Surrogacy of progression-free survival (PFS) for overall survival (OS) in rectal cancer trials with preoperative therapy: Literature-based meta-analysis.

Abstract

734 Background: Disease-free survival (DFS) after adjuvant chemotherapy in colon cancer has been validated as a surrogate endpoint for OS. An analysis of individual patient data from two randomized controlled trials evaluating preoperative chemoradiotherapy for rectal cancer has already shown that PFS might be a surrogate endpoint for OS (Eur J Cancer, 2012). However, since the report was derived from only two trials, the further evaluation of surrogacy with a larger number of trials is desirable. This study was conducted to explore the trial level correlations between PFS and OS in resectable rectal cancer with preoperative therapy. Methods: A systematic literature search of randomized trials of preoperative chemoradiotherapy, radiotherapy or chemotherapy for rectal cancer reported from January 1990 to September 2014 was conducted using PubMed and the Cochrane library. Weighted linear regression using sample size of each trial as a weight was used to estimate coefficient of determination (R2) within PFS and OS. 95% confidence interval of R2 was estimated by bootstrap method. The primary analysis included trials in which the HR of PFS and OS were reported. A sensitivity analysis included the trials in which either HR or 5 years point estimates of PFS and OS was reported. In the sensitivity analysis, HR was estimated from the point estimates of PFS and OS assuming exponential distribution. Results: Of 1,048 articles, 14 trials (7,233 patients) were selected for the primary analysis and 18 (8,020) for the sensitivity analysis. Median numbers of enrolled patients were 332 (range 113-1,805) in primary analysis, and 296 (113-1,805) in sensitivity analysis. The primary analysis did not show the correlation between treatment effects on PFS and OS(R2 0.408; 95% CI [0.133-0.845]). The sensitivity analysis showed modest correlation between PFS and OS (R2 0.603; 95% CI [0.296-0.871]). Conclusions: The primary analysis showed that PFS was not suitable as a surrogate endpoint, which is contradictory to the previous report. Based on our results, OS is still preferable as the primary endpoint in the trials of preoperative therapy for resectable rectal cancer.