ABSTRACT
PURPOSE: Level 1 evidence supports the use of neoadjuvant chemotherapy (NAC) for the treatment of muscle-invasive bladder cancer (MIBC), but observational data demonstrate that this approach is underused. A barrier to shared decision making is difficulty in predicting and communicating survival estimates after cystectomy with or without NAC. METHODS: We included patients with MIBC from the National Cancer Database treated with cystectomy. A state-transition model was constructed for calculating 5-year death risk using baseline patient-, tumor-, and facility-level variables. Internal-external cross-validation by geographic region was performed. The effect of NAC was integrated using a literature-derived hazard ratio. Bladder cancer-specific and other-cause mortality was estimated from all-cause mortality rates from US life tables. From the state-transition model, a Web-based tool was developed and pilot usability testing performed. RESULTS: A total of 9,824 patients with MIBC who underwent cystectomy were eligible for inclusion. Median overall survival was 39.6 months (95% CI, 37.4 to 42.4 months). Increasing age, higher clinical T stage, higher comorbidity index, and black race were associated with shorter survival. Private insurance, higher income, and cystectomy at a high-volume facility were associated with longer survival. The prediction model was well calibrated across geographic regions, with observed-to-predicted 5-year death risks ranging from 0.85 to 1.17. Absolute risk reductions with NAC varied from 8.6% to 10.1%. The Web-based tool allowed input of the predictor variables and a user-defined hazard ratio associated with the effect of NAC to generate individualized survival estimates. The tool demonstrated good usability with clinicians. CONCLUSION: A Web-based tool was developed to individualize outcome prediction and communication in patients with MIBC treated with cystectomy with or without NAC to facilitate shared decision making.
