mutational profiles in this series are consistent with those of previously published cohorts of cytologically ITNs.10,18,19,26 We believe that despite the inherent design limitations, this single-center retrospective cohort is representative, and thus results should be widely applicable. The known interinstitutional variability in cancer rates for the indeterminate categories of thyroid cytopathology should not affect the interpretation of the results for several reasons.27
First, most malignant ITNs are low-risk malignant tumors. Histologic and molecular findings of malignant ITNs have been suggested to be different from other malignant tumors with more definitive cytologic results (ie, suspicious for cancer or diagnostic of cancer).8-11 In the former, most mutations are RAS like, which are generally associated with better differentiated follicular pattern tumors, whereas in the latter, BRAF-like mutations are the most prevalent and frequently associated with a poorer thyroid differentiation score and PTC.28-32 In previous studies from our institution,17,20,22 the genetic background of ITNs was consistent with this expectation, with approximately 60% of identified mutations being in the RAS genes. The present study found that most malignant tumors among the indeterminate cytologic categories were low-risk follicular-pattern lesions (including NIFTPs and minimally invasive follicular thyroid carcinomas), which is also consistent with previous series.8-11 This study adds to the current knowledge that, among larger tumors, low-risk cancers are the most prevalent malignant tumors.
Second, the NIFTP reclassification affected the institutional cancer rates. Soon after the introduction of the Bethesda system, it became clear that there were wide interinstitutional variations in the rate of cancer of the indeterminate categories.27 The interpretation of noninvasive follicular pattern lesions as follicular adenomas or follicular variant of papillary thyroid carcinomas (now NIFTPs) seem to be a significant source of disagreement.12,33-35 For this reason, the interinstitutional variability in the rates of cancer of the indeterminate cytologic categories should be significantly decreased if NIFTPs are no longer considered to be malignant. This speculation, however, remains to be proven. The present study did not find differences in the rate of cancer calculated with or without NIFTPs between the 2 groups. Unlike this study, a recent meta-analysis4 of 646 ITNs from 5 studies found an increased risk of cancer for ITNs larger than 4 cm compared with smaller ITNs. The studies in that meta-analysis were published before the reclassification of encapsulated noninvasive FVPTCs as NIFTPs, however, and the heterogeneity in the rates of cancer was significant. A new evaluation that excludes NIFTPs from the calculation of the rate of cancer might be necessary to clarify how size affects cancer risk among ITNs.
Third, total thyroidectomy is no longer recommended for all DTCs larger than 1 cm. The 2015 ATA guidelines consider that a lobectomy may be sufficient initial treatment for intrathyroidal DTCs smaller than 4 cm, NIFTPs of any size, and minimally invasive follicular thyroid carcinomas and intrathyroidal encapsulated FVPTCs.3,36 This treatment de-escalation for low-risk malignant tumors is supported by 2 facts: surgical extent does not seem to affect survival or recurrence rates37,38 and the complication rate of thyroid lobectomies is half that of total thyroidectomies (11% vs 20%).7 Furthermore, a more conservative resection allows preserving thyroid function in approximately 80% of cases,39 and, if needed, the outcomes and complication rate of a 2-step thyroidectomy are similar to those of a total thyroidectomy.40
In this context, in which the presence of aggressive clinicopathologic features rather than the expected rate of cancer drives the surgical plan to minimize overtreatment, it seems contradictory that a total thyroidectomy be recommended on the sole basis of size larger than 4 cm given that 70% of ITNs are expected to be histologically benign. In this study, we found no association between tumor size and cancer aggressiveness, using size as a continuous variable or as a binary variable with the currently accepted size threshold (4 cm). On the basis of histologic findings, a lobectomy could have been sufficient initial treatment for more than 90% of the ITNs in this cohort regardless of size.
We recognize that a type II error is possible because of the limited number of malignant tumors 4 cm or larger and the limited number of high-risk cancers in this cohort. The only patient with distant metastasis had a 6.8-cm FVPTC with capsular and vascular invasion and mutations in the NRAS, TERT (OMIM 187270), and EIF1AX (OMIM 300186) genes detected in a presurgical fine-needle aspiration biopsy specimen. Follow-up time might be short to detect recurrences in DTCs. The risk of recurrence in this series, however, is expected to be low because 84% of the malignant tumors were in patients with no evidence of disease at last follow-up visit.3
The cytologic features of all ITNs included in this series were consecutively evaluated, and 97% of the histologic diagnoses were issued by board-certified pathologists at an academic cancer center, which strengthens the reliability of the pathologic diagnoses. Furthermore, the histologic specimens of 96% of the FVPTCs diagnosed before the proposed change in nomenclature for encapsulated and noninvasive tumors into NIFTPs were re-reviewed to ensure adherence with current diagnostic criteria. Nonetheless, we recognize the limitations of light microscopy.24
Malignant tumors of any size with indeterminate cytologic findings are usually well-differentiated, low-risk follicular-pattern tumors with excellent outcomes. Thus, recommending a total thyroidectomy for ITNs on the basis of tumor size may lead to overtreatment in a significant proportion of patients, unnecessarily exposing them to surgical risks and complications and to the need for lifelong thyroid hormone–replacement therapy. Because the expected risk from cancer (rather than the expected risk of cancer) drives the current surgical recommendations, this study suggests that in the absence of other indications for total thyroidectomy, a therapeutic or diagnostic lobectomy would be appropriate for more than 90% of solitary ITNs regardless of tumor size.
Accepted for Publication: May 24, 2018.
Corresponding Author: Pablo Valderrabano, MD, PhD, Department of Head and Neck–Endocrine Oncology, H. Lee Moffitt Cancer Center and Research Institute, 12902 Magnolia Dr, Tampa, FL 33612 (email@example.com).
Published Online: July 19, 2018. doi:10.1001/jamaoto.2018.1070
Author Contributions: Dr Valderrabano had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Valderrabano, Otto, McIver.
Acquisition, analysis, or interpretation of data: Valderrabano, Khazai, Thompson, Hallanger-Johnson, Chung, Centeno.
Drafting of the manuscript: Valderrabano, Otto.
Critical revision of the manuscript for important intellectual content: Valderrabano, Khazai, Thompson, Hallanger-Johnson, Chung, Centeno, McIver.
Statistical analysis: Thompson.
Obtained funding: Chung.
Administrative, technical, or material support: Otto, Hallanger-Johnson, Chung, Centeno.
Supervision: Chung, McIver.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.
Funding/Support: This work has been supported in part by grant P30-CA076292 from the Biostatistics Core Facility at the H. Lee Moffitt Cancer Center and Research Institute, a National Cancer Institute–designated comprehensive cancer center.
Role of the Funder/Sponsor: The funding source had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.