Benchmarking homologous recombination deficiency algorithms for prediction of clinical outcome in ovarian cancer

Abstract

Homologous recombination deficiency (HRD) is a predictive biomarker for response to platinum-based chemotherapy in ovarian cancer. In this retrospective study, we benchmarked three HRD detection algorithms - CHORD, ShallowHRD, and OvaHRDscar - alongside BRCA1/2 mutation status in a cohort of 100 patients with high-grade serous ovarian carcinoma (HGSC). HRD status was derived from whole genome sequencing of tumor/normal samples, and progression-free survival (PFS) was used as the primary endpoint. All three HRD algorithms showed a statistically significant association with improved PFS. In multivariate Cox regression models adjusted for age, FIGO stage, tissue type, and neoadjuvant chemotherapy, HRD-positive status was significant associated with reduced hazard of progression or death: OvaHRDscar (HR = 0.41, 95% CI: 0.25-0.67), ShallowHRD (HR = 0.50, 95% CI: 0.31-0.79), and CHORD (HR = 0.47, 95% CI: 0.24-0.94). In contrast, BRCA1/2 mutation status did not show a significant association (HR = 0.64, 95% CI: 0.35-1.16). These findings support that HRD algorithms may aid in the diagnostic assessment of HRD and support broader genomic profiling to enhance clinical decision-making. Future studies should focus on refining algorithm thresholds and validating these results in larger, multi-center cohorts to facilitate clinical translation.