BACKGROUND: The validity of progression-free survival (PFS) as a surrogate endpoint for overall survival (OS) in metastatic colorectal cancer (mCRC) trials has been studied extensively, but primarily in first-line treatment (for example, Buyse et al. J Clin Oncol. 2007;25:5218-5224 and Saad et al. Ann Oncol. 2010;21:7-12. Epub 2009 Nov 9). We sought to confirm and extend this research by investigating the influence of such factors as line of therapy on the relationship between OS and PFS in mCRC treatment trials.
METHODS: In a systematic literature review, mCRC phase 2 and 3 trials that presented OS and PFS (or time-to-progression) results were eligible. Correlation between these endpoints was estimated by single treatment arm and by study. Treatment effect in each study was defined as the ratio of the median time to event (either OS or PFS) in the two treatment arms. Meta-regression analyses were conducted using least squares metaregression models weighted by study sample size. Statistically significant factors were used to create subgroup analyses.
RESULTS: A total of 66 articles met the initial search criteria and 62 were included in the analysis (total of 23,527 patients). High positive correlation was found between the median PFS and median OS within treatment arms [Pearson coefficient 0.87, (95% confidence interval [CI] 0.82-0.91)] and also between the treatment effects for OS and PFS by study [0.69, (0.53-0.80)]. The regression equation for the relationship between treatment effects was: OS Effect = 0.60 + 0.41*PFS Effect. Thus, a 1-unit increase in PFS treatment effect predicts a 0.41 unit increase in OS treatment effect. Line of therapy was determined to be a significant factor with the R2 higher for first-line (R2 = 0.54) compared to second-line studies (R2 = 0.38).
CONCLUSIONS: Our results demonstrate a strong and consistent linear relationship between treatment effects for PFS and OS. The relationship appears to be stronger in first-line studies but is still evident in second-line studies.
