Access the full text.
Sign up today, get DeepDyve free for 14 days.
Loading next page...
/lp/ou_press/gene-expression-classifier-vs-targeted-next-generation-sequencing-in-cnGLZbPSwS
- Publisher
- Oxford University Press
- Copyright
- Copyright © 2018 Endocrine Society
- ISSN
- 0021-972X
- eISSN
- 1945-7197
- DOI
- 10.1210/jc.2017-02754
- Publisher site
- See Article on Publisher Site
Abstract
Abstract Context Molecular testing has reduced the need for diagnostic hemithyroidectomy for indeterminate thyroid nodules. No studies have directly compared molecular testing techniques. Objective Compare the diagnostic performance of Afirma Gene Expression Classifier (GEC) with that of ThyroSeq v2 next-generation sequencing assay. Design Parallel randomized trial, monthly block randomization of patients with Bethesda III/IV cytology to GEC or ThyroSeq v2. Setting University of California, Los Angeles. Participants Patients who underwent thyroid biopsy (April 2016 to June 2017). Intervention Testing with GEC or ThyroSeq v2. Main Outcome Measure Molecular test performance. Results Of 1372 thyroid nodules, 176 (13%) had indeterminate cytology and 149 of 157 eligible indeterminate nodules (95%) were included in the study. Of nodules tested with GEC, 49% were suspicious, 43% were benign, and 9% were insufficient. Of nodules tested with ThyroSeq v2, 19% were mutation positive, 77% were mutation negative, and 4% were insufficient. The specificities of GEC and ThyroSeq v2 were 66% and 91%, respectively (P = 0.002); the positive predictive values of GEC and ThyroSeq v2 were 39% and 57%, respectively. Diagnostic hemithyroidectomy was avoided in 28 patients tested with GEC (39%) and 49 patients tested with ThyroSeq v2 (62%). Surveillance ultrasonography was available for 46 nodules (45 remained stable). Conclusions ThyroSeq v2 had higher specificity than Afirma GEC and allowed more patients to avoid surgery. Long-term surveillance is necessary to assess the false-negative rate of these particular molecular tests. Further studies are required for comparison with other available molecular diagnostics and for newer tests as they are developed. The widespread use of medical imaging has spurred substantial growth in the detection of thyroid nodules, to the point that some authors have called attention to an epidemic of thyroid nodule overdiagnosis and overtreatment (1–4). More than 500,000 fine-needle aspiration (FNA) biopsies are performed annually in the United States (5). However, because the risk of malignancy in thyroid nodules is only 5% to 15% (6), a major management challenge is the avoidance of unnecessary surgery. Approximately 20% of FNAs yield an indeterminate (Bethesda class III or IV) result, which confers an ∼25% risk of malignancy (7, 8). Although the conventional treatment algorithm for indeterminate nodules includes repeated biopsy or diagnostic hemithyroidectomy, the majority of resected nodules are ultimately proven benign on histopathology. Recent advances in molecular testing have enabled more patients with indeterminate cytology results to avoid unnecessary surgery (9, 10). The Afirma Gene Expression Classifier (GEC) (Veracyte, Inc., San Francisco, CA) uses microarray technology to analyze messenger RNA expression of 167 genes and was designed as a “rule out” test with a reported sensitivity of 92% and specificity of 52% (11–13). The competing technology ThyroSeq v2 (CBLPath, Inc., Rye Brook, NY) relies on next-generation sequencing to detect DNA and RNA alterations in 14 genes and 42 gene fusions. With a reported sensitivity of 90% and specificity as high as 93% in limited validation studies, ThyroSeq v2 has the potential to perform as both a “rule in” test and a rule out test (14–16). No prior studies have compared the sensitivity and specificity of different molecular testing techniques, and long-term follow-up is lacking for patients with benign molecular test results who do not undergo surgery. We performed a prospective randomized comparison of GEC and ThyroSeq v2 using a pragmatic study design incorporating real-world decision making within a large, geographically dispersed health system. The objective of this study was to compare test performance and the number of diagnostic hemithyroidectomies avoided as a result of molecular testing. Materials and Methods Setting and participants All patients who underwent thyroid FNA throughout the University of California, Los Angeles (UCLA) Health System were eligible for enrollment (May 2016 to June 2017). Thyroid FNAs were performed at five sites by 14 practitioners. Clinical decisions were left to the discretion of the treating physicians, who included 63 endocrinologists, endocrine surgeons, general internists, and oncologists. Patients younger than 18 years were excluded because of limited data on the accuracy of molecular testing for pediatric patients. Patients with a known history of thyroid cancer who were undergoing surveillance ultrasonography were also excluded, as were patients who had other thyroid nodules or lymph nodes that underwent concurrent FNA yielding a malignant cytology result. This study was approved by the UCLA institutional review board. Molecular tests The GEC was developed using a machine-learning algorithm to classify whole-genome transcriptional profiles from indeterminate FNA samples with surgically confirmed benign or malignant histopathology. GEC was initially designed to have a high sensitivity (92%) and negative predictive value (NPV; 96%) for function as a rule out test (17). These performance characteristics were confirmed in independent validation studies (11–13). ThyroSeq was developed using next-generation sequencing to identify point mutations and gene fusions commonly found in thyroid malignancies. Thyroseq v1 targeted 12 cancer genes with 284 mutational hot spots but had insufficient sensitivity to serve as an effective rule out test in the classification of indeterminate thyroid nodules (18). Thyroseq v2 targets 13 genes in addition to 42 gene fusion products, with an improved sensitivity (90%) and specificity (93%), enabling its use as both rule in and rule out tests (14). Independent validation studies confirmed the high NPV of ThyroSeq v2 but reported lower positive predictive values (PPVs) (16, 19). Study design and cytopathology All eligible patients were informed about the study and consented to participation before biopsy. Patients were block randomized by month to a single molecular test (either GEC or ThyroSeq v2), such that the same molecular test was performed throughout the UCLA Health System for a given month. At the time of the FNA, an additional sample was routinely collected for molecular testing. All specimens underwent centralized cytopathologic review by six head and neck cytopathologists, each with an average of 15 years of experience. Challenging cases were reviewed by a second cytopathologist or in a weekly consensus conference. When the FNA result was indeterminate [Bethesda III: atypia of undetermined significance (AUS) or follicular lesion of undetermined significance (FLUS) or Bethesda IV: follicular neoplasm (FN) or suspicious for follicular neoplasm (SFN)], the molecular test sample was reflexively sent by the pathology department. Treatment recommendations were made by the treating physician (in most cases an endocrinologist; see Supplemental Fig. 1) using clinical judgment and incorporating molecular test results. Outcomes and follow-up Variables analyzed included patient age, sex, largest nodule diameter reported by preoperative cervical ultrasonography, presence of Hürthle cell predominance on FNA cytology, Bethesda diagnostic category, and molecular test result. We also recorded the initial recommendation for management on the basis of molecular test results, whether the patient underwent subsequent surgical resection, and the presence of malignancy on final pathology in the index nodule. Patients who had surgical resection underwent histopathologic evaluation to determine whether the index nodule was benign or malignant. Pathologists were not blinded to molecular test results. When histopathology revealed microcarcinoma (i.e., at least one focus of thyroid carcinoma measuring <1 cm in diameter), the index thyroid nodule was considered malignant only if the malignancy was found in the same thyroid quadrant where the FNA was performed. Nodules were classified as Hürthle cell–predominant according to Bethesda guidelines (20). For AUS/FLUS lesions, this was defined as “a moderately or markedly cellular sample composed of a virtually exclusive population of Hürthle cells, yet the clinical setting suggests a benign Hürthle cell nodule” or “there is a predominance of Hürthle cells in a sparsely cellular aspirate with scant colloid.” For SFN/FN lesions, this was defined as moderate to markedly cellular samples “consisting exclusively (or almost exclusively) of Hürthle cells.” A lesion was classified as Hürthle cell–predominant when the following cytology diagnosis was noted: predominant Hürthloid cell changes, predominance of Hürthle cells, and suspicion for Hürthle cell neoplasm. Patients who did not undergo surgery were followed up with surveillance ultrasonography every 6 months. Significant nodule growth was defined as >50% increase in volume or >20% growth in two dimensions (9). The patient’s treating physician made recommendations, including repeated FNA or surgery, on the basis of follow-up data. Statistical analysis Clinical data including the presence of Hürthle cell predominance was compared between the GEC and ThyroSeq v2 groups. Standardized differences were used to represent differences in effect sizes in baseline characteristics between patients who underwent testing with GEC and ThyroSeq v2. Standardized differences of 0.2, 0.5, and 0.8 represented small, medium, and large effect sizes, respectively. The performance characteristics of GEC and ThyroSeq v2 were evaluated by determining the sensitivity, specificity, and NPVs and PPVs, of identifying a malignant nodule. Patients with benign or negative molecular testing results who did not have surgery were considered to have benign nodules when calculating performance characteristics. Cases of noninvasive follicular thyroid neoplasm with papillarylike nuclear features (NIFTP) were considered malignant when calculating the PPV. Patients with suspicious or positive molecular testing results who did not undergo surgery were excluded when calculating performance characteristics. Fisher’s exact test for binomial proportions was used to estimate 95% confidence intervals (CIs) of each performance characteristic. Observed PPVs of GEC and ThyroSeq v2 were plotted as a function of malignancy prevalence and compared with expected PPVs on the basis of reported performance characteristics of GEC (sensitivity 90%, specificity 53%) (11) and ThyroSeq v2 (sensitivity 91%, specificity 92%) (15). The number of diagnostic hemithyroidectomies avoided was calculated on the basis of patients with negative molecular test results who did not have surgery. Because of the small number of Bethesda IV nodules, all nodules with indeterminate cytology (Bethesda III and IV) were analyzed together. P values <0.05 were considered statistically significant. Analyses were performed using SAS 9.3 (SAS Institute, Inc.; Cary, NC). Results Patients Between 1 May 2016 and 30 June 2017, 1157 patients underwent FNA of 1372 thyroid nodules. Of these, 1030 nodules (75.1%) had benign cytology, 176 nodules (12.8%) had indeterminate cytology (Bethesda III or IV), and 81 nodules (5.9%) were suspicious for malignancy or were malignant (Fig. 1). Of the 176 nodules with indeterminate cytology, 159 nodules were Bethesda III and 17 nodules were Bethesda IV. Nineteen indeterminate nodules were excluded from the study, including six nodules that underwent FNA at a newly acquired radiology facility without resources to participate in the study, five nodules in patients with coexisting thyroid cancer, and six nodules that underwent repeated FNA after an initial insufficient molecular test result. Of 157 eligible indeterminate nodules, the enrollment rate was 94.9% (149 nodules randomly assigned to GEC or ThyroSeq v2 and included in the analysis). Figure 1. View largeDownload slide Flow diagram for all patients who underwent thyroid FNA biopsy during the study period. Four patients with suspicious GEC results are undergoing observation because of patient preference, and an additional four patients are planning to have surgery (no date scheduled). One patient with a positive ThyroSeq v2 mutation is planning to have surgery, but no date has been scheduled. F/u, follow-up. Figure 1. View largeDownload slide Flow diagram for all patients who underwent thyroid FNA biopsy during the study period. Four patients with suspicious GEC results are undergoing observation because of patient preference, and an additional four patients are planning to have surgery (no date scheduled). One patient with a positive ThyroSeq v2 mutation is planning to have surgery, but no date has been scheduled. F/u, follow-up. The baseline characteristics for patients with indeterminate cytology who underwent molecular testing with GEC or ThyroSeq v2 showed no differences with respect to age, sex, or nodule size (Table 1). The median patient age was 59 years, and 81.9% of patients were female. The median nodule size was 2.0 cm (interquartile range 1.5 to 3.0 cm). There was a preponderance of Bethesda III nodules (89.9%) in both the GEC and ThyroSeq v2 groups and a similar rate of Hürthle cell cytology (14.8%). Table 1. Baseline Characteristics of Patients With Indeterminate Thyroid Nodules (n = 149) All GEC ThyroSeq v2 Standardized Difference Median age (IQR), y 59 (45–68) 61 (45–69) 57 (45–65) 0.15 Sex, n (%) Male 27 (18.1%) 13 (18.6%) 14 (17.7%) Female 122 (81.9%) 57 (81.4%) 65 (82.3%) 0.11 Median tumor size (IQR), cm 2.0 (1.5–3.0) 2.0 (1.5–3.1) 2.1 (1.5–3.0) 0.03 0–1 cm, n (%) 16 (10.7%) 8 (11.4%) 8 (10.1%) >1–2 cm, n (%) 62 (41.6%) 31 (44.3%) 31 (39.2%) >2–4 cm, n (%) 54 (36.2%) 24 (34.3%) 30 (38.0%) >4 cm, n (%) 17 (11.4%) 7 (10.0%) 10 (12.7%) Bethesda category, n (%) III (AUS/FLUS) 134 (89.9%) 63 (90.0%) 71 (89.9%) IV (FN/SFN) 15 (10.1%) 7 (10.0%) 8 (10.1%) 0.01 Hürthle cell predominance, n (%) 22 (14.8%) 10 (14.1%) 12 (15.2%) 0.03 All GEC ThyroSeq v2 Standardized Difference Median age (IQR), y 59 (45–68) 61 (45–69) 57 (45–65) 0.15 Sex, n (%) Male 27 (18.1%) 13 (18.6%) 14 (17.7%) Female 122 (81.9%) 57 (81.4%) 65 (82.3%) 0.11 Median tumor size (IQR), cm 2.0 (1.5–3.0) 2.0 (1.5–3.1) 2.1 (1.5–3.0) 0.03 0–1 cm, n (%) 16 (10.7%) 8 (11.4%) 8 (10.1%) >1–2 cm, n (%) 62 (41.6%) 31 (44.3%) 31 (39.2%) >2–4 cm, n (%) 54 (36.2%) 24 (34.3%) 30 (38.0%) >4 cm, n (%) 17 (11.4%) 7 (10.0%) 10 (12.7%) Bethesda category, n (%) III (AUS/FLUS) 134 (89.9%) 63 (90.0%) 71 (89.9%) IV (FN/SFN) 15 (10.1%) 7 (10.0%) 8 (10.1%) 0.01 Hürthle cell predominance, n (%) 22 (14.8%) 10 (14.1%) 12 (15.2%) 0.03 Standardized differences of 0.2, 0.5, and 0.8 represent small, medium, and large effect sizes, respectively. Abbreviation: IQR, interquartile range. View Large Table 1. Baseline Characteristics of Patients With Indeterminate Thyroid Nodules (n = 149) All GEC ThyroSeq v2 Standardized Difference Median age (IQR), y 59 (45–68) 61 (45–69) 57 (45–65) 0.15 Sex, n (%) Male 27 (18.1%) 13 (18.6%) 14 (17.7%) Female 122 (81.9%) 57 (81.4%) 65 (82.3%) 0.11 Median tumor size (IQR), cm 2.0 (1.5–3.0) 2.0 (1.5–3.1) 2.1 (1.5–3.0) 0.03 0–1 cm, n (%) 16 (10.7%) 8 (11.4%) 8 (10.1%) >1–2 cm, n (%) 62 (41.6%) 31 (44.3%) 31 (39.2%) >2–4 cm, n (%) 54 (36.2%) 24 (34.3%) 30 (38.0%) >4 cm, n (%) 17 (11.4%) 7 (10.0%) 10 (12.7%) Bethesda category, n (%) III (AUS/FLUS) 134 (89.9%) 63 (90.0%) 71 (89.9%) IV (FN/SFN) 15 (10.1%) 7 (10.0%) 8 (10.1%) 0.01 Hürthle cell predominance, n (%) 22 (14.8%) 10 (14.1%) 12 (15.2%) 0.03 All GEC ThyroSeq v2 Standardized Difference Median age (IQR), y 59 (45–68) 61 (45–69) 57 (45–65) 0.15 Sex, n (%) Male 27 (18.1%) 13 (18.6%) 14 (17.7%) Female 122 (81.9%) 57 (81.4%) 65 (82.3%) 0.11 Median tumor size (IQR), cm 2.0 (1.5–3.0) 2.0 (1.5–3.1) 2.1 (1.5–3.0) 0.03 0–1 cm, n (%) 16 (10.7%) 8 (11.4%) 8 (10.1%) >1–2 cm, n (%) 62 (41.6%) 31 (44.3%) 31 (39.2%) >2–4 cm, n (%) 54 (36.2%) 24 (34.3%) 30 (38.0%) >4 cm, n (%) 17 (11.4%) 7 (10.0%) 10 (12.7%) Bethesda category, n (%) III (AUS/FLUS) 134 (89.9%) 63 (90.0%) 71 (89.9%) IV (FN/SFN) 15 (10.1%) 7 (10.0%) 8 (10.1%) 0.01 Hürthle cell predominance, n (%) 22 (14.8%) 10 (14.1%) 12 (15.2%) 0.03 Standardized differences of 0.2, 0.5, and 0.8 represent small, medium, and large effect sizes, respectively. Abbreviation: IQR, interquartile range. View Large Performance of GEC and ThyroSeq v2 Of 70 nodules tested with GEC, 42.9% (n = 30) had a benign molecular test result, 48.6% (n = 34) were suspicious, and 8.6% (n = 6) were insufficient for analysis. Of 26 nodules with a suspicious GEC result that were surgically resected, four nodules had malignant histopathology and six nodules were NIFTP (Table 2). Of 79 nodules tested with ThyroSeq v2, 77.2% (n = 61) had a negative molecular test result, 19.0% (n = 15) had a positive mutation identified, and 3.8% (n = 3) were insufficient for analysis. Of 14 mutation-positive nodules that were surgically resected, six nodules had malignant histopathology (including two nodules with BRAF mutations and one nodule with a TERT promoter mutation) (Table 3) and two nodules were NIFTPs. All nodules with benign/negative molecular testing results that were surgically resected because of patient or physician preference had benign histopathology (three nodules for GEC and nine nodules for ThyroSeq v2). Table 2. Performance of GEC and Next-Generation Sequencing in Indeterminate Thyroid Nodules (n = 149) Molecular Test Result GEC ThyroSeq v2 Follow-up Result Malignant Benign Malignant Benign Benign/mutation negative, n 0 31 (3a + 28b) 0 58 (9a + 49b) Suspicious/mutation positive, n 10 16 8 6 Test Performance (Assuming Nodules With Benign/Negative Molecular Testing Are Benign), % (95% CI) Sensitivity 100.0% (69.2%–100%) 100.0% (63.1%–100%) Specificity 66.0% (50.7%–79.1%) 90.6% (80.7%–96.5%) PPV 38.5% (20.2%–59.4%) 57.1% (28.9%–82.3%) NPV 100.0% (88.8%–100%) 100.0% (93.8%–100%) Molecular Test Result GEC ThyroSeq v2 Follow-up Result Malignant Benign Malignant Benign Benign/mutation negative, n 0 31 (3a + 28b) 0 58 (9a + 49b) Suspicious/mutation positive, n 10 16 8 6 Test Performance (Assuming Nodules With Benign/Negative Molecular Testing Are Benign), % (95% CI) Sensitivity 100.0% (69.2%–100%) 100.0% (63.1%–100%) Specificity 66.0% (50.7%–79.1%) 90.6% (80.7%–96.5%) PPV 38.5% (20.2%–59.4%) 57.1% (28.9%–82.3%) NPV 100.0% (88.8%–100%) 100.0% (93.8%–100%) Test Performance (Only Surgically Confirmed Cases), % (95% CI) Sensitivity 100.0% (69.2%–100%) 100.0% (63.1%–100%) Specificity 15.8% (3.4%–39.6%) 60.0% (32.3%–86.7%) PPV 38.5% (20.2%–59.4%) 57.1% (28.9%–82.3%) NPV 100.0% (29.2%–100%) 100.0% (66.4%–100%) Test Performance (Only Surgically Confirmed Cases), % (95% CI) Sensitivity 100.0% (69.2%–100%) 100.0% (63.1%–100%) Specificity 15.8% (3.4%–39.6%) 60.0% (32.3%–86.7%) PPV 38.5% (20.2%–59.4%) 57.1% (28.9%–82.3%) NPV 100.0% (29.2%–100%) 100.0% (66.4%–100%) a Nodules with surgically confirmed benign histopathology. b Nodules with benign/negative molecular testing results undergoing surveillance and assumed to be benign. View Large Table 2. Performance of GEC and Next-Generation Sequencing in Indeterminate Thyroid Nodules (n = 149) Molecular Test Result GEC ThyroSeq v2 Follow-up Result Malignant Benign Malignant Benign Benign/mutation negative, n 0 31 (3a + 28b) 0 58 (9a + 49b) Suspicious/mutation positive, n 10 16 8 6 Test Performance (Assuming Nodules With Benign/Negative Molecular Testing Are Benign), % (95% CI) Sensitivity 100.0% (69.2%–100%) 100.0% (63.1%–100%) Specificity 66.0% (50.7%–79.1%) 90.6% (80.7%–96.5%) PPV 38.5% (20.2%–59.4%) 57.1% (28.9%–82.3%) NPV 100.0% (88.8%–100%) 100.0% (93.8%–100%) Molecular Test Result GEC ThyroSeq v2 Follow-up Result Malignant Benign Malignant Benign Benign/mutation negative, n 0 31 (3a + 28b) 0 58 (9a + 49b) Suspicious/mutation positive, n 10 16 8 6 Test Performance (Assuming Nodules With Benign/Negative Molecular Testing Are Benign), % (95% CI) Sensitivity 100.0% (69.2%–100%) 100.0% (63.1%–100%) Specificity 66.0% (50.7%–79.1%) 90.6% (80.7%–96.5%) PPV 38.5% (20.2%–59.4%) 57.1% (28.9%–82.3%) NPV 100.0% (88.8%–100%) 100.0% (93.8%–100%) Test Performance (Only Surgically Confirmed Cases), % (95% CI) Sensitivity 100.0% (69.2%–100%) 100.0% (63.1%–100%) Specificity 15.8% (3.4%–39.6%) 60.0% (32.3%–86.7%) PPV 38.5% (20.2%–59.4%) 57.1% (28.9%–82.3%) NPV 100.0% (29.2%–100%) 100.0% (66.4%–100%) Test Performance (Only Surgically Confirmed Cases), % (95% CI) Sensitivity 100.0% (69.2%–100%) 100.0% (63.1%–100%) Specificity 15.8% (3.4%–39.6%) 60.0% (32.3%–86.7%) PPV 38.5% (20.2%–59.4%) 57.1% (28.9%–82.3%) NPV 100.0% (29.2%–100%) 100.0% (66.4%–100%) a Nodules with surgically confirmed benign histopathology. b Nodules with benign/negative molecular testing results undergoing surveillance and assumed to be benign. View Large Table 3. Histopathology Diagnosis in Patients With Suspicious Molecular Test Results Who Underwent Surgery (n = 53) GEC Suspicious ThyroSeq v2 Positive Mutation Benign Adenomatous 13 6 (2 NRAS, 1 HRAS, 1 TERT, 1 TSHR, 1 PAX8/PPARG fusion) Hyperplastic 3 0 NIFTP 6 2 (1 NRAS, 1 THADA/IGF2BP3 fusion) Papillary thyroid cancer Classic 1 2 (BRAF) Follicular variant 1 1 (NRAS) Minimally invasive Hürthle cell carcinoma 1 1 (EIF1AX) Minimally invasive follicular thyroid carcinoma 0 1 (PAX8/PPARG fusion) Medullary thyroid cancer 1 0 Poorly differentiated 0 1 (TERT) GEC Suspicious ThyroSeq v2 Positive Mutation Benign Adenomatous 13 6 (2 NRAS, 1 HRAS, 1 TERT, 1 TSHR, 1 PAX8/PPARG fusion) Hyperplastic 3 0 NIFTP 6 2 (1 NRAS, 1 THADA/IGF2BP3 fusion) Papillary thyroid cancer Classic 1 2 (BRAF) Follicular variant 1 1 (NRAS) Minimally invasive Hürthle cell carcinoma 1 1 (EIF1AX) Minimally invasive follicular thyroid carcinoma 0 1 (PAX8/PPARG fusion) Medullary thyroid cancer 1 0 Poorly differentiated 0 1 (TERT) View Large Table 3. Histopathology Diagnosis in Patients With Suspicious Molecular Test Results Who Underwent Surgery (n = 53) GEC Suspicious ThyroSeq v2 Positive Mutation Benign Adenomatous 13 6 (2 NRAS, 1 HRAS, 1 TERT, 1 TSHR, 1 PAX8/PPARG fusion) Hyperplastic 3 0 NIFTP 6 2 (1 NRAS, 1 THADA/IGF2BP3 fusion) Papillary thyroid cancer Classic 1 2 (BRAF) Follicular variant 1 1 (NRAS) Minimally invasive Hürthle cell carcinoma 1 1 (EIF1AX) Minimally invasive follicular thyroid carcinoma 0 1 (PAX8/PPARG fusion) Medullary thyroid cancer 1 0 Poorly differentiated 0 1 (TERT) GEC Suspicious ThyroSeq v2 Positive Mutation Benign Adenomatous 13 6 (2 NRAS, 1 HRAS, 1 TERT, 1 TSHR, 1 PAX8/PPARG fusion) Hyperplastic 3 0 NIFTP 6 2 (1 NRAS, 1 THADA/IGF2BP3 fusion) Papillary thyroid cancer Classic 1 2 (BRAF) Follicular variant 1 1 (NRAS) Minimally invasive Hürthle cell carcinoma 1 1 (EIF1AX) Minimally invasive follicular thyroid carcinoma 0 1 (PAX8/PPARG fusion) Medullary thyroid cancer 1 0 Poorly differentiated 0 1 (TERT) View Large The performance characteristics of GEC and ThyroSeq v2 were evaluated in patients who underwent surgery or who had benign/negative molecular testing results and were treated nonoperatively (GEC, n = 57; ThyroSeq v2, n = 72). The overall prevalence of malignancy in Bethesda III and IV nodules in this group was 13.9% (18 of 129). Assuming that nodules with benign molecular testing that were not surgically resected were truly benign, the sensitivity and NPV for both GEC and ThyroSeq v2 were 100% (Table 2). The specificity was 66.0% (95% CI, 50.7% to 79.1%) for GEC and 90.6% (95% CI, 80.7% to 96.5%) for ThyroSeq v2. The PPV was 38.5% (95% CI, 20.2% to 59.4%) for GEC and 57.1% (95% CI, 28.9% to 82.3%) for ThyroSeq v2. On the basis of previously reported performance characteristics of GEC (11) and ThyroSeq v2 (15) and a prevalence of malignancy of 13.9%, expected PPVs of GEC and ThyroSeq v2 were 23.9% and 65.1%, respectively. The observed PPVs of GEC (38.5%) and ThyroSeq v2 (57.1%) approximated the expected rates (Fig. 2). Figure 2. View largeDownload slide Expected PPV of GEC and ThyroSeq v2 based on previously reported sensitivity and specificity (11, 15). Figure 2. View largeDownload slide Expected PPV of GEC and ThyroSeq v2 based on previously reported sensitivity and specificity (11, 15). Approximately two-thirds of the indeterminate nodules were managed nonoperatively, largely on the basis of benign/negative molecular test results. The remaining one-third was surgically excised (20.1% with lobectomy and 15.4% with total thyroidectomy). Surgery was performed in 30 of 70 nodules tested with GEC (43%) and 23 of 79 nodules tested with ThyroSeq v2 (29%). Diagnostic surgery was avoided on the basis of a negative molecular test in 28 patients tested with GEC (39.1%) and 49 patients tested with ThyroSeq v2 (62.0%). Hürthle cell predominance was present in 22 nodules with indeterminate cytology. Of the 10 nodules tested with GEC, five were benign and five were suspicious. Only one of the four nodules with a suspicious GEC result that were surgically resected had malignant histopathology (minimally invasive Hürthle cell carcinoma). Of the 12 nodules tested with ThyroSeq v2, eight were mutation negative and four were mutation positive. All four mutation-positive nodules were surgically resected, and two nodules had malignant histopathology (one minimally invasive Hürthle cell carcinoma and one poorly differentiated carcinoma), whereas one nodule was an NIFTP. Surveillance of patients treated nonoperatively Nonoperative management was pursued for 87 nodules with benign/negative molecular testing results. Of these, 23 nodules had <6-month follow-up from the time of FNA and thus did not undergone surveillance ultrasonography. Surveillance ultrasonography was available for 46 of 64 nodules (71.9%) after initial FNA (eight nodules at 6 months, 27 nodules at 12 months, and 11 nodules at 18 months). Of these, only one nodule with an initial ThyroSeq v2 negative result had grown significantly at 12 months and underwent repeated FNA (cytology was AUS with a benign GEC result, and the nodule is being observed). Discussion In this pragmatic randomized trial, ThyroSeq v2 had a higher specificity than GEC and allowed more patients to avoid diagnostic thyroid surgery on the basis of a negative molecular test result. In short-term follow-up, all of the patients who were initially treated nonoperatively remained under observation and did not require surgery. The increased utilization of neck imaging has created a worldwide epidemic of thyroid nodules. In the United States alone, ∼100,000 cytologically indeterminate thyroid nodules are managed annually (5). The advent of molecular testing has enhanced preoperative diagnosis and allowed more patients with benign nodules to avoid diagnostic surgery. GEC and ThyroSeq v2 are the two best-validated molecular tests for indeterminate thyroid nodules. GEC analyzes changes in the expression of 167 genes using a proprietary machine-learning algorithm. A prospective multi-institution validation study of 210 Bethesda III and IV nodules tested with GEC reported a sensitivity of 92%, a specificity of 52%, an NPV of 95%, and a PPV of 38% (11). Independent validation studies have reported an NPV ranging between 70% and 100% and a PPV ranging between 14% and 44% (13, 21, 22). ThyroSeq v2 quantitatively assesses the proportion of cells carrying common genes mutated in thyroid cancer (e.g.,BRAF and RAS), as well as gene fusions (e.g.,RET/PTC). ThyroSeq v2 has a reported sensitivity of 90%, a specificity of 93%, an NPV of 96%, and a PPV as high as 77% to 83% (14, 15), although external validation has suggested that the specificity and PPV may be lower than initially reported (77% and 40% to 50%, respectively) (16, 23). In our study, ThyroSeq v2 displayed a higher specificity and PPV than did GEC. Diagnostic hemithyroidectomy was avoided in 39% of patients tested with GEC and 62% tested with ThyroSeq v2 on the basis of a benign/negative molecular test result. Only 12 of 89 nodules with benign/negative molecular testing results were surgically resected (three with GEC and nine with ThyroSeq v2), and all had benign histopathology. Because the majority of nodules with benign molecular testing were managed nonoperatively, false negatives may be present that would decrease the sensitivity and NPV of either GEC or ThyroSeq v2. Thus far, surveillance ultrasonography is available for 46 nodules after initial FNA, and only one nodule has grown and undergone repeated FNA with a benign GEC result. All nonoperatively treated patients are continuing to undergo biannual ultrasonography surveillance, which should ultimately detect the majority of false-negative cases (24). The institutional prevalence of malignancy in nodules with Bethesda III and IV cytology varies widely between 6% and 50% (25). This variability limits direct comparison of performance characteristics between GEC and ThyroSeq v2 across prior studies. In contrast, our study comparing GEC with ThyroSeq v2 at a single institution with centralized cytopathologic review permits valid comparison of molecular testing techniques. The 13.9% prevalence of malignancy in our study lies in the middle of the range reported in the literature and makes our results generalizable for a wide range of practice settings. A large proportion of patients in our study had Bethesda III cytology. However, this is most likely due to institutional standards for interpreting thyroid FNAs, which are known to have interobserver variability (26). Three nodules with indeterminate cytology in the ThyroSeq v2 group had oncogenic mutations associated with a very high risk of malignancy (BRAF or TERT). Although these mutations are much more commonly observed in nodules that are suspicious for malignancy or frankly malignant cytology, they have been reported in 5% to 15% of nodules with AUS/FLUS or SFN/FN cytology (27–29). The recent reclassification of encapsulated follicular variant of papillary thyroid carcinoma (previously considered malignant) to NIFTP (now considered premalignant) also affects molecular test performance (30, 31). We considered NIFTP as malignant when calculating molecular test performance because the recommended treatment of NIFTP is surgical excision (32). Previous reports have cited high rates of total thyroidectomy for patients with suspicious molecular test results (33). In our study, 57% of surgically treated patients underwent lobectomy, whereas 43% underwent total thyroidectomy on the basis of contralateral thyroid nodules or preexisting hypothyroidism. Consistent with previous reports (34–36), 44% of surgically resected nodules that had suspicious/positive molecular testing results were diagnosed as NIFTPs on histopathology. This finding highlights the need to avoid unnecessary total thyroidectomy in the absence of appropriate clinical factors. Our study has several limitations. We did not perform both molecular tests in all indeterminate thyroid nodules. Although this would have allowed a more direct comparison between GEC and ThyroSeq v2 in each nodule, such a study design would have been cost prohibitive. Pathologists were also not blinded to the molecular tests, and histopathologic diagnosis may have been influenced by prior knowledge of the molecular test results. An additional limitation was the nonoperative treatment of most patients with benign or negative molecular test results. Although the pragmatic study design demonstrated the real-world utility of molecular testing, the lack of histopathologic confirmation for all nodules may decrease the accuracy of test performance measures. Some patients with benign or negative molecular test results may represent false negatives, which would decrease the sensitivity and NPV. Recent data show that undiagnosed thyroid cancers may grow with a mean doubling time of 2.2 years during long-term follow-up; therefore, ultrasonography surveillance will likely detect false-negative cases with longer follow-up (37). Because of the slow rate of progression of most thyroid cancers, however, ≥10 years may be necessary to establish a definitive benign diagnosis in nodules managed nonoperatively. Recently, the Genomic Sequencing Classifier (Veracyte, Inc.) and ThryoSeq v3 (CBLPath, Inc.) were introduced (38, 39). The results reported in this study apply only to the GEC and Thyroseq v2, and a comparison between updated versions may have different results. In conclusion, the advent of molecular testing has allowed many patients to avoid unnecessary diagnostic hemithyroidectomy. On the basis of our findings, Thyroseq v2 had a higher specificity than the Afirma GEC and allowed more patients to avoid surgery. Long-term ultrasonography surveillance is necessary to detect potential false-negative cases that may decrease the sensitivity of molecular testing. Abbreviations: Abbreviations: AUS atypia of undetermined significance CI confidence interval FLUS follicular lesion of undetermined significance FN follicular neoplasm FNA fine-needle aspiration GEC gene expression classifier NIFTP noninvasive follicular thyroid neoplasm with papillarylike nuclear features NPV negative predictive value PPV positive predictive value SFN suspicious for follicular neoplasm UCLA University of California, Los Angeles Acknowledgments Clinical Trial Information: ClinicalTrials.gov no. NCT02681328 (registered 12 February 2016). Disclosures: The authors have nothing to disclose. References 1. Lim H , Devesa SS , Sosa JA , Check D , Kitahara CM . Trends in thyroid cancer incidence and mortality in the United States, 1974-2013 . JAMA . 2017 ; 317 ( 13 ): 1338 – 1348 . Google Scholar CrossRef Search ADS PubMed 2. Davies L , Welch HG . Increasing incidence of thyroid cancer in the United States, 1973-2002 . JAMA . 2006 ; 295 ( 18 ): 2164 – 2167 . Google Scholar CrossRef Search ADS PubMed 3. Ahn HS , Kim HJ , Welch HG . Korea’s thyroid-cancer “epidemic”: screening and overdiagnosis . N Engl J Med . 2014 ; 371 ( 19 ): 1765 – 1767 . Google Scholar CrossRef Search ADS PubMed 4. Ross DS . Nonpalpable thyroid nodules: managing an epidemic . J Clin Endocrinol Metab . 2002 ; 87 ( 5 ): 1938 – 1940 . Google Scholar PubMed 5. Sosa JA , Hanna JW , Robinson KA , Lanman RB . Increases in thyroid nodule fine-needle aspirations, operations, and diagnoses of thyroid cancer in the United States . Surgery . 2013 ; 154 ( 6 ): 1420 – 1426, discussion 1426–1427 . Google Scholar CrossRef Search ADS PubMed 6. Hegedüs L . Clinical practice: the thyroid nodule . N Engl J Med . 2004 ; 351 ( 17 ): 1764 – 1771 . Google Scholar CrossRef Search ADS PubMed 7. Cibas ES , Ali SZ . The Bethesda System for reporting thyroid cytopathology . Thyroid . 2009 ; 19 ( 11 ): 1159 – 1165 . Google Scholar CrossRef Search ADS PubMed 8. Valderrabano P , Khazai L , Thompson ZJ , Leon ME , Otto KJ , Hallanger-Johnson JE , Wadsworth JT , Wenig BM , Chung CH , Centeno BA , McIver B . Cancer risk stratification of indeterminate thyroid nodules: a cytological approach . Thyroid . 2017 ; 27 ( 10 ): 1277 – 1284 . Google Scholar CrossRef Search ADS PubMed 9. Haugen BR , Alexander EK , Bible KC , Doherty GM , Mandel SJ , Nikiforov YE , Pacini F , Randolph GW , Sawka AM , Schlumberger M , Schuff KG , Sherman SI , Sosa JA , Steward DL , Tuttle RM , Wartofsky L . 2015 American Thyroid Association management guidelines for adult patients with thyroid nodules and differentiated thyroid cancer: the American Thyroid Association Guidelines Task Force on Thyroid Nodules and Differentiated Thyroid Cancer . Thyroid . 2016 ; 26 ( 1 ): 1 – 133 . Google Scholar CrossRef Search ADS PubMed 10. Zhang M , Lin O . Molecular testing of thyroid nodules: a review of current available tests for fine-needle aspiration specimens . Arch Pathol Lab Med . 2016 ; 140 ( 12 ): 1338 – 1344 . Google Scholar CrossRef Search ADS PubMed 11. Alexander EK , Kennedy GC , Baloch ZW , Cibas ES , Chudova D , Diggans J , Friedman L , Kloos RT , LiVolsi VA , Mandel SJ , Raab SS , Rosai J , Steward DL , Walsh PS , Wilde JI , Zeiger MA , Lanman RB , Haugen BR . Preoperative diagnosis of benign thyroid nodules with indeterminate cytology . N Engl J Med . 2012 ; 367 ( 8 ): 705 – 715 . Google Scholar CrossRef Search ADS PubMed 12. Alexander EK , Schorr M , Klopper J , Kim C , Sipos J , Nabhan F , Parker C , Steward DL , Mandel SJ , Haugen BR . Multicenter clinical experience with the Afirma gene expression classifier . J Clin Endocrinol Metab . 2014 ; 99 ( 1 ): 119 – 125 . Google Scholar CrossRef Search ADS PubMed 13. McIver B , Castro MR , Morris JC , Bernet V , Smallridge R , Henry M , Kosok L , Reddi H . An independent study of a gene expression classifier (Afirma) in the evaluation of cytologically indeterminate thyroid nodules . J Clin Endocrinol Metab . 2014 ; 99 ( 11 ): 4069 – 4077 . Google Scholar CrossRef Search ADS PubMed 14. Nikiforov YE , Carty SE , Chiosea SI , Coyne C , Duvvuri U , Ferris RL , Gooding WE , Hodak SP , LeBeau SO , Ohori NP , Seethala RR , Tublin ME , Yip L , Nikiforova MN . Highly accurate diagnosis of cancer in thyroid nodules with follicular neoplasm/suspicious for a follicular neoplasm cytology by ThyroSeq v2 next-generation sequencing assay . Cancer . 2014 ; 120 ( 23 ): 3627 – 3634 . Google Scholar CrossRef Search ADS PubMed 15. Nikiforov YE , Carty SE , Chiosea SI , Coyne C , Duvvuri U , Ferris RL , Gooding WE , LeBeau SO , Ohori NP , Seethala RR , Tublin ME , Yip L , Nikiforova MN . Impact of the multi-gene ThyroSeq next-generation sequencing assay on cancer diagnosis in thyroid nodules with atypia of undetermined significance/follicular lesion of undetermined significance cytology . Thyroid . 2015 ; 25 ( 11 ): 1217 – 1223 . Google Scholar CrossRef Search ADS PubMed 16. Valderrabano P , Khazai L , Leon ME , Thompson ZJ , Ma Z , Chung CH , Hallanger-Johnson JE , Otto KJ , Rogers KD , Centeno BA , McIver B . Evaluation of ThyroSeq v2 performance in thyroid nodules with indeterminate cytology . Endocr Relat Cancer . 2017 ; 24 ( 3 ): 127 – 136 . Google Scholar CrossRef Search ADS PubMed 17. Chudova D , Wilde JI , Wang ET , Wang H , Rabbee N , Egidio CM , Reynolds J , Tom E , Pagan M , Rigl CT , Friedman L , Wang CC , Lanman RB , Zeiger M , Kebebew E , Rosai J , Fellegara G , LiVolsi VA , Kennedy GC . Molecular classification of thyroid nodules using high-dimensionality genomic data . J Clin Endocrinol Metab . 2010 ; 95 ( 12 ): 5296 – 5304 . Google Scholar CrossRef Search ADS PubMed 18. Nikiforova MN , Wald AI , Roy S , Durso MB , Nikiforov YE . Targeted next-generation sequencing panel (ThyroSeq) for detection of mutations in thyroid cancer . J Clin Endocrinol Metab . 2013 ; 98 ( 11 ): E1852 – E1860 . Google Scholar CrossRef Search ADS PubMed 19. Taye A , Gurciullo D , Miles BA , Gupta A , Owen RP , Inabnet WB III , Beyda JN , Marti JL . Clinical performance of a next-generation sequencing assay (ThyroSeq v2) in the evaluation of indeterminate thyroid nodules . Surgery . 2018 ; 163 ( 1 ): 97 – 103 . Google Scholar CrossRef Search ADS PubMed 20. Cibas ES , Ali SZ ; NCI Thyroid FNA State of the Science Conference . The Bethesda System for reporting thyroid cytopathology . Am J Clin Pathol . 2009 ; 132 ( 5 ): 658 – 665 . Google Scholar CrossRef Search ADS PubMed 21. Al-Qurayshi Z , Deniwar A , Thethi T , Mallik T , Srivastav S , Murad F , Bhatia P , Moroz K , Sholl AB , Kandil E . Association of malignancy prevalence with test properties and performance of the gene expression classifier in indeterminate thyroid nodules . JAMA Otolaryngol Head Neck Surg . 2017 ; 143 ( 4 ): 403 – 408 . Google Scholar CrossRef Search ADS PubMed 22. Marti JL , Avadhani V , Donatelli LA , Niyogi S , Wang B , Wong RJ , Shaha AR , Ghossein RA , Lin O , Morris LG , Ho AS . Wide inter-institutional variation in performance of a molecular classifier for indeterminate thyroid nodules . Ann Surg Oncol . 2015 ; 22 ( 12 ): 3996 – 4001 . Google Scholar CrossRef Search ADS PubMed 23. Shrestha RT , Evasovich MR , Amin K , Radulescu A , Sanghvi TS , Nelson AC , Shahi M , Burmeister LA . Correlation between histological diagnosis and mutational panel testing of thyroid nodules: a two-year institutional experience . Thyroid . 2016 ; 26 ( 8 ): 1068 – 1076 . Google Scholar CrossRef Search ADS PubMed 24. Angell TE , Vyas CM , Medici M , Wang Z , Barletta JA , Benson CB , Cibas ES , Cho NL , Doherty GM , Doubilet PM , Frates MC , Gawande AA , Heller HT , Kim MI , Krane JF , Marqusee E , Moore FD Jr , Nehs MA , Zavacki AM , Larsen PR , Alexander EK . Differential growth rates of benign vs. malignant thyroid nodules . J Clin Endocrinol Metab . 2017 ; 102 ( 12 ): 4642 – 4647 . Google Scholar CrossRef Search ADS PubMed 25. Iskandar ME , Bonomo G , Avadhani V , Persky M , Lucido D , Wang B , Marti JL . Evidence for overestimation of the prevalence of malignancy in indeterminate thyroid nodules classified as Bethesda category III . Surgery . 2015 ; 157 ( 3 ): 510 – 517 . Google Scholar CrossRef Search ADS PubMed 26. Clary KM , Condel JL , Liu Y , Johnson DR , Grzybicki DM , Raab SS . Interobserver variability in the fine needle aspiration biopsy diagnosis of follicular lesions of the thyroid gland . Acta Cytol . 2005 ; 49 ( 4 ): 378 – 382 . Google Scholar CrossRef Search ADS PubMed 27. Kleiman DA , Sporn MJ , Beninato T , Crowley MJ , Nguyen A , Uccelli A , Scognamiglio T , Zarnegar R , Fahey TJ III . Preoperative BRAF(V600E) mutation screening is unlikely to alter initial surgical treatment of patients with indeterminate thyroid nodules: a prospective case series of 960 patients . Cancer . 2013 ; 119 ( 8 ): 1495 – 1502 . Google Scholar CrossRef Search ADS PubMed 28. Rowe LR , Bentz BG , Bentz JS . Utility of BRAF V600E mutation detection in cytologically indeterminate thyroid nodules . Cytojournal . 2006 ; 3 ( 1 ): 10 . Google Scholar CrossRef Search ADS PubMed 29. Censi S , Cavedon E , Bertazza L , Galuppini F , Watutantrige-Fernando S , De Lazzari P , Nacamulli D , Pennelli G , Fassina A , Iacobone M , Casal Ide E , Vianello F , Barollo S , Mian C . Frequency and significance of Ras, Tert promoter, and Braf mutations in cytologically indeterminate thyroid nodules: a monocentric case series at a tertiary-level endocrinology unit . Front Endocrinol (Lausanne) . 2017 ; 8 : 273 . Google Scholar CrossRef Search ADS PubMed 30. Liu J , Singh B , Tallini G , Carlson DL , Katabi N , Shaha A , Tuttle RM , Ghossein RA . Follicular variant of papillary thyroid carcinoma: a clinicopathologic study of a problematic entity . Cancer . 2006 ; 107 ( 6 ): 1255 – 1264 . Google Scholar CrossRef Search ADS PubMed 31. Nikiforov YE , Seethala RR , Tallini G , Baloch ZW , Basolo F , Thompson LD , Barletta JA , Wenig BM , Al Ghuzlan A , Kakudo K , Giordano TJ , Alves VA , Khanafshar E , Asa SL , El-Naggar AK , Gooding WE , Hodak SP , Lloyd RV , Maytal G , Mete O , Nikiforova MN , Nosé V , Papotti M , Poller DN , Sadow PM , Tischler AS , Tuttle RM , Wall KB , LiVolsi VA , Randolph GW , Ghossein RA . Nomenclature revision for encapsulated follicular variant of papillary thyroid carcinoma: a paradigm shift to reduce overtreatment of indolent tumors . JAMA Oncol . 2016 ; 2 ( 8 ): 1023 – 1029 . Google Scholar CrossRef Search ADS PubMed 32. Haugen BR , Sawka AM , Alexander EK , Bible KC , Caturegli P , Doherty GM , Mandel SJ , Morris JC , Nassar A , Pacini F , Schlumberger M , Schuff K , Sherman SI , Somerset H , Sosa JA , Steward DL , Wartofsky L , Williams MD . American Thyroid Association guidelines on the management of thyroid nodules and differentiated thyroid cancer task force review and recommendation on the proposed renaming of encapsulated follicular variant papillary thyroid carcinoma without invasion to noninvasive follicular thyroid neoplasm with papillary-like nuclear features . Thyroid . 2017 ; 27 ( 4 ): 481 – 483 . Google Scholar CrossRef Search ADS PubMed 33. Hang JF , Westra WH , Cooper DS , Ali SZ . The impact of noninvasive follicular thyroid neoplasm with papillary-like nuclear features on the performance of the Afirma gene expression classifier . Cancer Cytopathol . 2017 ; 125 ( 9 ): 683 – 691 . Google Scholar CrossRef Search ADS PubMed 34. Samulski TD , LiVolsi VA , Wong LQ , Baloch Z . Usage trends and performance characteristics of a “gene expression classifier” in the management of thyroid nodules: An institutional experience . Diagn Cytopathol . 2016 ; 44 ( 11 ): 867 – 873 . Google Scholar CrossRef Search ADS PubMed 35. Wong KS , Angell TE , Strickland KC , Alexander EK , Cibas ES , Krane JF , Barletta JA . Noninvasive follicular variant of papillary thyroid carcinoma and the Afirma gene-expression classifier . Thyroid . 2016 ; 26 ( 7 ): 911 – 915 . Google Scholar CrossRef Search ADS PubMed 36. Sahli ZT , Umbricht CB , Schneider EB , Zeiger MA . Thyroid nodule diagnostic markers in the face of the new NIFTP category: time for a reset ? Thyroid . 2017 ; 27 ( 11 ): 1393 – 1399 . Google Scholar CrossRef Search ADS PubMed 37. Tuttle RM , Fagin JA , Minkowitz G , Wong RJ , Roman B , Patel S , Untch B , Ganly I , Shaha AR , Shah JP , Pace M , Li D , Bach A , Lin O , Whiting A , Ghossein R , Landa I , Sabra M , Boucai L , Fish S , Morris LGT . Natural history and tumor volume kinetics of papillary thyroid cancers during active surveillance . JAMA Otolaryngol Head Neck Surg . 2017 ; 143 ( 10 ): 1015 – 1020 . Google Scholar CrossRef Search ADS PubMed 38. Patel KN , Angell TE , Barbiarz J , Barth NM , Blevins TC , Duh Q-Y , Ghossein RA , Harrell RM , Huang J , Imtiaz U , Kennedy GC , Kim SY , Kloos RT . LiVolsi VA , Randolph GW , Sadow PM , Shanik MH , Sosa JA , Traweek ST , Walsh PS , Whitney D , Yeh MW , Ladenson PW . Clinical validation of an improved genomic classifier for cytologically indeterminate thyroid nodules using an Ngs platform and machine learning algorithms in an independent prospective multicenter blinded cohort demonstrates improved performance . In: Proceedings of the 3rd World Congress on Thyroid Cancer; July 27–30, 2017 ; Boston, MA. Abstract OP87. 39. Nikiforova MN , Mercurio S , Wald AI , Barbi de Moura M , Callenberg K , Santana-Santos L , Gooding WE , Yip L , Ferris RL , Nikiforov YE . Analytical performance of the ThyroSeq v3 genomic classifier for cancer diagnosis in thyroid nodules [published online ahead of print January 18, 2018]. Cancer . doi: 10.1002/cncr.31245. Copyright © 2018 Endocrine Society
Journal
Journal of Clinical Endocrinology and Metabolism – Oxford University Press
Published: Mar 23, 2018