Association between combination of VI-RADS based on T2WI and morphological features and pathological outcome and prognosis in bladder cancer.

Abstract

690 Background: VI-RADS is widely used to identify muscle-invasive bladder cancer (MIBC). However, it is not known whether structural category (SC; VI-RADS based on T2-weighted images) alone is useful in diagnosing clinicopathological features and prognosis of bladder cancer. This study investigates the diagnostic performance of SC in the detection of MIBC. We also measured tumour contact length (TCL), difference in signal intensity ratio (dSIR) and coefficient of variation (CV) within the tumour on T2WI. We evaluated whether these factors could improve the diagnostic performance of SC for MIBC and explored the correlation between these morphological features and clinicopathological characteristics and prognosis. Methods: Between August 2018 and July 2023, we performed 587 transurethral resections of bladder tumours. This study includes 236 patients who underwent preoperative MRI. We first evaluated the diagnostic performance of VI-RADS and SC for MIBC. TCL, dSIR and CV were measured on T2WI. The dSIR was defined as the difference between the signal intensity (SI) of the tumour and that of the underlying muscle layer, divided by the SI of the normal muscle layer. The CV was calculated as the standard deviation of the SI within the tumour divided by the mean SI. All patients were divided into high and low SC groups and the relationship between MIBC and TCL, dSIR and CV was examined. Cut-off values for the diagnosis of MIBC and overall survival (OS) were calculated based on ROC curves for the most correlated factors in each group, which were added to the SC score to evaluate the diagnostic value. We also examined the correlation between each factor and pathological findings. Results: The diagnostic performance of VI-RADS with a cut-off score of 4 was an AUC of 0.888. Meanwhile, the diagnostic performance of SC with a cut-off score of 4 was AUC of 0.885. The association between MIBC and each factor was compared in the SC≥4 and SC≤3 groups. In the SC≥4, multivariate analysis showed significant differences in dSIR (p<0.001). In the SC≤3, multivariate analysis showed TCL is significantly longer in MIBC (p<0.001). The diagnostic performance was an AUC of 0.912 for suspected MIBC with SC≥4 and dSIR<1.03 and SC≤3 and TCL≥36mm. In pathological features, TCL is significantly longer in high-grade (p=0.018), necrosis(p<0.001), variant(p=0.002), lymph node metastasis (p<0.001) and organ metastasis (p<0.035). Univariate COX analysis showed that SC≥4 was risk factors for OS (HR = 2.15, P = 0.031). In the SC≥4, patients with TCL≥40 mm had worse OS (HR = 3.06, P < 0.015). In the SC≤3, patients with TCL≥31 mm had worse OS (HR = 10.59, P < 0.001). Conclusions: The present study showed that SC combined with dSIR and TCL has good diagnostic performance for MIBC. We also found that TCL correlates with high-grade tumor, necrosis, and variant histology. The combination of SC and TCL effectively predicts OS in bladder cancer.