Disease-free survival (DFS) as a surrogate for overall survival (OS) in patients (pts) with HR+/HER2− early breast cancer (EBC): A correlation analysis.

Abstract

535 Background: DFS has been widely adopted as a clinically meaningful primary endpoint in EBC trials in place of OS, which needs extensive follow-up to observe sufficient events and fails to assess survival benefit early. However, a rigorous validation of DFS as a surrogate endpoint for OS is required for each given setting. The objective of this analysis was to evaluate DFS as surrogate for OS in the adjuvant treatment of HR+/HER2− EBC. Methods: A systematic literature review (SLR) was conducted using biomedical literature databases (ie, Embase, PubMed, and Medline [from database inception to January 2023]) and key conferences to identify published studies in EBC. The eligibility criteria for the SLR were RCTs that included ≥80% of adult pts with HR+/HER2− EBC and assessing different treatment options (ie, cyclin-dependent kinase 4/6 inhibitors [CDK4/6i], endocrine therapies [ET], and chemotherapies [CT]). Studies were considered for correlation analysis if hazard ratios (HRs) were reported for both DFS and OS. The analysis excluded observational studies and trials with immature OS data. Spearman rank correlation (rs) and weighted linear regression analyses (measured by coefficient of determination [R2]) were performed to evaluate the correlation between HRs for OS and DFS. Considering heterogeneity, multiple scenario analyses were conducted. Surrogate threshold effect (STE) was also estimated; STE is defined as the maximum value of the HR for DFS that needs to be observed in a trial to ensure the possibility of concluding that a significant effect on OS occurred. Correlation was also assessed using pt-level data from the FACE trial (letrozole vs anastrozole in postmenopausal pts with HR+/HER2− EBC) to validate findings of the trial-level analysis. Results: A total of 14 RCTs (N = 31,668) were included (CDK4/6, 2; ET, 7; CT, 5). Trial-level analysis showed that rs between OS and DFS was 0.81 (95% CI, 0.56-0.94), which when weighted by variance was 0.81. Regression analysis showed that 84% of the variability in OS was explained by DFS (R2 = 0.84). A meta-regression model estimated an STE of 0.82 for HR of DFS, which can be leveraged to validate the surrogacy based on the DFS estimated from a future trial. Analysis of pt-level data from the FACE trial demonstrated a correlation between DFS and OS (rs = 0.75). Additionally, correlation coefficients estimated by the Pearson method and iterative imputation were 0.83 and 0.89, respectively. Conclusions: Both trial-level and pt-level analyses demonstrate a positive correlation between DFS and OS. These results support the use of DFS as a reliable surrogate endpoint for OS in HR+/HER2− EBC trials to give an early access to innovative therapies.