The effect of mutational processes and local determinants on cancer somatic mutation formation

Abstract

Somatic mutations play essential roles in cancer initiation and development. Formation of somatic mutation can be jointly caused by defect of endogenous repair mechanism like deficiency of mismatch repair and exposure of mutagens such as ultraviolet radiation, cigarette smoke and chemotherapeutic agents. Footprints left by repair and damage shape the cancer genome into unique mutational signatures, representing specific mutational processes. The distribution of mutations is uneven across the cancer genome and mutation density shows high variability at a local scale. This is largely attributed to genomic and epigenetic local determinants, including trinucleotides composition, cytosine methylation, replication timing, transcription factor binding, chromatin organization and gene expression level. By taking advantage of the vast amount of multi-omics data generated by next-generation sequencing technology, we can explore potential factors affecting cancer somatic mutation formation, thus advancing the understanding of the underlying etiology for cancer development. In this dissertation, I leverage bioinformatic approaches on multi-omics data from samples with known mutational processes to investigate somatic mutation formation. Three projects with different perspectives were carried out during my PhD study and they are detailed in the following paragraphs. First, we investigate the genome-wide mutation landscape of distinct polymerase epsilon (POLE) mutant cancers. We find that POLE mutants display different mutation profiles, and this can be explained by the discordant fidelities of these mutants in replicating specific DNA sequences. Significantly, these differences have important implications in cancer formation as we found that a POLE mutation is strongly associated with a specific truncating mutation of the TP53 cancer driver gene. This study furthers our understanding of the POLE mutagenic process in cancer and provides important insights into carcinogenesis in cancers with such mutations. Second, based on mismatch repair (MMR) deficient cancer samples across multiple cohorts, we find cancer genomes with deficiency of MMR proteins MSH2/MSH6 (MutSα) exhibit mutational signature contributions distinct from those that are deficient in MLH1/PMS2 (MutLα). This disparity arises from unrepaired 5-methylcytosine (5mC) deamination, i.e. methylation damage, rather than replicative DNA polymerase errors, suggesting a non-canonical role of MMR in the protection against methylation damage in non-dividing cells and ultimately cancer development. Third, we evaluate the influence of local determinants on panel-based tumor mutational burden (TMB) assessment. We found that cancer gene-based panels overestimate TMB calculation, potentially leading to misclassification of a subset of patients for immune checkpoint inhibitor therapy. The overestimation is largely due to increased frequency of positively selected mutations at cancer genes, which cannot simply be addressed by the removal of known hotspot mutations. We propose a cancer type and panel region-specific model that can harmonize TMB evaluation, which may have benefits for clinical practice. In summary, interplay of DNA damage and repair, combined with local determinants, has impactful influences on cancer somatic mutation formation. These studies broaden our understanding of how certain cancers develop and provide potential support for clinical efficiency.