BackgroundThere is limited data comparing the performance of Afirma Genomic Sequencing Classifier (GSC) in thyroid nodules carrying an initial versus a repeat diagnosis of atypia of undetermined significance (AUS). This study reported an institutional experience in this regard.Materials and MethodsThis retrospective study included consecutive thyroid nodules that had an initial or a repeat AUS diagnosis and had a subsequent GSC diagnostic result (benign or suspicious) from 2017 to 2021. All nodules were followed by surgical intervention or by clinical and/or ultrasound monitoring. GSC's benign call rate (BCR), rate of histology‐proven malignancy associated with a suspicious GSC result, and diagnostic parameters of GSC were calculated and compared between the two cohorts (initial versus repeat AUS). Statistical significance was defined with a p‐value of <.05 for all analysis.ResultsA total of 202 cases fulfilled inclusion criteria, including 67 and 135 thyroid nodules with an initial and a repeat AUS diagnosis, respectively. BCR was 67% and 66% in initial and repeat AUS cohorts, respectively. Rate of histology‐proven malignancy associated with a suspicious GSC result were 22% and 24% in initial and repeat AUS cohorts, respectively. Compared with the repeat AUS cohort, the initial AUS cohort showed slightly lower sensitivity (83% vs. 100%), specificity (70% vs. 73%), PPV (23% vs. 24%), NPV (98% vs. 100%), and diagnostic accuracy (72% vs. 75%). Nevertheless, these differences did not reach statistical significance.ConclusionGSC demonstrated comparable performance in thyroid nodules with a repeat AUS diagnosis versus nodules with an initial AUS diagnosis.