Drivers of homologous recombination deficiency (HRD) in metastatic colorectal cancer (mCRC).

Abstract

3577 Background: HRD is linked to sensitivity to platinum-based chemotherapy and poly (ADP-ribose) polymerase inhibitors across various tumor types. However, the presence of HRD, driver events (such as pathogenic mutations in HR genes or other genomic alterations), and its clinical relevance in mCRC remain underexplored. Methods: We performed the VHIO-300 test, an ISO15189 accredited custom NGS panel profiling over 450 genes (including HR-related genes: ATM, BRCA1, BRCA2, BRIP1, CHEK2 or PALB2 ) on 356 Stage IV mCRC patients (corresponding to 247 primary colorectal and 109 to metastatic samples) enrolled in the Vall d’Hebron Institute of Oncology’s Molecular Prescreening Program from June 2021 to December 2024. All samples had a tumor cellularity > 40% as per pathologist evaluation . An HR score (sHR) based on genome-wide copy number alterations (CNA) and loss of heterozygosity (LOH) patterns is generated. After cross-validation with Myriad MyChoice, sHR ≥ 56 was established based on a cohort of ovarian tumors and used to identify HRD in mCRC tumor samples. Results: HRD prevalence was 3.4% in our mCRC cohort (12/356), but, much higher in metastatic lesions, (6.4%, 7/109) than in primary samples (2.03%, 5/247) (p = 0.05). In fact, the median sHR between primary (21) vs. metastases (30) in CRC was significantly different (p < 0.01). Regarding HR gene status, HR-mutated samples were not significantly within the HRD group (p = 0.27) and only 6.5% (2/31) were HRD. Noteworthy, BRCA2 exhibits a frameshift deletion in a homopolymer stretch, that is a frequent hotspot in microsatellite instable (MSI) tumors, but this event was not found to be associated with HRD. In fact, all HRD tumors (n = 12) were microsatellite stable (MSS). Other frequent alterations in mCRC were studied and BRAF mutations were found to be present in 42% of HRD tumors (p < 0.01). Inversely, HRD was rare in KRAS -mutated samples (0.7%; 1/140) and, in fact, highly correlated with non-HRD status (p = 0.02), especially the G12 mutation (p < 0.01). Interestingly, CNA profiles also revealed a strong association between the BCL2L1 loci gain and HRD (p < 0.01). Clinically, HRD was not significantly associated with prognostic value nor clinical benefit to oxaliplatin-based combinations. However, the limited sample size and heterogeneous treatment lines restricted robust statistical analysis. Conclusions: This study identified a small, yet significant subset of mCRC that displays HRD. sHR and HRD rates were higher in metastatic lesions vs primary tumors, indicating HR scarring could be accumulating over time in some mCRC patients. No clear association between pathogenic HR gene mutations and HRD were found, suggesting the involvement of alternative molecular mechanisms in this process. Frequent co-occurring events, such as BRAF mutations or BCL2L1 gains could be drivers in CRC HRD, and shape, eventually, new therapeutic options for these patients in the metastatic setting.