Background: Germline mutations in the BRCA1 or BRCA2 (BRCA) genes predispose to hereditary breast and ovarian cancer and, mostly in the case of BRCA2, are also prevalent in cases of pancreatic and prostate malignancies. Tumours from these patients tend to lose both copies of the wild-type BRCA gene, which makes them exquisitely sensitive to platinum drugs and poly(ADP-ribose) polymerase inhibitors (PARPi), treatments of choice in these disease settings. Reversion secondary mutations with the capacity of restoring BRCA protein expression have been documented in the literature as bona fide mechanisms of resistance to these treatments. Patients and methods: We analysed published sequencing data of BRCA genes (from tumour or circulating tumour DNA) in 327 patients with tumours harbouring mutations in BRCA1 or BRCA2 (234 patients with ovarian cancer, 27 with breast cancer, 13 with pancreatic cancer, 11 with prostate cancer and 42 with a cancer of unknown origin) that progressed on platinum or PARPi treatment. Results: We describe 269 cases of reversion mutations in 86 patients in this cohort (26.0%). Detailed analyses of the reversion events highlight that most amino acid sequences encoded by exon 11 in BRCA1 and BRCA2 are dispensable to generate resistance to platinum or PARPi, whereas other regions are more refractory to sizeable amino acid losses. They also underline the key role of mutagenic end-joining DNA repair pathways in generating reversions, especially in those affecting BRCA2, as indicated by the significant accumulation of DNA sequence microhomologies surrounding deletions leading to reversion events. Conclusions: Our analyses suggest that pharmacological inhibition of DNA end-joining repair pathways could improve durability of drug treatments by preventing the acquisition of reversion mutations in BRCA genes. They also highlight potential new therapeutic opportunities when reversions result in expression of hypomorphic versions of BRCA proteins, especially with agents targeting the response to DNA replication stress.

Synthetic lethality is a promising concept in cancer research, potentially opening new possibilities for the development of more effective and selective treatments. Here, we present a computational method to predict and exploit synthetic lethality in cancer metabolism. Our approach relies on the concept of genetic minimal cut sets and gene expression data, demonstrating a superior performance to previous approaches predicting metabolic vulnerabilities in cancer. Our genetic minimal cut set computational framework is applied to evaluate the lethality of ribonucleotide reductase catalytic subunit M1 (RRM1) inhibition in multiple myeloma. We present a computational and experimental study of the effect of RRM1 inhibition in four multiple myeloma cell lines. In addition, using publicly available genome-scale loss-of-function screens, a possible mechanism by which the inhibition of RRM1 is effective in cancer is established. Overall, our approach shows promising results and lays the foundation to build a novel family of algorithms to target metabolism in cancer.