Accuracy of the Vesical Imaging-Reporting and Data System (VIRADS) for pre-treatment staging of bladder cancer in an Australian cohort.

Introduction: To evaluate the performance of the Vesical Imaging-Reporting and Data System (VIRADS) in differentiating muscle-invasive and non-muscle-invasive bladder cancer and whether this reporting system improves inter-reader agreement.

Methods: Sixty-four cases of multiparametric 3 tesla bladder MRI from January 2014 to May 2020 were reviewed retrospectively. T2-weighted, diffusion and post-contrast images were reviewed. All magnetic resonance images were reported by a radiologist with 15 years' experience (Reader 1) and a final year radiology trainee with a special interest in urogenital imaging with 3 years of experience (Reader 2). Both readers were blinded to clinical history and histopathology results when scoring each lesion.

Results: The sensitivity and specificity for differentiating MIBC and NMIBC were 91% and 68%, respectively, for Reader 1 and 91% and 63%, respectively, for Reader 2. The inter-reader agreement for assigning VIRADS scores was 0.79. The area under the receiver operator curve for Reader 1 and 2 were not significantly different (Reader 1 = 0.79, Reader 2 = 0.77, P = 0.83).

Conclusions: Staging of bladder cancer prior to treatment can be accurately and reliably diagnosed using VIRADS, a novel, standardised reporting system for bladder MRI.