Combining molecular testing and the Bethesda category III:VI ratio for thyroid fine-needle aspirates: A quality-assurance metric for evaluating diagnostic performance in a cytopathology laboratory.

ABSTRACT

BACKGROUND: Molecular testing (MT) of thyroid fine-needle aspiration (FNA)-derived genetic material is commonly used to assess malignancy risk for indeterminate cases. The Bethesda System for Reporting Thyroid Cytopathology (TBS) provides limited guidance for the appropriate use of category III (atypia of undetermined significance [AUS]). The authors combined MT with cytomorphology to monitor AUS diagnoses in a cytopathology laboratory. METHODS: Neoplasia-associated genetic alterations (NGAs) were determined by MT of preoperative FNA biopsies or resected malignancies and were categorized as BRAF V600E mutations, RAS-like mutations (HRAS, NRAS, or KRAS mutations or non-V600E BRAF mutations), or other mutations. RESULTS: Among 7382 thyroid FNA biopsies, the AUS rate was 9.3% overall and ranged from 4.3% to 24.2% among 6 cytopathologists (CPs) who evaluated >150 cases. The ratio of specimens falling into TBS category III to specimens falling into category VI (malignant) (the IIIVI ratio) was 2.4 overall (range, 1.1-8.1), and the ratio of specimens falling into TBS categories III and IV (follicular neoplasm or suspicious for follicular neoplasm) combined (III+IV) to specimens falling into category VI (the [III+IV]VI ratio) was 2.9 overall (range, 1.4-9.5). MT was performed on 588 cases from 560 patients (79% women) with a median age of 56 years (range, 8-89 years). BRAF V600E mutation was the most common (76% of cases) in TBS category VI and was rare (3%) in category III. RAS-like mutations were most common in TBS categories III (13%), IV (25%), and V (suspicious for malignancy) (17.5%). The NGA rate in AUS cases fell between 5% and 20% for 5 of 6 CPs and did not correlate with the IIIVI ratio or the (III+IV)VI ratio. CONCLUSIONS: Lack of correlation between the NGA rate and easily calculable diagnostic ratios enables the calibration of diagnostic thresholds, even for CPs who have normal metrics. Specifically, calculation of the NGA rate and the IIIVI ratio may allow individual CPs to determine whether they are overcalling or undercalling cases that other CPs might otherwise recategorize.