Abstract Background Several definitions of efficacy endpoints have been proposed for clinical trials in early breast cancer (EBC). Invasive disease-free survival (IDFS) and other endpoints have been used as surrogates for overall survival (OS) in adjuvant trials. The aim was to evaluate these endpoints in a comparative way in a large cohort of patients to assess whether they are appropriate surrogates for OS. Methods Individual patient data from 8 phase III randomized adjuvant chemotherapy (CT) trials for EBC were analyzed. Trial-level surrogacy analyses were conducted using weighted least squares regression to assess the association between treatment effects on surrogates at different time points—IDFS and other defined endpoints (by STEEP v2.0)—and OS. Correlation between the log hazard ratio (HR) of the endpoints and OS were quantified using R2 values. Preplanned subgroup analyses by tumor subtype were also performed. Results Results censoring at 5 years and at longer follow-up (median: 9.6 years) were similar. HR of IDFS and invasive BC-free survival (IBCFS) at 5 years showed a R2 0.7 relative to HR of OS. All other surrogates revealed correlations with R2 0.5. For patients with hormone receptor-positive/HER2-negative intermediate/high-risk tumors, correlation showed higher R2 for IDFS (R2=0.89; 95%CI = 0.80-0.98) and IBCFS (R2=0.84; 95%CI = 0.71-0.97) than distant-recurrence endpoints, while for those with triple negative tumors the highest correlation was seen with distant disease-free survival (R2=0.86; 95%CI = 0.75-0.98). Conclusions Although all studied endpoints are appropriate surrogates for OS when evaluating CT-based adjuvant interventions, investigators should consider the BC subtype to select the primary endpoint.
Blasco et al. (Wed,) studied this question.