Repository logo
 

The Intra-Tumour Heterogeneity Landscape of Human Cancers


Type

Thesis

Change log

Authors

Dentro, Stefan Christiaan 

Abstract

Tumours accumulate many somatic mutations in their lifetime. Some of these mutations, drivers, convey a selective advantage and can induce clonal expansions. Incomplete clonal expansions give rise to intra-tumour heterogeneity. Somatic mutations can be measured through massively parallel sequencing, where mutations that are supporting incomplete expansions will appear as subclonal. These mutations can be used as a marker of the existence of the expansion and allow for a window into the clonal and subclonal architecture of the tumour at diagnosis. During my Ph.D. I have developed computational methods to infer intra-tumour hetero- geneity from massively parallel sequencing data and applied these to the 2,778 tumour whole genome sequences in the International Cancer Genome Consortium Pan-Cancer Analysis of Whole Genomes initiative to paint the pan-cancer landscape of intra-tumour heterogeneity. I will first introduce the methods; a method to call somatic copy number alterations (Battenberg) and a method to infer subclones from single nucleotide variants (DPClust). Both are extensively validated on simulated and on real data, and I describe a rigorous quality control procedure. The methods are then applied to a single sample to showcase what can be learned about the life history of a cancer, before introducing additional computational methods for a pan-cancer study of heterogeneity. Finally, I describe the findings. I find that nearly all cancers, for which there is sufficient power, contain at least one subclone (96.7% of 1,801 primary tumours). The subclones contain driver mutations that are under positive selection, and known cancer genes contain subclonal driver mutations in low proportions. 9.5% of tumours contain only subclonal drivers that are clinically actionable, suggesting that heterogeneity could inform treatment choices. Finally, the analysis reveals that activity of smoking and UV-light associated mutational signatures goes down as the tumour evolves, while activity of the APOBEC associated signatures goes up.

Description

Date

2017-12-05

Advisors

Adams, David
Van Loo, Peter
Wedge, David

Keywords

Cancer, Sequencing, Copy Number Alterations, Subclonal Architecture

Qualification

Doctor of Philosophy (PhD)

Awarding Institution

University of Cambridge
Sponsorship
This thesis was funded by the Wellcome Trust.