Integrating Digital Ki-67 Labeling Index and K-TIRADS for Malignancy Risk Stratification in Thyroid Core Needle Biopsies

Background/Objectives: Preoperative risk stratification of indeterminate thyroid nodules, particularly category IV nodules diagnosed as follicular neoplasm, remains challenging. Molecular testing may support clinical decision-making, but its cost and limited availability restrict routine use. This study evaluated whether the digitally quantified Ki-67 labeling index, alone and combined with the Korean Thyroid Imaging Reporting and Data System (K-TIRADS), could predict malignancy in thyroid core needle biopsy (CNB) specimens. Methods: We retrospectively analyzed 130 thyroid nodules sampled by ultrasound-guided CNB. The Ki-67 labeling index was digitally quantified in immunohistochemically stained CNB sections, and ultrasound features were classified according to K-TIRADS. Receiver operating characteristic curve analysis was performed in the overall cohort. Logistic regression analyses were restricted to 57 category IV nodules. Results: In the overall cohort, the Ki-67 labeling index showed excellent diagnostic performance for predicting malignancy, with an area under the curve of 0.924. The optimal cutoff was 1.73%, with 88.7% sensitivity and 83.1% specificity. In category IV nodules, K-TIRADS category and Ki-67 labeling index were independently associated with malignancy. The odds ratios were 6.02 (95% CI, 1.53–23.74) for each one-category increase in K-TIRADS and 4.44 (95% CI, 1.69–11.70) for each percentage-point increase in Ki-67. In the Ki-67-only model, a Ki-67 index of 5% corresponded to a predicted malignancy probability of 98.1%. Conclusions: Digital Ki-67 quantification, integrated with K-TIRADS, may serve as a practical adjunct for malignancy risk stratification in thyroid CNB specimens, particularly for category IV nodules. Further external validation is warranted.