|dc.description.abstract||Aberrant DNA methylation patterns are a universal feature of cancer, yet we understand relatively little of the cause and consequence of DNA methylation in the cancer process. Colorectal cancer (CRC) morbidity and mortality is a massive burden on the New Zealand, and worldwide, health care system. CRC represents an exciting model for the study of DNA methylation in cancer owing to the CpG island methylator phenotype (CIMP) subtype, which has a high degree of cancer-specific DNA methylation. This PhD is comprised of three chapters investigating: a) aberrant DNA methylation in CRC b) the mechanisms underlying the observed methylation patterns and c) the potential clinical utility of cancer-specific DNA methylation. In the first chapter, genome-wide DNA methylation is characterised using the high-density Illumina 450K HumanMethylation beadchip. Three distinct methylation subtypes (CIMP-High, CIMP-Low and CIMP-Negative) were identified within the tumours analysed. Hypermethylation was detected at over 2,000 CpG islands across the cohort, however many of these CpG islands were hypermethylated in only a fraction of cases. In contrast to this, a set of over 250 CpG islands were hypermethylated in more than 90% of CIMP-H tumours. This was likened to the model of ‘mountains’ (the small number of highly consistent events) and ‘hills’ (the large number of events which occur at a very low frequency), which has previously been used to describe driver and passenger mutation frequencies in cancer. The consistently hypermethylated CpG islands are linked to biologically relevant genes which play important roles in cellular differentiation and cell fate decisions. During this PhD, another group demonstrated a mechanistic link between activating BRAF and KRAS mutations and the CIMP-High and CIMP-Low phenotypes. Prompted by this discovery, I investigated the activity of the RAS/RAF pathway in the 20-30% of CIMP-H tumours which lack characteristic BRAF/KRAS mutations. I show that activation of the RAS/RAF pathway occurs even in BRAF/KRAS wildtype tumours, and propose a model in which DNA methylation is critical to the progression of CIMP tumours. Finally, I investigate the capacity for aberrant DNA methylation marks to differentiate CRC from non-cancerous tissue and identify loci which could be developed into biomarkers for improved detection of CRC.
This thesis contributes to our understanding of CRC epigenetics through the identification of consistently hypermethylated regions and suggests a mechanism by which hypermethylation occurs in a subset of CIMP-High tumours. Loci identified in this study can accurately discriminate CRC from non-cancerous tissue and could be developed into a simple, cost-effective test for the detection of CRC.||