The timing of key events and mutational processes in tumour evolution

Abstract

Cancer develops from normal tissues through the process of somatic evolution over time. The cells in our body continually acquire errors in their DNA through the effects of varied exogenous mutagens and cell-intrinsic mechanisms. These mutations expose cells to the effects of natural selection; while many are phenotypically neutral, those which confer a fitness advantage can lead to clonal population expansions through increased cellular proliferation, or decreased cell death. The sequential accumulation of such advantageous mutations ultimately gives rise to cancer. To date, little is known of the transition between normal and cancerous tissues, either in terms of timescale, or in the typical sequence of genomic changes. The advent of next-generation sequencing technologies has allowed the comprehensive identification of the typical genomic alterations acquired by many different types of cancer. Furthermore, from sequencing data, it is now possible to reconstruct a partial sequence of mutations acquired by the tumour genome, even from a single biopsy. In this thesis, I describe the application of such \life history" analyses to a pan-cancer cohort of over 2,500 whole genome sequenced tumour samples, to learn more about cancer's evolutionary past. The timing of individual point mutations and larger chromosomal changes during tumour evolution may be estimated using their ordering relative to one another. I applied such an approach to estimate the timing of copy number gains across 2,778 cancers. This analysis reveals distinct patterns of genomic evolution between different tumour types, and identified recurrent, early gains of specific chromosomes, including chromosomes 7, 19 and 20 in glioblastoma, and isochromosome 17q in medulloblastoma. Signatures of mutational processes were extracted from groups of timed mutations to quantify the changing influence of mutagenic exposures on the tumour genome over time. Pan-cancer, I observe general patterns of signature activity, with an early role for mutational exposures such as smoking and UV light, while defective DNA repair mechanisms and chemotherapeutic agents contribute more to the later stages of tumour evolution. Intriguingly, I observe speci c patterns of activity for signatures with no known aetiology, which may give some insight into their biological derivation. These inferences were then integrated into typical timelines of tumourigenesis for 32 cancer types, which reveal the distinctive evolutionary narratives of tumours from different tissues. The incorporation of real time estimates into these timelines indicates that much of tumour evolution begins years, if not decades, before diagnosis. Using the timing of individual events, I also develop an approach that aims to characterise specific, recurring sequences of events, and apply this to one cohort as a proof-of-principle analysis. Overall, these results contribute to our understanding of tumour evolution across many different cancer types. Early events are identified which may provide suitable candidates for biomarkers, allowing earlier detection and therapeutic intervention. The cancer timelines provide a first glimpse at the timing of specific events and mutational processes during tumour development, and may perhaps form a basis for further study with a view to predicting the evolutionary future of cancer.

Date of Award	1 Jun 2020
Original language	English
Awarding Institution	King's College London
Supervisor	Peter Van Loo (Supervisor) & Francesca Ciccarelli (Supervisor)