Clinical application of TBSRTC in thyroid FNAC: a single-center retrospective study

Background Thyroid fine-needle aspiration cytology (FNAC) classified according to the Bethesda System for Reporting Thyroid Cytopathology (TBSRTC) is widely used for the preoperative evaluation of thyroid nodules. However, category distribution, observed risk of malignancy (ROM), and diagnostic performance may vary according to referral patterns, disease prevalence, and selective surgical follow-up. Methods This single-center retrospective study included 2,316 ultrasound-guided thyroid fine-needle aspiration (FNA) specimens from 1,791 patients evaluated at a tertiary cancer center between 2019 and 2022. Bethesda category distribution was assessed in the entire cohort. Observed ROM was calculated in 1,137 FNA sampling events with matched postoperative histopathology. Binary diagnostic performance was evaluated in 1,068 sampling events after exclusion of 42 category III and 27 category IV sampling events. Results Bethesda categories I–VI comprised 49 (2.1%), 762 (32.9%), 150 (6.5%), 89 (3.8%), 191 (8.2%), and 1,075 (46.4%) specimens, respectively. Among the 1,137 nodules with histopathologic follow-up, 156 were benign, 22 were borderline or low-risk neoplasms, and 959 were malignant. When borderline or low-risk neoplasms were classified as nonmalignant, the observed ROM values for categories I–VI were 50.0%, 20.5%, 61.9%, 37.0%, 84.2%, and 99.5%, respectively. Sensitivity, specificity, accuracy, positive predictive value, and negative predictive value were 95.6%, 90.0%, 94.9%, 98.6%, and 74.1%, respectively. False-negative diagnoses were predominantly associated with sampling limitations, including minute malignant foci, stromal fibrosis, thyroid follicular nodular disease, and cystic degeneration. False-positive diagnoses were mainly related to overlapping cytomorphologic features in reactive, inflammatory, hyperplastic, cystic, and treatment-related conditions. Conclusions Thyroid FNAC showed high diagnostic performance in this surgically selected tertiary cancer-center cohort. The observed ROM and predictive values were influenced by referral bias, surgical selection, and partial verification and should not be generalized to unselected thyroid FNA populations. Accurate targeting, representative sampling, cautious interpretation of focal atypia, and strict nodule-level cytology–histology correlation may reduce diagnostic discordance.