Institutional prevalence of malignancy of indeterminate thyroid cytology is necessary but insufficient to accurately interpret molecular marker tests

Abstract

Abstract Objective Several molecular marker tests are available to refine the diagnosis of thyroid nodules. Knowing the true prevalence of malignancy (PoM) within each cytological category is considered necessary to select the most appropriate test and to interpret results accurately. We describe our institutional PoM among cytological categories and report our experience with molecular markers. Design Single-center retrospective study. Methods We calculated the institutional PoM for each category of the Bethesda system (Bethesda) on all thyroid nodules with cytological evaluation from October 2008 to May 2014. We estimated the predictive values for Afirma, miRInform, and ThyroSeq v2, based on published sensitivity and specificity. Finally, we assessed our own experience with miRInform. Results The PoMs for Bethesda III and IV categories were 21 and 28%, respectively. ThyroSeq v2 achieves the highest theoretical negative and positive predictive values (NPV and PPV) in Bethesda III (98 and 75%) and Bethesda IV categories (96 and 83%). At our institution, miRInform detected a mutation in 16% of 109 indeterminate nodules tested, all in Bethesda IV specimens. Histology was available in 56 (51%) nodules. The observed sensitivity and specificity in Bethesda IV specimens were 63 and 86%, yielding an NPV and a PPV of 75 and 77%, respectively. Conclusions For our current Bethesda III and IV PoM, the actual performance of miRInform was worse than expected. Theoretically ThyroSeq v2 should have the best performance, but it could be affected in the same way as miRInform, given the similarities between the tests. Assessing the institutional performance of each test is necessary along with PoM individualization.