[3] predicting cancer evolution

By combining experimental and theoretical approaches, we scan cancer genomes for causal genes and for genetic interactions to better understand cancer evolution.

Needle in a haystack. Most somatic mutations found in cancer cells are ‘passengers’, which are not selected. Detecting the few mutations among those which are ‘drivers’ is challenging, yet crucial to understand carcinogenic transformation. We suggest that synonymous mutations sometimes drive cancer by affecting splicing patterns of oncogenes and the TP53 tumor suppressor gene (Supek et al. 2014 Cell).

Achilles' heel of cancers. Cancer is hard to treat. Ongoing hypermutation may present vulnerabilities particular to cancer cells, since high mutation rates rare in non-cancerous somatic cells. One example is mutagenesis due to the APOBEC3A enzyme in lung cancers. An experimental approach via CRISPR/Cas9 screens in isogenic cell lines, combined with a genomics approach, suggested a novel vulnerability of lung cancer cells (Biayna et al. 2020 PLOS Biology). HMCES, an abasic site sensor protein, protects from toxic repair intermediates of APOBEC3A DNA lesions, and inhibiting HMCES is synthetic-lethal with overexpression of APOBEC3A.

Cutting off the mutation supply. Beyond the particular case of lung and APOBEC3A, we find that mutation signatures in genomes of cancer cells are highly predictive of responses to various drugs, often exceeding conventional genomic markers such as driver mutations or copy number alterations (Levatic et al. 2021 bioRxiv). Targeted therapies directed towards hypermutating cells would slow down mutation rates of the tumor mass, preventing therapy resistance and occurrence of late-stage drivers.