BACKGROUND: Treatment (tx) in RCTs might yield different outcomes than in clinical practice due to protocol-mandated intervention and follow up. We compared overall survival (OS) with 1st-line (1L) FOLFOX/XELOX + bevacizumab (BV) between mCRC pts in an RCT and pts in an OCS with matching eligibility.
METHODS: Data sources were: 1) NO16966, an international RCT (2004 − 05 accrual; 2008 follow up) of FOLFOX/XELOX plus BV vs placebo for 1L mCRC; 2) ARIES, a US-based OCS of mCRC pts receiving 1L chemotherapy plus BV (2006–08 accrual; 2012 follow up). Analyses were restricted to pts who: 1) retrospectively met NO16966 eligibility criteria; 2) received FOLFOX/XELOX + BV; and 3) had no prior oxaliplatin. Cox proportional hazards models of OS were used to compare cohorts; adjusted analyses accounted for baseline (BL) and time-varying covariates including time on 1L tx, disease progression (PD), and use of BV beyond PD (BBP). Due to geographic differences in the cohorts, a sensitivity analysis compared all ARIES pts to NO16966 pts from US/Canada.
RESULTS: Mean 1L tx duration was 242 days in ARIES (n = 755) vs 193 days in NO16966 (n = 624). Unadjusted analysis showed lower OS (ie, hazard ratio [HR] >1) for NO16966 pts. However, after adjusting for imbalances in BL differences, PD, and tx to and beyond PD, OS appeared comparable between cohorts (table). ECOG status of 0 (HR 0.79; 95% CI 0.69–0.89), history of primary tumor resection (HR 0.67; 95% CI 0.57–0.78), longer 1L tx duration (HR 0.95; 95% CI 0.94–0.97), and BBP (HR 0.77; 95% CI 0.65–0.91) were associated with longer OS, but >1 metastatic site (HR 1.29; 95% CI 1.13–1.47) and incidence of PD (HR 7.52; 95% CI 6.08–9.29) were associated with reduced OS. Sensitivity analyses of US/Canada pts showed similar results.
CONCLUSIONS: When comparing similar pts and adjusting for differences in tx patterns, OS appeared to be comparable with 1L FOLFOX/XELOX + BV in the NO16966 RCT vs the ARIES OCS.
