Introduction/Background High-grade serous ovarian cancers with deficiency of homologous recombination DNA repair (HRD) are sensitive to the combination of bevacizumab and olaparib as maintenance therapy in PAOLA-1 trial (NCT02477644). HRD status is determined by mutational scars within tumor genome. Here, we developed a new method called GIScar (Genomic Instability Scar) suitable with the most of the academic molecular biology laboratory constraints.

Methodology We used sequencing data from a limited panel of 127 genes including genes involved in homologous recombination to detect mutational scars, i.e. chromosomic breaks, genomic deletion/duplication and allelic imbalance. The score was trained among 146 prospective samples from ovarian tumors with HRD status previously defined by Myriad Genetics^® (MG). For clinical validation, we sequenced 469 DNA tumor samples from the PAOLA-1 trial and correlated GIScar status with progression free survival (PFS).

Results On the 146 prospective samples, GIScar reached an accuracy of 92.46% compared to MG HRD status, with a sensitivity of 95.38% and specificity of 90.12%. On the 469 PAOLA-1 samples, patients with GIScar HRD positive (including tBRCAm) tumors showed a significant prolonged PFS in olaparib vs placebo arm (median PFS: 38.7 vs 20.1 months, hazard ratio (HR): 0.470 [95% CI, 0.334–0.661] as well those with GIScar HRD positive tBRCAwt tumors (23.9 vs 16.4 months, HR: 0.529 [95% CI, 0.323–0.866]). Patients with negative GIScar HRD tumors did not benefit from addition olaparib (median PFS: 16.6 vs 16.5 months, HR: 1.045 [95% CI 0.757–1.441]). Furthermore, our approach reduced by 90% (4 vs 47 tumors) the number of inconclusive status compared to MG.

Conclusion GIScar demonstrated high accuracy with MG data with less inconclusive results and identifies patients who could best benefit from maintenance olaparib added to bevacizumab. GIScar test performances allow the deployment of this test in academic molecular biology laboratories.