Multi-slice computed tomography radiomics combined with serum alpha-L-fucosidase: a potential biomarker for precise identification of pleomorphic adenoma and Warthin tumor.

Background: The rising incidence of parotid gland tumors, with a focus on pleomorphic adenomas (PMA) and Warthin tumors (WT), necessitates accurate preoperative distinction due to their treatment variability and PMA's malignant potential. Traditional imaging, while valuable, has limited accuracy. This study employs multi-slice computed tomography (MSCT) radiomics coupled with serum alpha-L-fucosidase (AFU) levels to develop a diagnostic model aimed at elevating clinical discernment and precision therapy delivery.

Methods: Ninety-one patients were randomly assigned to one of two cohorts: training or validation, at a ratio of 7:3 (64 vs. 27). The region of interest (ROI) on each tumor from the collected MSCT images was delineated to extract radiomics features. In the training cohort, the least absolute shrinkage and selection operator (LASSO) regression and 5-fold cross-validation were adopted to screen the extracted features and calculate Rad-score. Four diagnostic models were developed after univariate and multivariate logistic regression analysis of Rad-score and clinical factors. The performance of four models was then evaluated in the validation cohort by the comparison of receiver operating characteristic (ROC) curve and calibration curve to select the best one. Finally, the clinical application value of the best model was assessed via the nomogram and decision curve analysis (DCA) curve.

Results: Multivariate logistic regression analysis revealed serum AFU, Rad-score and gender as independent diagnostic factors for PMA and WT distinguishment. The nomogram combining the three factors had an area under the curve (AUC) of 0.934 [95% confidence interval (CI): 0.863-1.000] and 0.987 (95% CI: 0.956-1.000) in the training and validation cohorts, respectively, with great goodness-of-fit and clinical value.

Conclusions: The optimized nomogram combining MSCT radiomics and AFU improved the accuracy of distinguishing PMA from WT, suggesting its potential for developing new biomarkers.