Background: The gold standard end point in randomized clinical trials in metastatic breast cancer (MBC) is overall survival (OS). Although therapeutics have been approved based on progression-free survival (PFS), its use as a primary end point is controversial. We aimed to assess to what extent PFS may be used as a surrogate for OS in randomized trials of anti-HER2 agents in HER2+ MBC.
Methods: Eligible trials accrued HER2+ MBC patients in 1992-2008. A correlation approach was used: at the individual level, to estimate the association between investigator-assessed PFS and OS using a bivariate model and at the trial level, to estimate the association between treatment effects on PFS and OS. Correlation values close to 1.0 would indicate strong surrogacy.
Results: We identified 2545 eligible patients in 13 randomized trials testing trastuzumab or lapatinib. We collected individual patient data from 1963 patients and retained 1839 patients from 9 trials for analysis (7 first-line trials). During follow-up, 1072 deaths and 1462 progression or deaths occurred. The median survival time was 22 months [95% confidence interval (CI) 21-23 months] and the median PFS was 5.7 months (95% CI 5.5-6.1 months). At the individual level, the Spearman correlation was equal to ρ = 0.67 (95% CI 0.66-0.67) corresponding to a squared correlation value of 0.45. At the trial level, the squared correlation between treatment effects (log hazard ratios) on PFS and OS was provided by R(2) = 0.51 (95% CI 0.22-0.81).
Conclusions: In trials of HER2-targeted agents in HER2+ MBC, PFS moderately correlates with OS at the individual level and treatment effects on PFS correlate moderately with those on overall mortality, providing only modest support for considering PFS as a surrogate. PFS does not completely substitute for OS in this setting.
Keywords: HER2; lapatinib; metastatic breast cancer; progression-free survival; surrogate; trastuzumab.
© The Author 2016. Published by Oxford University Press on behalf of the European Society for Medical Oncology. All rights reserved. For permissions, please email: journals.permissions@oup.com.