Exploring arterial spin labeling for bladder cancer: noninvasive assessment of tumor aggressiveness and muscular invasion

Pathological grading and muscle invasion status are critical determinants of treatment strategy and prognosis in bladder cancer. Noninvasive, contrast-free multiparametric imaging approaches capable of preoperatively characterizing tumor biology are clinically needed. This prospective single-center study included 68 patients with pathologically confirmed bladder urothelial carcinoma, including 22 MIBC and 46 NMIBC cases. All patients underwent preoperative RESOLVE DWI and 3D-pCASL MRI. A multivariable logistic regression model integrating apparent diffusion coefficient (ADC) and blood flow (BF) was constructed for tumor grade prediction, and internal validation was performed using 2,000 bootstrap resamples. High-grade tumors showed higher BF and lower ADC values than low-grade tumors (BF: 94.2 ± 38.4 vs. 38.9 ± 14.6 mL/min/100 g; ADC: 0.954 [0.803–1.047] vs. 1.206 [1.097–1.295] × 10 ⁻³ mm²/s; both P < 0.0001). BF was significantly higher in MIBC than in NMIBC (83.90 [51.75–113.75] vs. 50.00 [34.25–68.50] mL/min/100 g, P = 0.0059), whereas ADC showed no significant difference between groups ( P = 0.6695). The ADC–BF model achieved an AUC of 0.980 (95% CI: 0.954–1.000), with 91.7% sensitivity and 96.9% specificity. Its performance was superior to ADC alone (DeLong test, P = 0.0004) but not significantly different from BF alone ( P = 0.0597). Bootstrap validation yielded an optimism-corrected AUC of 0.976. BF alone achieved AUCs of 0.931 for tumor grading and 0.736 for muscular invasion prediction. ADC was negatively correlated with Ki-67, whereas BF was positively correlated with CD34 expression. RESOLVE DWI and 3D-pCASL MRI provide complementary noninvasive biomarkers reflecting tumor cellularity and perfusion, supporting preoperative assessment of bladder cancer grade and muscle invasion. Further validation in larger multicenter cohorts is required to confirm the generalizability of these findings.