Molecular-derived risk of malignancy and the related positive call rate of indeterminate thyroid cytology diagnoses as quality metrics for individual cytopathologists.

BACKGROUND: Indeterminate thyroid cytopathology diagnoses represent differing degrees of risk that are corroborated by follow-up studies. However, traditional cytologic-histologic correlation may overestimate the risk of malignancy (ROM) because only a subset of cases undergo resection. Alternatively, some molecular tests provide probability of malignancy data to calculate the molecular-derived risk of malignancy (MDROM) and the positive call rate (PCR). The authors investigated MDROMs and PCRs of indeterminate diagnoses for individual cytopathologists as quality metrics. METHODS: This study was approved by the Department of Pathology Quality Improvement Program. Thyroid cytopathology diagnoses and ThyroSeq v3 results were retrieved for each cytopathologist for a 2-year period with at least 3 years of follow-up for the atypia of undetermined significance (AUS), follicular neoplasia (FN), and follicular neoplasia, oncocytic-type (ONC) cytopathologic diagnoses. MDROMs and PCRs were compared with reference ROMs and cytologic-histologic correlation outcomes. RESULTS: The overall MDROMs (and ranges for cytopathologists) for the AUS, FN, and ONC categories were 13.4% (range, 5.8%-20.8%), 28.1% (range, 22.1%-36.7%), and 27.0% (range, 19.5%-41.5%), respectively, and most individual cytopathologists' MDROMs were within reference ROM ranges. However, PCRs more effectively parsed the differences in cytopathologists' ROM performance. Although the overall PCRs were not significantly different across cytopathologists (p = .06), the AUS PCRs were quite different (p = .002). By cytologic-histologic correlation, six of 55 resected cases (10.9%) were falsely negative, and there were no false-positive cases. CONCLUSIONS: MDROMs and PCRs evaluate concordance with reference ROMs and with one another and provide individual feedback, which potentially facilitates quality improvement.