A malignancy risk prediction model for Bethesda category III thyroid nodules based on serological indicators, C-TIRADS, and BRAF V600E testing.

Objective: To investigate the diagnostic value of a multimodal prediction model integrating thyroglobulin antibody (TgAb), thyroid peroxidase antibody (TPOAb), the Chinese Thyroid Imaging Reporting and Data System (C-TIRADS), and BRAF V600E mutation status, for evaluating the malignancy risk in Bethesda Category III thyroid nodules. Methods: We retrospectively enrolled 93 patients whose thyroid nodules were diagnosed as Bethesda Category III via fine-needle aspiration and subsequently underwent surgical treatment at Northern Jiangsu People's Hospital between July 2020 and June 2025. Preoperative TgAb levels, TPOAb levels, C-TIRADS classification, and BRAF V600E mutation status data were collected. A malignancy risk prediction model was constructed using Firth's penalized likelihood logistic regression. The diagnostic performance of individual indicators and the combined model was evaluated using receiver operating characteristic (ROC) curves. A nomogram was developed for individualized risk prediction, and the model was internally validated using bootstrap resampling (1000 iterations), calibration curves, and decision curve analysis (DCA) to assess the clinical utility. Results: Among the 93 Bethesda III thyroid nodules, 84 (90.32%) were malignant. Among the individual predictors, BRAF V600E exhibited the highest diagnostic performance (AUC = 0.736, 95% CI: 0.589-0.884). Compared with individual modalities, the combined multimodal model demonstrated a significantly improved diagnostic trend, with an AUC of 0.821 (95% CI: 0.653-0.989). Conclusion: In a surgically enriched, high-risk cohort of patients with Bethesda III thyroid nodules, the multimodal model integrating TgAb, TPOAb, C-TIRADS, and BRAF V600E testing provides excellent auxiliary diagnostic value and can effectively guide preoperative risk stratification.