Evaluation of the Cancer of Bladder Risk Assessment (COBRA) Score in the Cancer Genome Atlas (TCGA) Bladder Cancer Cohort

Objective: To evaluate the Cancer of the Bladder Risk Assessment (COBRA) score in The Cancer Genome Atlas (TCGA) bladder cancer cohort. Second, to investigate the utility of the COBRA score within each bladder cancer molecular subtype following radical cystectomy (RC) and determine if it can help identify candidates for adjuvant therapies and clinical trials. Methods: Among the TCGA bladder cancer cohort (n = 412), RC pathology reports were reviewed to calculate COBRA scores. Kaplan-Meier survival curves along with univariable and multivariable Cox proportional hazard models were used to determine the clinical utility of the COBRA score to predict overall survival (OS) within the overall cohort and within each molecular subtype (if n>30 within subtype). Results: In the analytic cohort (n = 273) there was a median follow-up of 18 months. Higher COBRA score was associated with significant increased risk of death in both univariable (HR = 1.52 per point [PP] 95% CI [1.32, 1.75)] and multivariable models (HR = 1.54 PP 95% CI [1.32, 1.79]). This remained true in multivariable models stratified by molecular subtype for basal (HR = 1.37 PP 95% CI [1.07, 1.74]), luminal infiltrated (HR = 1.70 PP 95% CI [1.10, 2.64]), and luminal papillary (HR = 1.62 PP 95% CI [1.28, 2.06]) tumors. Conclusion: Our findings validate the COBRA score in the TCGA bladder cancer cohort. This suggests the COBRA score can be used in conjunction with molecular subtyping information to help guide clinical decision-making following RC to improve risk stratification and allow for earlier identification of candidates for adjuvant therapies and clinical trials.