Comparison of the Diagnostic Performances of ACR-TIRADS, EU-TIRADS, and Bethesda Classifications in Predicting Malignancy in Thyroid Nodules

Objective: To compare the diagnostic performance of ultrasound-based risk stratification systems (ACR-TIRADS and EU-TIRADS) with the Bethesda classification in predicting thyroid nodule malignancy. Methods: This prospective cohort study included consecutive patients undergoing ultrasound-guided fine-needle aspiration biopsy between January 3 and December 30, 2024. Among 239 enrolled patients, 154 with histopathological confirmation following thyroidectomy were analyzed. Nodules were classified using ACR-TIRADS and EU-TIRADS criteria; cytological findings were categorized via the Bethesda system. Diagnostic performance was evaluated using receiver operating characteristic (ROC) curve analysis, and independent predictors of malignancy were identified through logistic regression. Results: Among 154 operated patients, histopathology confirmed malignancy in 40.3% of nodules. Malignancy rates increased significantly with advancing categories across all 3 systems (P < .001). Individual performances were moderate and statistically similar: ACR-TIRADS (area under the curve [AUC], 0.72), EU-TIRADS (AUC, 0.71), and Bethesda (AUC, 0.72) (ACR-TIRADS vs EU-TIRADS, DeLong test, P = .70). Accuracy improved when all systems were combined (AUC, 0.78). On multivariable analysis, only Bethesda remained an independent predictor of malignancy (odds ratio, 1.96; P < .001). In the cytologically indeterminate Bethesda III–IV subgroup (n = 73), both TIRADS systems demonstrated high sensitivity (85.2%) but low specificity (37.0% to 39.1%; AUC range, 0.65 to 0.68). Conclusion: Bethesda cytology was the strongest independent predictor of thyroid malignancy. ACR-TIRADS and EU-TIRADS provide valuable complementary roles in determining biopsy indications, particularly for cytologically indeterminate nodules. Combined assessment optimizes diagnostic accuracy and supports clinical decision-making.