Surrogate endpoints for overall survival in randomized controlled trials evaluating adjuvant treatment for breast cancer: A meta-analysis

Abstract

Background: In cancer randomized controlled trials (RCT), endpoints other than overall survival (OS) such as disease-free survival (DFS) are increasingly being used as primary endpoint. Their development is influenced by the need to reduce the number of patients, the duration and ultimately the cost of the trials. Their use as primary endpoint however require a rigorous validation as surrogate endpoints for OS. In adjuvant breast cancer, only a few studies evaluated surrogate endpoints for OS, limited by the use of aggregate data. We present the results of a meta-analysis of individual-patient data assessing surrogate endpoints for OS in the context of adjuvant breast cancer. Methods: We evaluated four endpoints as potential surrogates for OS relying on a meta-analysis of 5 phase III trials: relapse-free survival (RFS), invasive DFS, locoregional RFS and distant DFS. At the patient level, we estimated the individual-level associations by jointly modeling each surrogate with OS using a copula function. At the trial level, we estimated the association between the treatment effects using (1) a linear regression model weighted by the trial size and (2) the two-step model proposed by Burzykowski, Molenbergh and Buyse. Results: We gathered individual-patient data from 11676 patients from 5 RCT. The endpoints were highly associated with OS at the patient level (r>=0.98). At the trial level, invasive DFS showed the higher association with OS (R 2 WLR =0,78; R2 2SM=0,87). Conclusions: This is the first meta-analysis on individual-patient data evaluating surrogate endpoints for OS in adjuvant breast cancer. These results suggest that the endpoints could be interesting candidate surrogates for OS, but further evaluation on a larger set of trials is required to improve the precision of our estimations of the triallevel associations.