Accurate assessment of muscle invasion is critical in determining appropriate management for bladder cancer. Under-staging remains a well-recognised limitation of transurethral resection of bladder tumour (TURBT), even when detrusor muscle is included in the specimen. Current guidelines support consideration of a second TURBT to address this issue. Multiparametric MRI (mpMRI) has emerged as a promising, non-invasive tool to enhance local staging accuracy. The Vesical Imaging Reporting and Data System (VI-RADS) offers a structured approach to determine the likelihood of muscle invasion based on T2-weighted, diffusion-weighted (DWI), and dynamic contrast-enhanced (DCE) imaging. Diffusion kurtosis imaging (DKI), an advanced extension of DWI, captures non-Gaussian water diffusion and may improve prediction of tumour complexity and grade. This study evaluates the diagnostic performance of VI-RADS and DKI in pre-resection staging of bladder cancer.

We evaluated the first 100 patients enrolled in a prospective, multi-centre study with suspicious bladder lesions on ultrasound or CT. All participants underwent pre-TURBT multiparametric MRI (mpMRI) using a standardised protocol incorporating T2-weighted, DWI and DCE imaging. VI-RADS scores were independently assigned by two expert radiologists. Diagnostic performance at VI-RADS cutoffs of ≥3 and ≥4 was assessed through sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). Diffusion kurtosis imaging (DKI) parameters were derived from DWI sequences using defined volumes of interest to calculate mean kurtosis, which was then correlated with tumour grade.

Muscle-invasive bladder cancer (MIBC) was identified in 23% of patients, while 69% had non-muscle-invasive disease. The remaining 8% were diagnosed with non-bladder cancer conditions. High-grade tumours were present in 67% of cases, and 72% had pure urothelial carcinoma without variant histology. Using a VI-RADS cutoff of ≥4, the sensitivity, specificity, PPV, and NPV for detecting MIBC were 78.3%, 85.7%, 64.3%, and 92.2%, respectively. At a cutoff of ≥3, sensitivity increased to 91.3%, specificity decreased to 54.5%, PPV was 37.5%, and NPV was 95.5%. Inter-reader agreement was strong (Cohen’s kappa = 0.81). The probability of MIBC increased progressively with higher VI-RADS scores, from 0% at a score of 1 to 90.9% at a score of 5. Diffusion kurtosis imaging (DKI) demonstrated a moderate correlation with tumour grade (r = 0.44, p = 0.058).

Pre-cystoscopic mpMRI using VI-RADS offers strong diagnostic accuracy for identifying MIBC. A low VI-RADS score can reassure clinicians when muscle invasion is not seen on TURBT. Diffusion kurtosis imaging shows promise as an adjunct tool for tumour grading and may support more accurate preoperative risk stratification.