OBJECTIVES : OS is an endpoint for measuring lifetime clinical benefit of oncology therapies; however, studies in the adjuvant setting may not detect OS benefit owing to the time required to observe sufficient events. DFS is commonly measured in primary (neo)adjuvant BC trials when OS may be prolonged. We used linear regression analysis to assess DFS as a surrogate for OS in HER2– primary BC.

METHODS : MEDLINE, Embase, CENTRAL, CDSR, DARE, major oncology congresses and clinical trial databases were searched in August 2019 to identify studies reporting OS and DFS in a HER2– primary BC population. Studies were included if they were conducted in North America and/or Europe, if hazard ratios (HRs) were reported for both outcomes, and if patients had completed local treatment and (neo)adjuvant chemotherapy. Spearman’s rank correlation and linear regression analyses were performed to evaluate the correlation between HRs for OS and DFS.

RESULTS : Among 2,596 records reviewed, eight studies (six RCTs, two comparative retrospective studies) were included. Studies comprised triple-negative BC populations (n=2), hormone receptor positive HER2– BC populations (n=2), and overall HER2– BC populations (n=4). The Spearman’s rank correlation coefficient between OS and DFS was 0.756 (p=0.030), which when weighted by study sample size was 0.803 (p=0.016). The regression coefficient on natural logarithm (ln) HR_OS versus ln HR_DFS was 1.75 (95% confidence interval: 1.49–2.02; R² = 0.972), suggesting a 10% relative risk reduction (RRR) for DFS was associated with an additional 7% RRR for OS.

CONCLUSIONS : A nominally significant positive correlation was found between OS and DFS in HER2– primary BC, indicating that DFS can be used as a valid surrogate for OS, enabling earlier assessment of clinical benefit. Limitations include the number of studies. Validation of this correlation including future studies is required.