A prospective cohort study to assess the diagnostic accuracy of VI-RADS score in a tertiary center in central India

Abstract

Purpose:

Accurate preoperative differentiation between non–muscle-invasive bladder cancer (NMIBC) and muscle-invasive bladder cancer (MIBC) is critical for selecting appropriate therapy. We prospectively evaluated the diagnostic performance of multiparametric MRI (mpMRI) using the Vesical Imaging–Reporting and Data System (VI-RADS) for predicting pathological muscle invasion in patients with suspected bladder tumors.

Methods:

In this prospective cohort study, 102 consecutive adults with suspected bladder mass were enrolled; 53 patients formed the final analytic cohort (24 MIBC, 29 NMIBC). mpMRI was performed on a 3T scanner; the highest VI-RADS score was used for patients with multiple lesions. Two blinded radiologists assigned VI-RADS scores. The reference standard was histopathology obtained after Transurethral resection of bladder tumor. Diagnostic performance metrics were calculated with 95% confidence intervals, and area under curve (AUC) analysis was performed.

Results:

All mpMRI scans were interpretable, with no serious adverse events. Muscle invasion was observed in 37.5% of VI-RADS 4 lesions and 94.7% of VI-RADS 5 lesions. At VI-RADS ⩾ 3, sensitivity was 100% and specificity 37.9% (accuracy 66.0%; AUC 0.69). At VI-RADS ⩾ 4, sensitivity remained 100% with improved specificity 62.1% (accuracy 79.2%; AUC 0.81). Using VI-RADS 5 as positive yielded specificity 96.6% with sensitivity 75% (accuracy 86.8%; AUC 0.86).

Conclusion:

VI-RADS on mpMRI demonstrated excellent sensitivity for ruling out MIBC at lower thresholds, while higher thresholds improved specificity for ruling in muscle invasion. These findings support VI-RADS-guided preoperative risk stratification, with larger multicenter validation warranted.