Biparametric Versus Multiparametric MRI for VI-RADS Assessment: Reproducibility Relative to Routine mpMRI Reporting and Impact of Radiologist Experience in a Single-Center Study

<b>Background</b>: We tested whether a contrast-free protocol can reproduce contrast-enhanced VI-RADS scoring and whether reader expertise influences results. <b>Methods</b>: In this retrospective single-center study (January-December 2024), 65 patients (69 lesions) underwent bladder multiparametric MRI. Two blinded radiologists assigned VI-RADS scores using only T2-weighted and diffusion-weighted imaging (biparametric, non-contrast MRI): an expert (>15 years in urogenital radiology) in genitourinary MRI and a non-expert (5 years of experience in genitorurinary radiology). Two complementary reference standards were used. For reproducibility analysis, the reference standard was the VI-RADS score from the original clinical report based on the full multiparametric examination including contrast-enhanced imaging. For diagnostic accuracy analysis, histopathology was used as the reference standard for muscle-invasive versus non-muscle-invasive disease. Agreement was evaluated with confusion matrices, overall agreement, and weighted Cohen's kappa. Discrimination for high likelihood of muscle invasion (VI-RADS ≥ 4) was assessed with receiver operating characteristic analysis. <b>Results</b>: Reference scores were VI-RADS 2 (34.8%), 3 (14.5%), 4 (20.3%), and 5 (30.4%). Agreement was higher for the expert than the non-expert (73.9% vs. 56.5%; weighted kappa 0.74 [95% confidence interval 0.56-0.89] vs. 0.58 [0.37-0.75]). The area under the curve for VI-RADS ≥ 4 was 0.87 (0.78-0.95) for the expert and 0.81 (0.69-0.91) for the non-expert. Sensitivity at a biparametric threshold of VI-RADS ≥ 4 was 88.6% for both readers; specificity was 85.3% vs. 73.5%. Post-resection cases showed more discrepancies, mainly overstaging. <b>Conclusions</b>: Contrast-free biparametric MRI may approximate multiparametric VI-RADS scoring only in treatment-naïve pre-TURBT cases with clearly low-risk, non-equivocal imaging features, but performance is reader-dependent and less reliable in equivocal, higher-risk, and post-resection examinations. Contrast-enhanced multiparametric MRI remains preferred for staging.