42: Recurrence-free Survival versus Overall Survival as a Primary End Point for Resectable Colorectal Liver Metastasis Studies: Analysis of 2,983 Curative-intent Hepatectomies

Introduction: Recurrence-free survival (RFS) has been used as a surrogate endpoint for overall survival (OS) in trials involving resectable colorectal liver metastasis (CRLM). However, the correlation between RFS and OS for this clinical context has not been previously examined.


Methods: Consecutive patients who underwent a complete resection of CRLM (1991-2019) were included from a single-center prospectively maintained database.  OS and RFS probabilities at various timepoints were estimated using the Kaplan-Meier method, and the pairwise associations were assessed by Spearman's rank correlation.  Simulated randomized trials with homogenous arms were performed to evaluate the within-trial concordance of RFS and OS.


Results: A total of 2,983 patients underwent complete resection of CRLM, with a median follow-up of 8.4 years.  There were 1,995 (67%) disease recurrences and 1,684 (56%) deaths.  Median RFS was 1.3 years (95% CI 1.3-1.4) and median OS was 5.2 years (95% CI 5.0-5.5).  A total of 1,428 deaths (85%) were preceded by recurrence, and median time from recurrence to death was 2.0 years (range 0-23 years).  Pairwise correlations between RFS and OS were weak to moderate, with a correlation estimate ranging from 0.30-0.56, and maximized for the pair of 5-year RFS and 7-year OS (0.56, SD 0.13).  Of the simulated randomized trials where a statistical difference in OS between trial arms was detected, only 29% also had a statistically different RFS between trial arms.


Conclusions: Following resection of CRLM, there is minimal correlation between RFS and OS.  RFS is an inadequate surrogate for OS.