RAD51 testing in patients with early HER2-negative breast cancer and homologous recombination deficiency: post-hoc analysis of the GeparOla trial

Abstract

Purpose: The randomized GeparOla trial reported comparable pathological complete response (pCR) rates with neoadjuvant containing olaparib vs. carboplatin treatment. Here, we evaluate the association between functional homologous repair deficiency (HRD) by RAD51 foci and pCR, and the potential of improving patient selection by combining RAD51 and stromal tumor infiltrating lymphocytes (sTILs). Patients and methods: This is a post-hoc blinded, biomarker analysis from the randomized GeparOla trial. Patients with early-stage HER2-negative breast cancer and HRD assessed by Myriad myChoice or BRCA1/BRCA2 mutation were randomized 1:1 to receive i) paclitaxel plus olaparib or ii) paclitaxel plus carboplatin, both followed by epirubicin/cyclophosphamide. Functional HRD was predefined as a RAD51 score ≤10% (RAD51-low). Results: Overall, 90/97 (92.8%) samples were evaluable for RAD51 testing and 72/90 (80.0%) were RAD51-low. The pCR rate in patients with RAD51-low tumors was 66.7% (48/72), while it decreased to 22.2% (4/18) in those with RAD51-high. In the multivariable model including clinicopathological factors and treatment, the RAD51 score remained significantly associated with pCR (OR=12.03, 95%CI 2.60–55.73, p=0.002). Patients with RAD51-low and high sTILs in their tumors achieved a pCR rate of 75.0% (27/36). Similar results were observed for olaparib or carboplatin. In the exploratory DFS analysis, no differences were observed between RAD51 groups (high vs. low: HR=0.85, 95% CI 0.25–2.97). Conclusions: In a pre-selected population with HRD according to a genetic test, RAD51 testing identifies patients with different pCR rates under PARPi or platinum-based therapies. Future biomarker-driven studies should consider this information to refine stratification factors and to improve patient selection.