Abstract
Background: Molecular analysis of indeterminate thyroid nodules has been shown to reduce unnecessary surgeries for benign thyroid nodules. Afirma Gene Expression Classifier (GEC) has a reported benign call rate (BCR) of 38-61%, and the newer generation Genomic Sequencing Classifier (GSC) has a reported BCR of 42-76%. Both GEC and GSC have a high sensitivity of 88-100% and negative predictive value (NPV) of 94-100%. Specificity and positive predictive value (PPV) have improved with GSC when compared to GEC, largely due to improvements in the classification of hurtle cell neoplasms. GSC PPV is reported as 47-76%. The aim of our study was to determine surgical and malignancy rates for patients at our institution with GEC, GSC, or for those who did not undergo molecular testing.
Methods: Retrospective analysis of all Bethesda III and IV nodules with available follow up data at our institution between January 2013 and December 2018. We evaluated cytopathology, molecular testing with Afirma GEC or GSC, surgical rate, and malignancy rate between groups. Univariate descriptive statistics and bivariate analyses were calculated using Chi-squared and Fisher’s exact testing for categorical variables and one-way analysis of variance for continuous variables.
Results: A total of 376 patients had Bethesda III and IV indeterminate thyroid nodule biopsies and met inclusion criteria for analysis. Of these, 262 patients did not undergo molecular testing, 50 underwent GEC, and 64 GSC testing. There was no difference between the groups for gender, age, or BMI. The overall surgical rate was 66.4, 60.0, and 46.9%, respectively (p=0.014). There was no difference in BCR for GEC or GSC with 44.0% versus 54.7% (p=0.38). There was no difference in malignancy rate with no molecular testing 19.5%, 22.7% if GEC suspicious, and 29.2% if GSC suspicious (p=0.619). GEC had a PPV of 22.7%, NPV of 100%, sensitivity of 100%, and specificity of 56.4% compared with GSC with PPV of 29.2%, NPV of 100%, the sensitivity of 100%, and specificity of 67.3%. Our overall rate of incidentally noted malignancy for indeterminate thyroid nodules who ultimately underwent surgery was 23 of 234 (9.9%).
Conclusions: At our institution, the surgical rate was not different between patients who did not have molecular testing and those with GEC, however, there was a significant reduction in overall surgeries when using GSC. GSC BCR improved when compared to GEC, consistent with previous studies, but PPV remained low at 29.2%, which is lower than previously reported.
Performance of the Afirma genomic sequencing classifier versus gene expression classifier: An institutional experience
