Building a multiomic homologous recombination deficiency clinical test using Adaptive Sampling

Abstract: Homologous recombination deficiency (HRD) affects breast, ovarian, pancreas and prostate cancers. Patients with HRD tumours can respond to poly ADP-ribose polymerase (PARP) inhibitors and platinum-based therapies. HRD results in a variety of single base, insertion and deletion, and large structural variants (SVs) that serve as biomarkers for detection. However, these mutations can occur years prior to diagnosis and might not reflect the true HRD status of the tumour and its response to therapy. We hypothesize that we can detect features of HRD through single base, large SV, and methylation signatures that provide a comprehensive set of HRD features, representative of current HRD status and predict treatment response. We compared the performance of mSigDetectHRD with whole-genome sequencing (WGS) of the tumours as ground truth, to build a cost-effective test using Adaptive Sampling. Mutations accumulate non-uniformly throughout the genome and we hypothesize that we can target regions of higher signature activity to obtain better detection. In addition, using multiple features of HRD we can distinguish BRCA1 and BRCA2 loss that can have clinical importance for familial genetic testing. Last, by training our classifier on public methylation data from cancer genomes, we developed a HRD methylation signature that corresponds with genomic features of HRD and predict patients’ response to platinum-based therapies. We envision our approach to develop multiomic diagnostic tests, based on Adaptive Sampling, to be able to address clinical challenges in predicting treatment responses and provide an unprecedented depth of information for cancer diagnosis in the future. Biography: Dr Alvin Ng is the Dean’s postdoctoral fellow in Lee Kong Chian School of Medicine. Alvin was trained in computational biology and genomics in his PhD in Duke-NUS Medical School and postdoctoral training in the Early Cancer Institute, Cambridge. He is focused on using long reads to detect mutational processes associated with DNA repair defects and understand how complex rearrangements arise in cancers. In the last year he developed methods to detect these mutational signatures and characterize highly complex cancer genomes.