Lack of RAD51 foci formation enables the identification of PARP inhibitor sensitive breast tumors

Poly-(ADP-ribose)-polymerase (PARP) inhibitors (PARPi) are active anti-cancer agents in tumors deficient in DNA repair by homologous recombination (HR). Clinical studies with single-agent PARPi are therefore focused on breast and ovarian cancers with BRCA1/2 mutations and/or those identified as HR-deficient by genomic scars. Nevertheless, there is a need to improve patient selection by characterizing the actual tumor's HR-status. Here, we sought to validate the RAD51 immunofluorescence assay as a functional readout of the tumor's capacity to complete HR DNA repair by examining its correlation with PARPi primary response and acquired resistance using patient-derived tumor xenografts (PDX). We developed two independent PDX panels, totaling 56 models. Specifically, the discovery panel consisted of 25 PDXs from patients harboring (n = 12) or not (n = 13) germline BRCA1/2 (gBRCA) mutations, namely from 23 primary or advanced breast cancer and 2 high-grade serous metastatic ovarian cancer (HGSOC). The antitumor activity of the PARP1/2 inhibitor olaparib as single agent (50 mg/kg) was assessed in this panel. In addition, in vivo acquired resistance was modeled in the PDXs by exposing 3 PARPi-sensitive PDXs to olaparib for >80 days, until individual tumors regrew. The tumor's HR capacity was evaluated by measuring the percentage of cells in the S/G2 phase of the cell cycle (geminin-positive) exhibiting RAD51 nuclear foci, both in PARPi- and in vehicle-treated formalin-fixed paraffin embedded (FFPE) tumors. The validation set is composed of 28 triple negative breast cancer PDXs including 5 gBRCA PDXs. The percentage of RAD51-positive cells in the discovery set ranged between 1 and 83% in PARPi-treated PDXs and between 0 and 62% in vehicle-treated tumors. Five out of 25 PDXs (20%) with low levels of RAD51 -positive cells (1-3% in PARPi-treated tumors) exhibited tumor regression upon olaparib treatment. The 20 PDXs that showed disease stabilization (n = 2), disease progression (n = 18) and those that developed acquired resistance to olaparib (n = 3) exhibited significantly higher levels of RAD51-positive cells (27-83% in PARPMreated tumors, p< 0.0001). Similar results were obtained in vehicle-treated tumors. Of note, PARPi responsiveness correlated with low levels of RAD51 foci independent of the BRCA status. The validation set is composed of 7 regressing-and 21 stabilized/progressing-models, whose correlation with the RAD51 predictive value will be presented. Our re