Background: Clinical benefit of PARP inhibitors or Platinum-based chemotherapy has been observed and well proved in BRCA mutation patients. However, there are other mechanisms to induce recombination repair deficient (HRD), such as BRCA methylation, ATM expression loss; HRR gene mutation in addition to BRCA mutation. Whether these complicated HRD events or phenotype could be measured by a unified model to accurately identify HRD positive patients is important.
Methods: In this study, we use support vector machine method to build genomic scar (GS) model to predict HRD events, which takes different types of chromosomal copy number into consideration to evaluate genomic instability. The accuracy of GS model was tested by different NGS platform from WGS to two captured SNP Panel. GS status association with Cisplatin-based chemotherapy was also preliminarily validated.
Results: GS model displayed more than 95% accuracy to detect BRCA-ness events and is also able to enrich different types of HRR dysfunction, such as HRR-related mutation or BRCA1 promoter methylation. Moreover, GS model can identify patient benefits from Cisplatin-based chemotherapy, and GS score (GSS) high group had longer progression free survival (PFS) than GSS low group (11 vs 8.5 months, HR=0.5, P=0.05). Meanwhile, GSS showed the high concordance among different NGS platform including WGS; two captured SNP Panels, which implied the high robustness of GS model.
Conclusion: GS was a robust model to predict HRD and had the clinical potential to enrich patients who will response to PARP inhibitors and/or Platinum-based chemotherapy.
Citation Format: Hao Wen, Shuang Yang, Xuejun Chen, Ying Weng, Changbin Zhu, Li Ruan, Hua Dong, Xiaohua Wu. Genomic scar score: a robust model to predict recombination repair deficient based on genomic instability [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 2044.