Combined clinicopathological and ultrasonographic features for predicting extrathyroidal extension in thyroid cancer

Extrathyroidal extension (ETE) critically impacts the treatment and prognosis of thyroid cancer, necessitating accurate preoperative prediction models. This study aimed to develop and validate such a model combining clinicopathological and ultrasonographic features.

This retrospective study enrolled 435 thyroid cancer patients from The First Affiliated Hospital of Nanchang University, categorized as non-ETE and ETE groups based on postoperative pathology. Patients were randomly divided into training (70%) and validation (30%) cohorts, with an external test cohort (n=70) from Huashan Hospital, Fudan University. Clinicopathological and ultrasonographic features were analyzed. Significant variables (P<0.05) identified via univariate logistic regression were incorporated into multivariate models: a clinicopathological model, an ultrasonographic model, and a combined clinicopathological-ultrasonographic model. Performance was evaluated using the area under the curve (AUC), calibration curves, and decision curve analysis (DCA).

The combined model demonstrated robust calibration and discrimination across cohorts, with AUCs of 0.872 (training cohort), 0.835 (validation cohort), and 0.858 (external test cohort). Subgroup analyses revealed slightly lower AUC for microcarcinomas vs. non-microcarcinomas (0.772 vs. 0.893, P<0.05) and superior discrimination of gross ETE vs. minimal ETE (AUC: 0.963 vs. 0.824, P<0.05). No significant differences were observed between Hashimoto's/non-Hashimoto's (AUC: 0.875 vs. 0.843) or euthyroid/dysfunction subgroups (AUC: 0.852 vs. 0.845) (P>0.05).

The clinicopathological-ultrasonographic model provides reliable preoperative ETE prediction, facilitating personalized surgical planning and improved patient outcomes.