Background: Increasingly, patients with thyroid nodule cytology labeled atypical (or follicular lesion) of undetermined significance (AUS/FLUS) or follicular neoplasm (FN) undergo diagnostic analysis with the Afirma gene expression classifier (GEC). No long-term, multisite analysis of Afirma GEC performance has yet been performed.
Methods: We analyzed all patients who had received Afirma GEC testing at five academic medical centers between 2010 and 2013. Nodule and patient characteristics, fine needle aspiration cytology, Afirma GEC results, and subsequent clinical or surgical follow-up were obtained for 339 patients. Results were analyzed for pooled test performance, impact on clinical care, and site-to-site variation.
Results: Three hundred thirty-nine patients underwent Afirma GEC testing of cytologically indeterminate nodules (165 AUS/FLUS; 161 FN; 13 suspicious for malignancy) and 174 of 339 (51%) indeterminate nodules were GEC benign, whereas 148 GEC were suspicious (44%). GEC results significantly altered care recommendations, as 4 of 175 GEC benign were recommended for surgery in comparison to 141 of 149 GEC suspicious (P<.01). Of 121 Cyto Indeterminate/GEC Suspicious nodules surgically removed, 53 (44%) were malignant. Variability in site-to-site GEC performance was confirmed, as the proportion of GEC benign varied up to 29% (P=.58), whereas the malignancy rate in nodules cytologically indeterminate/GEC suspicious varied up to 47% (P=.11). Seventy-one of 174 GEC benign nodules had documented clinical follow-up for an average of 8.5 months, in which 1 of 71 nodules proved cancerous.
Conclusions: These multicenter, clinical experience data confirm originally published Afirma GEC test performance and demonstrate its substantial impact on clinical care recommendations. Although nonsignificant site-to-site variation exists, such differences should be anticipated by the practicing clinician. Follow-up of GEC benign nodules thus far confirm the clinical utility of this diagnostic test.