Diagnostic prediction model using VI-RADS and pathological grade to predict muscle invasion in bladder cancer: a retrospective study

The efficacy of bladder cancer treatment depends strongly on an accurate distinction between muscle-invasive and non-muscle-invasive disease. While the Vesical Imaging-Reporting and Data System (VI-RADS) provides a structured MRI-based assessment of detrusor muscle invasion, equivocal categories and interobserver variation may limit stand-alone decision-making. This study developed a diagnostic prediction model combining VI-RADS with TURBT-derived pathological grade for predicting muscle invasion in patients with bladder cancer. This retrospective study analyzed multiparametric magnetic resonance imaging (mpMRI) findings and clinical data from 152 patients with histologically confirmed bladder cancer treated between January 2018 and April 2024. Patients were classified as having muscle-invasive bladder cancer (MIBC) or non-muscle-invasive bladder cancer (NMIBC). Independent predictors of muscle invasion were identified using univariate and multivariate binary logistic regression analyses. Predictive performance and clinical utility were assessed using receiver operating characteristic (ROC) curves and decision curve analysis (DCA). The study was designed and reported in accordance with the TRIPOD guidelines for the development of prediction models. Of the 152 patients, 58 had MIBC and 94 had NMIBC. Multivariate logistic regression demonstrated that both pathological grade and VI-RADS score were independent predictors of muscle invasion. ROC curves identified a VI-RADS score ≥ 3 as the optimal threshold for predicting MIBC. The combination of pathological grade and VI-RADS score significantly improved diagnostic accuracy compared with either variable alone ( p < 0.05) and provided greater net clinical benefit on DCA. The integration of pathological grade with VI-RADS scoring improves the preoperative assessment of muscle invasion in bladder cancer, enabling more accurate risk stratification and supporting informed clinical decision-making.