Optical Genome Mapping for detecting Homologous Recombination Deficiency (HRD) in human breast and ovarian cancers

Homologous recombination deficiency (HRD) leads to genomic instability that marks HRD tumor genome with a specific genomic scar. Present in many cancers, HRD is important to be detected as it is associated with a hyper-sensitivity to some classes of drugs, in particular the PARP inhibitors. Here, we investigated the use of structural variants (SVs) detected by the Optical Genome Mapping (OGM) technology as biomarkers to identify HRD tumors. We analyzed SVs data obtained by OGM from 37 samples of triple-negative breast cancer or high grade ovarian cancer with the known HRD status. We found that HRD cases were enriched with duplications and reciprocal translocations, while nonHRD cases were enriched with inversions. The number of translocations, defined as inter-chromosomal or intra-chromosomal rearrangements of more 5Mb were similar in HRD and nonHRD cases. We defined isolated translocations as the subset of translocations having no other translocation within 2 megabase zone around both junctions, and demonstrated that the number of isolated translocations perfectly discriminated HRD and nonHRD cases in the training series. Validation series consisting from 26 cases showed 20% false positive and zero false negative error rate, which proved isolated translocations to be 100% sensitive and 80% specific SV marker of HRD. Our results demonstrate that the OGM technology is an affordable way of getting an insight of the structural variants present in solid tumors, even with low tumoral cellularity. It represents a promising technology for HRD diagnosis, where a single marker already gives 80% correct recognition. ### Competing Interest Statement This work was supported by a Sponsored Research Agreement with Bionano Genomics and S.V. was supported by this grant. T.P. and M.-H.S. are co-inventors of the LST method (US20170260588, US20150140122 and exclusive license to Myriad Genetics). T.P., M.-H.S. C.C. and A.E. are co-inventors of the shallowHRDv2 method (EP23170829).