Our aim was to review the studies which assessed potential surrogate endpoints for overall survival (OS) in breast cancer trials. A Literature search in PubMed database of studies which assessed potential surrogate endpoints for OS in breast cancer trials was conducted. The surrogacy was assessed with the German institute of Quality and efficiency in Health Care's (IWQiG) framework and the Fleming hierarchy. Thirteen studies were identified. At the neoadjuvant setting, two individual patient data (IPD) meta-analyses and one aggregate data meta-analysis assessing surrogacy of pathological complete response (PCR) were identified. Trial-level association was calculated in one study and the squared correlation was 0.24. Therefore PCR was not judged to be valid surrogate for OS at the neoadjuvant setting according to the IWQiG framework and Fleming hierarchy. At the adjuvant setting, one meta-analysis on aggregate data was identified. 2-year DFS was not judged to be valid surrogate for OS at the neoadjuvant setting according to the IWQiG framework and Fleming hierarchy. At the metastatic setting, six meta-analyses based on aggregate data, three IPD meta-analyses and one retrospective study were identified. Within the IPD meta-analyses, at the trial-level association the squared correlation between the potential surrogates and OS ranged from 0.10 to 0.57 and no endpoint was judged to be valid surrogate for OS at the metastatic setting. The level of evidence available supporting a relationship between OS and potential surrogate endpoints in breast cancer trials is low.
