Three-year disease-free survival (DFS) as surrogate end-point for predicting five-year overall survival (OS) benefit in adjuvant taxane-based chemotherapy for breast cancer (BC): Analysis of 10 randomized clinical trials (RCTs)

Abstract

584 Background: The issue regarding the eventual correlation between DFS at earlier follow-up (i.e. 3-yrs) with 5-yrs OS has not actually been explored in trials addressing the role of taxanes in BC. All RCTs in which patients were randomized to receive a standard or a taxane-based regimen for early BC were analyzed to evaluate this topic. Methods: All phase III trials with at least 60 month follow-up were considered eligible. The correlation has been explored according to a linear regression model considering both each single outcome pair (DFS/OS) for all arms (extracted by curves), their differences, and each outcome Hazard Ratio (HR) or calculated Relative Risk (RRs), following 2 steps: 1) correlation between 5-yrs DFS and OS (to confirm the evidence); 2) correlation between 3-yrs DFS and 5-yrs OS (predictive role). The correlation was estimated according to Pearson (r) and R2 coefficients (parametric) and Spearman (Rho) coefficient (non- parametric). A model to calculate the target sample size to determine 5-yrs OS benefit of 3%, 5% and 7%, respectively, was calculated as well. Results: Ten RCTs (17,067 patients) with available data for outcomes were gathered. For 5-yrs DFS/OS, a linear correlation was found between rates (r=0.74, R2=0.55; p<0.0001; Rho=0.83; p<0.0001), and HRs (r=0.90, R2=0.81; p<0.0001; Rho=0.91; p<0.0001). Three-yrs DFS correlates with 5-yrs OS, with both rates (r=0.81, R2=0.66; p<0.0001; Rho=0.92; p<0.0001), and RRs (r=0.84, R2=0.71; p=0.002; Rho=0.85; p=0.002). Three-yrs DFS and 5-yrs OS absolute differences strongly correlate (r=0.86, R2=0.74; p=0.001; Rho=0.84; p=0.002). The sample size model (on the basis of the r-coefficient=0.81), calculates 2,733, 863, and 389 pts to improve 3-yrs DFS of 4%, 7% and 10%, which means to improve 5-yrs OS of 3.2%, 5.7% and 8.1%, respectively. Conclusions: By these data, 3-yrs DFS is a reliable surrogate end-point for OS when testing new drugs in early BC, and is able to predict a late survival benefit. Thanks to the smaller patient sample size, RCTs with this design will provide early results in a shorter time period, allowing a faster data transfer to clinical practice. No significant financial relationships to disclose.