Homologous recombination deficiency prediction using low-pass whole genome sequencing in breast cancer

Homologous recombination repair deficiency (HRD) results in a defect in DNA repair and is a frequent driver of tumorigenesis. Poly(ADP-ribose) polymerase inhibitors (PARPi) or platinum-based therapies have increased theraputic effectiveness when treating HRD positive cancers. For breast cancer and ovairan cancer HRD companion diagnostic tests are commonly used. However, the currently used HRD tests are based on high-depth genome sequencing or hybridization-based capture sequencing, which are technically complex and costly. In this study, we modified an existing method named shallowHRD, which uses low-pass whole genome sequencing (WGS) for HRD detection, and estimated the performance of the modified shallowHRD pipeline. Our shallowHRD pipeline achieved an AUC of 0.997 in simulated low-pass WGS data, with a sensitivity of 0.981 and a specificity of 0.964; and achieved a higher HRD risk score in clinical BRCA-deficient breast cancer samples (p = 5.5 × 10^-5, compared with BRCA-intact breast cancer samples). We also estimated the limit of detection the shallowHRD pipeline could accurately predict HRD status with a minimum sequencing depth of 0.1 ×, a tumor purity of > 20%, and an input DNA amount of 1 ng. Our study demostrates using low-pass sequencing, HRD status can be determined with high accuracy using a simple approach with greatly reduced cost.