Neuroradiology/Head and Neck Imaging Original Research Zhu et al. Ultrasound Versus Afirma Testing of FNA-Indeterminate Thyroid Nodules Neuroradiology/Head and Neck Imaging Original Research Qing-Li Zhu 1,2 William C. Faquin 3 Anthony E. Samir 2 Zhu QL, Faquin WC, Samir AE Keywords: Afirma, fine-needle aspiration, indeterminate, thyroid nodules, ultrasound DOI:10.2214/AJR.14.13984 Received October 9, 2014; accepted after revision March 24, 2015. 1 Department of Ultrasound, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences, Beijing, China. 2 Department of Radiology, Abdominal Imaging and Intervention, Massachusetts General Hospital, Harvard Medical School, White 270, 55 Fruit St, Boston, MA 02114. Address correspondence to A. E. Samir (asamir@mgh.harvard.edu). 3 Department of Pathology, Massachusetts General Hospital, Harvard Medical School, Boston, MA. AJR 2015; 205:861 865 0361 803X/15/2054 861 American Roentgen Ray Society Relationship Between Sonographic Characteristics and Afirma Gene Expression Classifier Results in Thyroid Nodules With Indeterminate Fine-Needle Aspiration Cytopathology OBJECTIVE. The purpose of this article is to investigate whether specific clinical and sonographic characteristics are predictive of a benign Afirma test result. MATERIALS AND METHODS. We conducted a retrospective study of Afirma gene expression classifier analysis performed in 44 patients with 45 indeterminate thyroid fineneedle aspiration (FNA) cytologic results between March 2013 and April 2014. Of these, 33 of 45 nodules (73.3%) were repeat atypia of undetermined significance (AUS) and follicular lesions of undetermined significance (FLUS), or follicular neoplasm (FN) and suspicious for a follicular neoplasm (SFN) before Afirma testing. RESULTS. Of the 45 nodules, 21 (46.7%) were cytologically diagnosed as FLUS, 16 (35.6%) were diagnosed as AUS, and eight (17.8%) were diagnosed as FN or SFN. By Afirma testing, 23 of the 45 nodules (51.1%) were benign, 21 (46.7%) were suspicious, and one (2.2%) had nondiagnostic results. The mean (± SD) nodule size was smaller in the Afirma-benign group than in the Afirma-suspicious group (1.8 ± 0.8 cm [95% CI, 1.4 2.1] vs 2.2 ± 0.8 cm [95% CI, 1.8 2.6]; p < 0.035). No sonographic feature was statistically significantly different between the Afirma-benign and -suspicious groups, including nodule solidity (p = 0.225), echogenicity (p = 0.543), calcification (p = 0.542), and hypervascularity (p = 0.976). All nodules were ovoid shaped and had circumscribed margins in both Afirma groups. CONCLUSION. Smaller nodule size was the only characteristic associated with a benign diagnosis on Afirma testing. Sonographic characteristics are not helpful in cases that had a repeat indeterminate FNA finding before Afirma testing. F ine-needle aspiration (FNA) cytology has become an essential diagnostic tool in the management of thyroid nodules [1 3]. The Bethesda System for Reporting Thyroid Cytopathology provides a revised six-tiered reporting system with an emphasis on the risk of malignancy associated with each category [4]. The indeterminate groups atypia of undetermined significance (AUS) and follicular lesion of undetermined significance (FLUS), or follicular neoplasm (FN) and suspicious for a follicular neoplasm (SFN) comprise 15 30% of all FNA findings [5] and are associated with an intermediate risk of malignancy (5 15% for AUS-FLUS; 15 30% for FN-SFN) [4 6]. However, most patients with intermediate-risk nodules ultimately undergo diagnostic thyroid surgery and approximately three quarters of the nodules are benign on final surgical pathologic examination [7 9]. It is therefore clinically relevant to improve preoperative malignancy risk stratification in patients with indeterminate cytologic findings on FNA. Recently, a molecular test, the Afirma gene expression classifier (Veracyte), has been developed for further risk stratification in cytologically indeterminate thyroid nodules [10]. This test classifies nodules as either benign or suspicious for malignancy, with a reported high negative predictive value (96%). Because the risk of malignancy in a thyroid nodule with AUS-FLUS or FN-SFN and a benign gene expression classifier diagnosis is reported to be comparable to that of a nodule with benign cytopathologic findings, clinical follow-up has been recommended instead of surgery [11]. However, the Afirma test is costly, and the number of cytologically indeterminate thyroid FNA results is high. It is, therefore, reasonable to consider whether a specific subgroup of patients with indeterminate FNA results might benefit more from Afirma testing than others. For example, one potential AJR:205, October 2015 861
Zhu et al. Fig. 1 59-year-old woman with complex nodule in left thyroid lobe. A, Sagittal (left) and transverse (right) gray-scale ultrasound images show 3.0-cm complex oval nodule with circumscribed margin. B, Color Doppler flow image shows peripheral hypovascularity. Initial ultrasoundguided fine-needle aspiration (FNA) biopsy yielded diagnosis of atypia of undetermined significance (AUS) on cytologic analysis. Repeat ultrasound-guided FNA biopsy also yielded diagnosis of AUS. Afirma (Veracyte) analysis result was benign. Nodule remained unchanged on follow-up. strategy might be to perform confirmatory Afirma testing in patients whose indeterminate nodules on FNA are considered more likely to be benign on the basis of nodule sonographic characteristics and patient clinical characteristics. Clinical and nodule sonographic characteristics are known to be predictive of malignancy risk. For example, in one study of 165 AUS-FLUS nodules, the malignancy rate of nodules was much higher (79.3%) in a selected patient population with suspicious sonographic findings than in a population with indeterminate features on ultrasound (24.7%) [12]. The goal of this study was to investigate whether specific clinical and sonographic characteristics are predictive of a benign Afirma test result. Materials and Methods Patients The institutional review board at Massachusetts General Hospital waived the requirement to obtain informed consent for the retrospective study. Our hospital is a quaternary referral center providing comprehensive care for patients with thyroid cancer, including dedicated thyroid endocrinology, surgery, and nuclear medicine facilities. Afirma gene expression classifier analysis has been performed since March 2013, which was chosen as the starting point of our study. From March 2013 through April 2014, the electronic medical record was reviewed for patients who underwent thyroid FNA biopsy, had received a diagnosis of AUS-FLUS or FN-SFN, according to the Bethesda System for Reporting Thyroid Cytopathology [4], and then were sent for Afirma testing. A total of 44 patients (27 women and 17 men) with 45 thyroid nodules were analyzed in the current study. Of these, 33 of 45 nodules (73.3%) were repeat AUS-FLUS or FN-SFN before Afirma testing. For each subject, we collected demographic, sonographic, cytologic, Afirma gene expression classifier, histopathologic, and subsequent management and outcome data. Thyroid Ultrasound One radiologist with 10 years of experience in thyroid ultrasound reviewed each patient s images. The reviewer took the lesion size measured on the original images as the nodule size and determined the ultrasound features by image review. The original reports were not used to determine image features. The reviewer was blinded to FNA or Afirma results when classifying ultrasound results. All patients had their electronic medical records or charts reviewed for the following sonographic variables: nodule solidity, echogenicity, shape, margin, calcifications, and vascularity. Nodule solidity was classified as solid, predominantly solid (> 50% solid for a mixed nodule), and predominantly cystic (> 50% cyst for a mixed nodule). Nodule shape was classified as ovoid, irregular, or taller than wide. Nodule margin was classified as circumscribed or noncircumscribed (i.e., ill defined). Echogenicity was classified into markedly hypoechoic (i.e., more hypoechoic than the adjacent strap muscle), hypoechoic, isoechoic to thyroid, or hyperechoic. Calcification was classified as no calcifications, microcalcifications (< 1 mm in diameter and visualized as tiny punctuate hyperechoic foci without acoustic shadows), or macrocalcifications (hyperechoic foci > 1 mm in size). A B 862 AJR:205, October 2015
Ultrasound Versus Afirma Testing of FNA-Indeterminate Thyroid Nodules Nodule vascularity was assessed by color Doppler imaging and was classified as hypervascular (i.e., greater vascularity in the nodule than in the normal thyroid parenchyma), hypovascular, or none. Fine-Needle Aspiration Procedure and Afirma Gene-Expression Classifier Analysis Under ultrasound guidance, multiple FNA specimens were obtained using 25-gauge needles. Samples were preserved on alcohol-fixed slides in Cyto- Rich Red (BD Diagnostics-TriPath) and in Afirma testing medium. FNA cytologic findings were read by board-certified subspecialist cytopathologists and were classified according to the Bethesda system [4]. When classified as cytologically indeterminate, either on initial or repeat FNA biopsy, depending on the preference of the referring physician, aspirates were sent for Afirma analysis. Statistical Analysis SPSS statistics software (version 13, IBM) was used for the data analysis. Univariate analysis was done using the Fisher exact test for categoric variables (patient sex), and the independent samples t test was used to compare means of the continuous normal data (patient age). Nodule size was not normally distributed and, thus, the Mann-Whitney U test was used. A p < 0.05 was considered to indicate a statistically significant difference. Results Forty-five Afirma gene expression classifier analyses were performed for 44 patients with 45 indeterminate thyroid nodules between March 2013 and April 2014. The 44 patients in our cohort consisted of 27 (61.4%) women and 17 (38.6%) men, with a mean age of 56.8 ± 13.1 years. None of the patients had a history of ionizing radiation exposure in childhood, thyroid carcinoma, or family history of thyroid carcinoma. Of the 45 nodules, 51.1% (23/45) were reported as Afirma benign, 46.7% (21/45) as Afirma suspicious, and the remaining 2.2% (1/45) as nondiagnostic. The mean nodule size in the Afirma-suspicious group was larger than that in the Afirma-benign group (2.2 ± 0.8 cm [95% CI, 1.8 2.6 cm] vs 1.8 ± 0.8 cm [95% CI, 1.4 2.1 cm]; p = 0.035). No statistically significant differences were observed for patient age (58.3 ± 13.7 years vs 56.6 ± 11.8 years; p = 0.648), sex (p = 0.537), or thyroid function test results (thyroid-stimulating hormone level, 1.7 ± 0.9 miu/l vs 2.7 ± 1.9 miu/l; p = 0.097). Of the 45 nodules, 21 (46.7%) were cytologically diagnosed as FLUS, 16 (35.6%) were diagnosed as AUS, and eight (17.8%) were FN-SFN. For the 21 FLUS nodules, TABLE 1: Ultrasound Features and Afirma (Veracyte) Results in 44 Thyroid Nodules With Indeterminate Fine-Needle Aspiration Cytologic Findings Ultrasound Features Total (n = 44) Afirma analyses were benign in 12 cases and suspicious in nine cases. For the 16 AUS nodules, Afirma analyses were benign in seven cases (Fig. 1), suspicious in eight cases (Fig. 2), and nondiagnostic in one case. This nondiagnostic case was diagnosed as malignant on a subsequent FNA biopsy and then confirmed as papillary carcinoma on thyroidectomy specimen histologic analysis. In eight FN-SFN, Afirma analyses were benign in four cases and suspicious in four cases. Patients with indeterminate cytologic findings and suspicious Afirma results were recommended for surgery, and 10 of 21 (47.6%) completed the surgery. Of these, three of 21 (14.3%) nodules proved to be follicular variant papillary thyroid carcinoma, one (4.8%) was a classic papillary thyroid carcinoma, one (4.8%) was oncocytic follicular variant papillary thyroid carcinoma, one (4.8%) was a follicular carcinoma, two (9.5%) were Hurthle cell adenomas, one (4.8%) was nodular hyperplasia, and one (4.8%) was adenomatous nodule with oncocytic features. Benign (n = 23) Afirma Results Suspicious (n = 21) a Nodule size (cm), mean ± SD (95% CI) 1.8 ± 0.8 (1.4 2.1) 2.2 ± 0.8 (1.8 2.6) 0.035 Component 0.225 Solid 32 19 13 (3 cancers, 2 benign) Complex cystic < 50% 11 4 7 (3 cancers, 2 benign) Complex cystic 50% 1 0 1 Ovoid shape 44 23 21 (6 cancers, 4 benign) Smooth margin 44 23 21 (6 cancers, 4 benign) Echo 0.543 Marked hypoechoic 2 0 2 Hypoechoic 38 20 18 (6 cancers, 3 benign) Isoechoic 4 3 1 (benign) Calcification 0.542 None 38 21 17 (5 cancers, 4 benign) Microcalcification 3 1 2 Mixed calcification 3 1 2 (1 cancer) Vascularity 0.976 Hypervascular 18 10 8 (2 cancers, 2 benign) Hypovascular 15 7 8 (3 cancers) None 3 2 1 (1 cancer) Unknown 8 4 4 (2 benign) Note Except for nodule size, data are number of thyroid nodules. One nondiagnostic Afirma case was not included. a Ten of 21 patients with suspicious Afirma results completed the surgery, and the final pathologic outcomes of the nodules are included in parentheses. Eleven patients did not complete surgery. Of these, five had committed to future surgery that has not yet been performed, four were lost to follow-up, and two declined the recommendation. Patients with a benign Afirma result were all recommended for follow-up, and none of them had undergone thyroid surgery at the end of the data analysis period. Ultrasound features and Afirma Results are shown in Table 1. On preoperative sonography, 73% (32/44) of nodules were solid. Of these, 59% (19/32) were Afirma benign and 41% (13/32) were Afirma suspicious. Twenty-seven percent (12/44) of the nodules were complex nodules. Of these, 33% (4/12) were Afirma benign and 67% (8/12) were Afirma suspicious. This difference was not statistically significant (p = 0.179). Seven of eight complex cystic nodules with an Afirma suspicious diagnosis were more solid than cystic. Surgery revealed three cancers and two benign entities in these seven nodules. One nodule with an Afirma-suspicious diagnosis was predominantly cystic. This nodule was lost p AJR:205, October 2015 863
Zhu et al. A B Fig. 2 27-year-old woman with solid nodule in right thyroid lobe. A, Sagittal gray-scale ultrasound image shows 2.3-cm solid nodule (between calipers) with oval shape and circumscribed margin. B, Color Doppler flow image shows peripheral hypervascularity. Initial ultrasound-guided fine-needle aspiration biopsy yielded diagnosis of atypia of undetermined significance. Afirma (Veracyte) analysis result was suspicious. Nodule was confirmed as follicular variant papillary thyroid carcinoma on surgical pathologic examination. to follow-up. All nodules had an ovoid shape and circumscribed margin. No sonographic feature was associated with Afirma-benign or -suspicious status. Logistic regression with forward stepwise selection showed no predictive power for any combination of factors. Discussion According to the Bethesda system, the malignancy risk of nodules diagnosed as AUS-FLUS on FNA cytology is estimated as 5 15%, and repeat FNA is recommended [4]. However, marked variability in incidence and malignancy in resection specimens has been reported [13]. In a retrospective multicenter study, Alexander et al. [14] reported a malignancy rate of 44% in patients with cytologically indeterminate and suspicious Afirma nodules. Our 70.0% malignant rate in this subgroup was higher than that reported by Alexander et al. This can be explained by the approach used in our institution for selecting patients for Afirma testing. Most (33/45) of our patients underwent repeat FNA biopsy before Afirma testing. Because a repeat FNA biopsy can accurately reclassify more than 50% of nodules in the AUS-FLUS category as benign without the use of ancillary molecular tests [15 17], it is likely that our subject sample contained a higher proportion of malignant nodules than the subjects in the article by Alexander et al. Although clinical and ultrasound features have been shown to be predictive of malignancy in thyroid nodules in general, the literature concerning their utility for the prediction of malignancy in patients with an indeterminate FNA diagnosis is mixed. Some authors have found that ultrasound appearance is helpful to further stratify indeterminate FNA cases into high and low risk for malignancy [12, 18 20]. In a study of 165 AUS-FLUS nodules, Jeong et al. [12] found that the malignancy rate was 79.3% for nodules with suspicious ultrasound features and 24.7% for nodules with indeterminate ultrasound features. Others have shown that sonographic features alone may be insufficient to predict malignancy in these patients [7, 9, 21, 22]. Mehta et al. [21] found that one or more suspicious ultrasound features was identified only in 33% of nodules and occurred regardless of histologic findings. Of the 21 nodules with suspicious Afirma results in our series, none showed suspicious sonographic findings such as taller-than-wide dimension or ill-defined margins, and only two of these nodules showed microcalcification. The reason for the lack of sonographic characteristics associated with malignancy in this group is uncertain. It is possible that nodules with indeterminate cytologic findings are less likely to show sonographic characteristics considered suspicious for malignancy; however, this is speculative, and our study is not powered to make this determination. Nonetheless, this finding highlights the lack of utility of sonographic features for malignancy risk stratification in thyroid nodules indeterminate on FNA biopsy. Limitations Our study has several limitations, including small sample size, limited pathologic followup, and retrospective study design. In addition, it is possible that there is an unrecognized selection bias in the subgroup of patients referred for Afirma analysis. At our institution, radiologists, endocrinologists, and endocrine surgeons perform approximately 2500 thyroid biopsies per year. The decision to proceed to Afirma testing is made by the referring physician and varies considerably among referring physicians. Typically, endocrinologists perform two FNA procedures before Afirma testing, whereas primary care physicians commonly perform Afirma testing after a single nondiagnostic aspirate. Most (33/45) of our subjects underwent two FNA biopsies of the indeterminate nodule before Afirma testing. We did not review all cytopathology reports to isolate the number of subjects with indeterminate FNA results, and we did not compare the characteristics of nodules in patients with indeterminate biopsy samples who were referred for Afirma testing with those who were not referred for Afirma testing. Conclusion Except for nodule size, sonographic characteristics are not useful in cases that had an indeterminate FNA finding before Afirma testing and cannot be used as a substitute for the Afirma gene expression classifier in such patients. Acknowledgments We thank Ronald N. Arpin III and Beverly Pizzi for their help in data collection. References 1. Cooper DS, Doherty GM, Haugen BR, et al.; American Thyroid Association (ATA) Guidelines Taskforce on Thyroid Nodules and Differentiated Thyroid Cancer. Revised American Thyroid Association management guidelines for patients with thyroid nodules and differentiated thyroid 864 AJR:205, October 2015
Ultrasound Versus Afirma Testing of FNA-Indeterminate Thyroid Nodules cancer. Thyroid 2009; 19:1167 1214 rate of thyroid nodules defined as follicular lesion 16. Nayar R, Ivanovic M. The indeterminate thyroid 2. Gharib H, Papini E, Paschke R, et al.; AACE/ of undetermined significance and atypia of unde- fine-needle aspiration: experience from an aca- AME/ETA Task Force on Thyroid Nodules. Amer- termined significance in thyroid cytopathology demic center using terminology similar to that ican Association of Clinical Endocrinologists, As- and its relation with ultrasonographic features. proposed in the 2007 National Cancer Institute sociazione Medici Endocrinologi, and European Endocr Pathol 2014; 25:248 256 Thyroid Fine Needle Aspiration State of the Sci- Thyroid Association. Medical guidelines for clini- 10. Alexander EK, Kennedy GC, Baloch ZW, et al. ence Conference. Cancer 2009; 117:195 202 cal practice for the diagnosis and management of thyroid nodules: executive summary of recommendations. Endocr Pract 2010; 16:468 475 3. Baloch ZW, Cibas ES, Clark DP, et al. The National Cancer Institute thyroid fine needle aspiration state of the science conference: a summation. Cytojournal 2008; 5:6 4. Cibas ES, Ali SZ. The Bethesda system for reporting thyroid cytopathology. Thyroid 2009; 19:1159 1165 5. Daniels GH. What is the role of molecular markers in the management of indeterminate thyroid nodules? Cancer Cytopathol 2013; 121:223 224 6. Dhyani M, Faquin W, Lubitz CC, Daniels GH, Samir AE. How to interpret thyroid fine-needle aspiration biopsy reports: a guide for the busy radiologist in the era of the Bethesda classification system. AJR 2013; 201:1335 1339 7. Ho AS, Sarti EE, Jain KS, et al. Malignancy rate in thyroid nodules classified as Bethesda Category III (AUS/FLUS). Thyroid 2014; 24:832 839 8. Choi YJ, Baek JH, Ha EJ, et al. Differences in risk of malignancy and management recommendations in subcategories of thyroid nodules with atypia of undetermined significance or follicular lesion of undetermined significance: the role of ultrasound-guided core-needle biopsy. Thyroid 2014; 24:494 501 9. Çuhaci N, Arpaci D, Üçler R, et al. Malignancy Preoperative diagnosis of benign thyroid nodules with indeterminate cytology. N Engl J Med 2012; 367:705 715 11. Duick DS, Klopper JP, Diggans JC, et al. The impact of benign gene expression classifier test results on the endocrinologist-patient decision to operate on patients with thyroid nodules with indeterminate fine-needle aspiration cytopathology. Thyroid 2012; 22:996 1001 12. Jeong SH, Hong HS, Lee EH, Cha JG, Park JS, Kwak JJ. Outcome of thyroid nodules characterized as atypia of undetermined significance or follicular lesion of undetermined significance and correlation with ultrasound features and BRAF V600E mutation analysis. AJR 2013; 201:[web]W854 W860 13. Ohori NP, Schoedel KE. Variability in the atypia of undetermined significance/follicular lesion of undetermined significance diagnosis in the Bethesda System for Reporting Thyroid Cytopathology: sources and recommendations. Acta Cytol 2011; 55:492 498 14. Alexander EK, Schorr M, Klopper J, et al. Multicenter clinical experience with the Afirma gene expression classifier. J Clin Endocrinol Metab 2014; 99:119 125 15. Yassa L, Cibas ES, Benson CB, et al. Long-term assessment of a multidisciplinary approach to thyroid nodule diagnostic evaluation. Cancer 2007; 111:508 516 17. Faquin WC, Baloch ZW. Fine-needle aspiration of follicular patterned lesions of the thyroid: diagnosis, management, and follow-up according to National Cancer Institute (NCI) recommendations. Diagn Cytopathol 2010; 38:731 739 18. Gweon HM, Son EJ, Youk JH, Kim JA. Thyroid nodules with Bethesda system III cytology: can ultrasonography guide the next step? Ann Surg Oncol 2013; 20:3083 3088 19. Rosario PW. Thyroid nodules with atypia or follicular lesions of undetermined significance (Bethesda category III): importance of ultrasonography and cytological subcategory. Thyroid 2014; 24:1115 1120 20. Yoo WS, Choi HS, Cho SW, et al. The role of ultrasound findings in the management of thyroid nodules with atypia or follicular lesions of undetermined significance. Clin Endocrinol (Oxf) 2014; 80:735 742 21. Mehta RS, Carty SE, Ohori NP, et al. Nodule size is an independent predictor of malignancy in mutation-negative nodules with follicular lesion of undetermined significance cytology. Surgery 2013; 154:730 736 22. Lee SW, Lee HJ, Kim HJ, et al. Combined categorical reporting systems of US and cytology findings for thyroid nodules: guidance on repeat fine-needle aspiration cytology. Radiology 2013; 266:956 963 AJR:205, October 2015 865