Prediction of homologous recombination deficiency identifies colorectal tumors sensitive to PARP inhibition

Abstract

Abstract The synthetic lethal effect observed with the use of PARP inhibitors (PARPi) with tumors characterized by the loss of key players in the homologous recombination (HR) pathway, commonly referred to as “BRCAness”, is maintaining high interest in oncology. While BRCAness is a well-established feature in breast, ovarian, prostate, and pancreatic carcinomas, our recent findings indicate that up to 15% of colorectal cancers (CRC) also harbor defects in the HR pathway, presenting promising opportunities for innovative therapeutic strategies in CRC patients. We developed a new tool called HRDirect, which builds upon the HRDetect algorithm and is able to predict HR deficiency (HRD) from reference-free tumor samples. We validated HRDirect using matched breast cancer and CRC patient samples. Subsequently, we assessed its efficacy in predicting response to the PARP inhibitor olaparib by comparing it with two other commercial assays: AmoyDx HRD by Amoy Diagnostics and the TruSight Oncology 500 HRD (TSO500-HRD) panel by Illumina NGS technology. While all three approaches successfully identified the most PARPi-sensitive CRC models, HRDirect demonstrated superior precision in distinguishing resistant models compared to AmoyDX and TSO500-HRD, which exhibited overlapping scores between sensitive and resistant cells. Furthermore, we propose integrating HRDirect scoring with ATM and RAD51C immunohistochemical analysis as part of our “composite biomarker approach” to enhance the identification of HRD tumors, with an immediate translational and clinical impact for CRC personalized treatment.