Real-world endpoints as surrogates for overall survival in metastatic bladder cancer.

Abstract

693 Background: Overall survival (OS) remains the gold-standard endpoint in solid tumors, including metastatic bladder cancer (mBC), but requires extensive follow-up and resources. Real-world endpoints (RWE) such as real-world progression-free survival (rwPFS), time to next therapy (TTNT), time to end of first-line therapy (TTEFL), second-line rwPFS (rwPFS2), and treatment-free survival (TFS) may accelerate evidence generation if validated as OS surrogates. Methods: We analyzed 5,296 patients with mBC from who received first-line chemotherapy (n = 3,831) or single-agent immune checkpoint inhibitors (ICI, n = 1,465) using the Flatiron Health database. Patient-level surrogacy was assessed using Kendall’s tau. For group-level surrogacy, patients were stratified into 20 risk-based sub-cohorts via Cox regression–derived risk scores. The R² from linear regression quantified the association between 12-month OS rates and 3-month event-free rates for rwPFS, TTNT, and TTEFL across sub-cohorts. TFS was evaluated as the restricted mean survival time between TTEFL and TTNT over 12 months. Sensitivity analyses tested alternative timepoints and cohort sizes. Results: Median OS was 14.5 months (18-month OS, 43%) for chemotherapy and 8.9 months (18-month OS, 34%) for ICI. At patient level, rwPFS (τ = 0.67 chemotherapy; 0.71 ICI), TTNT (0.58; 0.78), and rwPFS2 (0.94; 0.96) showed robust correlations with OS, whereas TTEFL was weaker (0.30, 0.59). At group level, 3-month rwPFS (R² = 0.88) and TTNT (0.85) were strongly correlated with 12-month OS in chemotherapy, while in ICI, TTEFL (0.80) and TTNT (0.78) outperformed rwPFS (0.59) at 3 months. Extending rwPFS to 6 months against 18-month OS improved its correlation in ICI (0.70), reflecting delayed immunotherapy benefit. TFS paralleled OS in chemotherapy (R² = 0.85) but was less informative for ICI (0.03). Sensitivity analyses confirmed robustness across varied time horizons and cohort sizes. Conclusions: This study suggests RWEs as potential surrogates for OS in mBC, with optimal endpoints varying by treatment type. For chemotherapy, rwPFS and TTNT serve as reliable proxies. In ICI-treated patients, TTNT and TTEFL provide robust early signals, with rwPFS gaining predictive strength over longer follow-ups. These insights pave the way for faster, more effective evaluations of treatment outcomes in mBC, ultimately enhancing clinical and regulatory decision-making.