Although defects in cancer susceptibility genes within the DNA-damage response (DDR) machinery including BRCA1 and BRCA2 account for only 5-10% of all breast cancer cases, these defects are highly penetrant and significantly increase the risk of breast (60-80%) and also ovarian (35%) cancers [1]. Together with defects in the DNA-damage sensor ATM, apoptosis effector TP53, and PTEN and CDH1 with roles in regulation of DDR, these account for considerable proportions of sporadic breast (63%) and ovarian (85%) cancers. To compensate for these DDR defects and to avoid cell death triggered from a genomic catastrophe, cancer cells rewire their DDR network while also selecting (during clonal expansion) the optimal combination of oncogenic events. Deciphering these combinations of events would aid in mapping the vulnerabilities of cancer cells harbouring defects in DDR.

While there have been several studies screening for essentiality of genes across DDR-deficient cell-lines, the essential genes so identified are either restricted only to these cell-line models or are not frequently (over)expressed in cancers. Here, we observe that oncogenic events that are mutually exclusive to DDR defects in large proportions of cancers constitute the (clonally) selected combinations that are amenable to cancer-cell survival, and therefore by systematically mining for these events, we infer vulnerability genes that if targeted in conjunction with DDR defects could induce a genomic catastrophe and trigger cancer-cell death.

Using data from DNA copy-number and mRNA-expression profiles we infer vulnerability genes that are mutually exclusive to defects in six DDR genes ATM, BRCA1, BRCA2, CDH1, PTEN and TP53 across four cancers (total 3980 samples) – breast (2029), prostate (623), ovarian (828) and uterine (500) from The Cancer Genome Atlas. Interestingly, across the four cancers these vulnerability genes form the most combinations with BRCA2 (59.02%), followed by CDH1 (24.59%), PTEN (8.20%) and TP53 (8.19%) at p<0.01 (1-hypergeometric test), whereas these show distinct patterns within the individual cancers: combinations dominated by CDH1 (90%) in breast, PTEN (78.38%) and BRCA2 (16.82%) in prostate, and BRCA1 (71.94%) and TP53 (16.21%) in ovarian cancers. Validation using GARP (Gene Activity Rank Profile)-score data from essentiality screens [2] from ten breast cancer cell lines (HCC1143, HCC1187, HCC1395, HCC1419, HCC1428, HCC1500, HCC1806, HCC1954, HCC38, MCF7) which harbour defects in at least one of the six DDR genes shows remarkable agreement between the GARP rankings and our inferred vulnerabilities. Our inferred genes are significantly enriched (p<0.0001 X2 test) in the top quartile of the entire set of profiled (∼16000) essential genes in these screens. Moreover, Kaplan-Meier analysis using survival data from 1000 breast cancer patients shows considerable overexpression of these genes (e.g.TLK2 in 37% luminal cases) which correlates significantly (TLK2: p<0.0006; Grade 3 hazard ratio 2.5) with poor prognosis. Experimental validation of these genes using single- and double knockout with DDR in breast cancer cell lines is currently underway.

[1] Liu & Srihari et al., Nucl Acids Res 2014, 42(10):6106-27.

[2] Marcotte et al., Cancer Discov 2012, 2(2):172-89.

Citation Format: Srihari S, Singla J, Wong L, Simpson PT, Khanna KK, Ragan MA. Identifying genetic vulnerabilities in cancers driven by defects in DNA-damage response. [abstract]. In: Proceedings of the Thirty-Eighth Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2015 Dec 8-12; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2016;76(4 Suppl):Abstract nr PD6-05.