Validation of vesical imaging reporting and data system for assessing muscle invasion in bladder tumor

To retrospectively determine the diagnostic values of vesical imaging reporting and data system (VI-RADS) score for detecting muscle-invasive bladder tumors.

This study included 297 consecutive patients with 339 tumors who previously diagnosed and subsequently underwent multiparametric MR imaging between January 2015 and March 2019. Two radiologists assessed the scores of muscle-invasive tumors using cutoff values of ≥ 4 and ≥ 3. Cutoff values for VI-RADS scores were estimated from the best operating points of the areas under the receiver operating characteristic curve analyses using the Youden J statistic. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy were calculated to assess the utility of VI-RADS for diagnosing muscle-invasive tumors.

Inter-observer agreement was excellent for three different MR imaging type at lesion level (k = 0.89 for T2W, k = 0.82 for DW, and k = 0.85 for DCE). At a cutoff value of 4, T2W and DW imaging had a diagnostic accuracy of 79.3% (269/339) for tumor lesions with muscle invasion, which was similar to an overall score of 80.2% (272/339). The overall VI-RAD score showed 80.2% accuracy (272/339), with a cutoff value of ≥ 4, yielding 91.3% sensitivity (85/93), 76.0% specificity (187/246), 83.3% PPV (85/102), and 78.9% NPV (187/237). When we considered an arbitrary overall score of ≥ 3 as the cutoff value, the accuracy was 63.7% (216/339); sensitivity, 94.6% (125/132); specificity, 43.9% (91/207); PPV, 51.6% (125/242); and NPV, 63.7% (91/97).

VI-RADS has an overall good performance in the diagnosis of muscle-invasive tumors.