Diagnostic Accuracy and Interobserver Agreement for the Vesical Imaging-Reporting and Data System for Muscle-invasive Bladder Cancer: A Multireader Validation Study

Abstract

A Vesical Imaging-Reporting and Data System (VI-RADS) based on multiparametric magnetic resonance imaging (mp-MRI) was developed to provide accurate information for the diagnosis of muscle-invasive bladder cancers (MIBCs). In this study we evaluated the interobserver agreement and diagnostic performance of VI-RADS. We retrospectively evaluated data for 74 consecutive patients with bladder cancer who had undergone mp-MRI before transurethral resection in a single institution from January 2010 to August 2018. Five readers assessed the probability of the presence of MIBC using VI-RADS scores. The interobserver agreement was assessed by measuring intraclass correlation coefficients (ICCs). Receiver operating characteristic (ROC) analysis was used to evaluate the MIBC diagnostic performance. The area under the ROC curve (AUC) was estimated for ordinal score assessments. Our study demonstrated that interobserver agreement was excellent among five readers (ICC 0.85, 95% confidence interval 0.80-0.89) and the diagnostic performance of VI-RADS was represented as a pooled AUC of 0.90 (95% confidence interval 0.87-0.93). PATIENT SUMMARY: We evaluated the interobserver agreement and diagnostic performance of a new scoring system based on magnetic resonance imaging called a Vesical Imaging-Reporting and Data System (VI-RADS) for muscle-invasive bladder cancer. Our results indicate that VI-RADS is suitable as a comprehensive tool for appropriate treatment planning for patients with bladder cancer.

Keywords: Bladder cancer; Multiparametric magnetic resonance imaging; Scoring; Staging; Vesical Imaging-Reporting and Data System.