Comparison of regorafenib and immunotherapy in colorectal cancer survival: A multicenter cohort study

BACKGROUND: Real-world comparative outcomes of regorafenib versus immune checkpoint inhibitors (ICI) after chemotherapy in colorectal cancer (CRC) are limited. METHODS: Using de-identified multicenter University of California Health EHR data (2012-2025), we identified chemotherapy-exposed CRC patients initiating regorafenib (C/R, n = 653) or ICI (C/I, n = 382) with mutually exclusive exposure groups. Time-to-event analyses were anchored at therapy initiation (index date). Overall survival was compared using Kaplan-Meier and multivariable Cox models. To enhance interpretability and bias control, we also estimated 3- and 5-year restricted mean survival time (RMST) differences and conducted a propensity-score overlap-weighted Cox sensitivity analysis with balance diagnostics. Post-treatment carcinoembryonic antigen (CEA) levels were compared using Wilcoxon testing. RESULTS: Survival differed between groups (log-rank p = 0.0195). Five-year Kaplan-Meier survival was 33.2% (95% CI 29.8-37.0) for C/R and 41.5% (36.7-46.9) for C/I (absolute difference 8.2 percentage points). In adjusted Cox models, C/R was associated with higher mortality risk than C/I (HR 1.217, 95% CI 1.021-1.450; p = 0.0282). RMST favored C/I at 3 years (difference 95.6 days, 95% CI 39.4-148.6) and 5 years (168.0 days, 95% CI 69.1-262.8). Overlap-weighted estimates were consistent (HR for C/I vs C/R 0.802, 95% CI 0.672-0.958; p = 0.0151). Post-treatment CEA was higher in C/R (Hodges-Lehmann difference 8.70; p < 0.0001). CONCLUSION: In this real-world cohort, initiating immunotherapy was associated with improved survival versus regorafenib after chemotherapy, and post-treatment CEA differed between groups, supporting its potential prognostic value.