Development and validation of a clinical prediction model to diagnose Warthin tumor based on non-contrast computed...

This study aims to develop and validate a clinical prediction model that integrates clinical features with non-contrast CT imaging characteristics to identify Warthin tumor (WT) in the parotid gland.

A total of 289 patients who underwent surgical resection of parotid tumors at the Affiliated hospital of Jiangnan University from June 2018 to December 2024 were consecutively and randomly divided into training (n = 202) and validation (n = 87) cohorts at a 7:3 ratio. Demographic and non-contrast CT imaging variables were collected. Logistic regression identified predictors, and a nomogram was constructed. To further validate the model, an independent dataset comprising 84 patients from a second hospital was used. The model's performance was evaluated through receiver operating characteristic (ROC) curves, calibration curves, the Hosmer-Lemeshow test, and decision curve analysis (DCA).

Age, smoking history, tumor distribution, earlobe position, and longitudinal-to-transverse ratio (LTR) were identified as independent predictors for differentiating WT from other parotid gland tumors. The nomogram showed high diagnostic accuracy, with the area under the curve (AUC) values of 0.942 (training), 0.937 (validation), and 0.953 (external validation). Calibration curves indicated good agreement with ideal predictions, supported by the Hosmer-Lemeshow test (P > 0.05). DCA further demonstrated the superior clinical utility of the nomogram model.

The nomogram model incorporating clinical and non-contrast CT features demonstrates high accuracy for differentiating WT from other parotid gland tumors in clinical practice.