Evaluation of different methods to assess homologous recombination status and sensitivity to PARP inhibitors in ovarian cancer

Abstract

Methods: Matched samples of ascites and tumor tissue were taken from patients undergoing surgery for epithelial ovarian cancer. Tumor samples were formalin fixed and paraffin embedded (FFPE); ascites samples were used to generate primary cultures (PC). HR status was determined in PCs as previously described.[1] IC50 for the PARP inhibitor Rucaparib was estimated using SRB assays. DNA was extracted from the FFPE tissue. The following techniques were evaluated in PCs or paired FFPE samples: DR-GFP reporter assay, PARP activity assay, BRCA1 expression on immunohistochemistry, BRCA1 methylation status and BRCA1/2 mutation analysis. A next generation sequencing based assay was used to detect mutations and other genomic alterations in a large panel of cancer-associated genes, including BRCA1/2. Results: Paired samples were collected from 64 patients and characterized for HR function. 47/64 (76%) were high grade serous. 44% (28/64)) were HR defective (HRD) by Rad51 assay and correlated with Rucaparib sensitivity (PPV-92%, NPV-100%). Molecular analysis revealed that all mutations and other genomic alterations detected in ascites derived PCs were also found in matched FFPE tumor tissues. All tumors with serous histology contained p53 mutations, whilst the remaining tumors without p53 mutations were non-serous in histology. DR-GFP assay was unreliable in PC due to poor transfection. In a subset of 50 cancers there was reduced BRCA1 expression in the HRD vs. HRC tumours (34.8% vs. 22.7%, ns) whilst in a further subset of 30 cases there was no difference in endogenous or stimulated PARP activity between HRD and HRC tumours. Deleterious BRCA2 mutations were identified in 7 tumors, 6 of which were HRD. Only 1 deleterious BRCA1 mutation was detected but methylation of BRCA1 was identified in 13 of 64 (20%) tumors, 7 of which were HRD. Mutation of BRCA2 was mutually exclusive to methylation of BRCA1. HRD vs. HRC tumours showed BRCA1 methylation (25% vs. 17%) and BRCA1/2 mutation (21% vs. 0.3%). 14/28 (50%) HR defective tumors do not have BRCA1/2 mutations or BRCA1 methylation, suggesting other mechanisms can also result in a HR defective phenotype. 28/64 (43%) of samples demonstrated the HR defective phenotype. In all cases HR status correlated with sensitivity to Rucaparib. Conclusion: As expected, deleterious BRCA2 mutations conferred a HRD phenotype in cells but methylation of BRCA1 was not universally associated with HRD. This may be as a result of only partial silencing of the gene by methylation and further work is required to identify thresholds of methylation which may predict HR status. The use of BRCA1/2 mutation testing alone is unlikely to identify the majority of HR defective ovarian tumors. Assessment of functional status of HRD is the preferred option and further technologies should be developed to simplify the Rad51 assay.