bioRxiv2023
Although preclinical studies for drug discovery and biomarker development are extensively conducted using large publicly available cancer cell line databases, there have been no reports to date that clarify the association between homologous recombination repair deficiency (HRD) and drug sensitivity using these data. We comprehensively analyzed the molecular profiles and drug response screening data from the Cancer Cell Line Encyclopedia. Unexpectedly, gene alterations in BRCA1/2 or homologous recombination repair-related genes, HRD score, or mutational signature 3 were not significantly correlated with sensitivity to platinum agents or PARP inhibitors. Rather, higher HRD scores and mutational signature 3 were significantly associated with resistance to platinum and PARP inhibitors in multiple assays. These findings were consistent when analyses were restricted to breast and ovarian cancer cell lines and when data from the COSMIC Cell Line Project dataset were used. SignificanceIn existing cancer cell line databases, the association between HRD status and sensitivity to platinum or PARP inhibitors differs from that expected from clinical tumor data. This discrepancy may also apply to other tumor characteristics, and researchers should be aware of the potential limitations of cell line data. Structural abstractO_ST_ABSBackgroundC_ST_ABSComprehensive molecular profiling and drug sensitivity screening data from over 1000 cancer cell lines are currently available for preclinical studies including targeted drug discovery and biomarker development. However, there are no reports using these cell line databases to clarify the association between homologous recombination repair deficiency (HRD) and drug sensitivity. MethodsWe investigated the association between HRD status, including gene alterations in the homologous recombination repair (HR) pathway, HRD score, and mutational signature 3, and sensitivity to platinum agents and PARP inhibitors in the Cancer Cell Line Encyclopedia (CCLE) and the COSMIC Cell Line Project (CLP) datasets. ResultsIn the CCLE dataset (n=1182), samples with BRCA alterations, including BRCA1 methylation and BRCA1/2 mutations with locus-specific loss-of-heterozygosity, exhibited higher HRD scores and mutational signature 3. These two scores were positively correlated (r=0.475, p=1.25 x10-52). Unexpectedly, neither BRCA1/2 nor HR-related gene alterations correlated with sensitivity to platinum agents or PARP inhibitors. Instead, significantly positive correlations were observed between drug-response AUC values and HRD scores in 60% (6/10) of platinum agent assays, and in 43% (6/14) PARP inhibitor assays, while no significant negative correlation was observed. Similar results were obtained in the analysis with mutational signature 3. These findings were consistent in analyses limited to ovarian and breast cancer cell lines and in the CLP dataset, indicating samples with HRD showed resistance rather than sensitivity to these drugs. ConclusionIn existing cancer cell line databases, the association between HRD status and sensitivity to platinum and PARP inhibitors differs from that expected from clinical tumor data. This discrepancy may also apply to other tumor characteristics, and investigators should be aware of the potential limitations of cell line data.