The development of oesophageal adenocarcinoma (OAC) from Barrett’s oesophagus provides an excellent model of the step-wise progression of malignancy. This process is strongly associated with the reflux of stomach contents into the oesophagus. However the exact mechanism by which low pH and bile acids contribute to the development of OAC remains unclear. The disease mostly presents late and treatment options are limited resulting in poor outcomes. A paucity of information regarding the mutations and mutational processes that drive OAC is likely contributing to this. However to understand the development of a cancer it is not enough to simply identify commonly mutated genes, rather it is also crucial to identify the timings at which these mutations occur in the development of disease. My aims in this thesis were to develop pipelines for the identification of somatic mutations using next-generation sequencing and to utilize these to provide an initial insight into the mutational signatures and genes that drive OAC. Using the unique opportunity presented by having material from multiple stages of disease development I aimed to understand better the timing of mutations in the development of cancer. By studying single nucleotide variants from 43 tumours I was able to identify the signatures of 7 mutational processes acting on the OAC genome. These include ageing, enzymatic DNA damage (by the APOBEC enzymes) and homologous recombination deficiency. Two novel signatures dominated the genomes and were seen only very rarely in tumours from other sites. These signatures may represent the action of mutagens in the novel environment found around the oesophagus and stomach with bile acids and low pH being potential culprits. Previous work has suggested that OAC may harbour large numbers of complex rearrangements and that reflux may contribute to this. I have developed and validated a pipeline for the sensitive and specific detection of structural variants in cancer. As part of this I have explored the factors contributing to false positive structural variant calls in ‘next-generation’ sequencing and developed filters to remove these. I have shown that mismappings and germline variants are the greatest source of error and therefore that choice of a highly accurate aligner is essential. Most importantly I have shown that a simple filter using mismapped reads seen in a large panel of normals is capable of filtering the vast majority of false positive variants. I also provide here a detailed sensitivity estimate for our pipeline. Using this pipeline I was able to identify a further 5 structural variant mutational processes molding the OAC genome. Finally I have identified new potential OAC driver genes including members of the SWI/SNF complex and the toll-like receptor signaling pathway.
To understand the role these mutations play in the development of OAC I screened for these mutations in samples representing multiple stages in the progression from Barrett’s oesophagus to cancer. Intriguingly almost all putative driver genes were found mutated at the earliest stages of disease development at the same frequency as seen in cancer. Importantly this questions the role of these mutations in the development of the malignant phenotype. This first pass analysis of the OAC genome has highlighted novel mutational signatures that point to the central role of the unique mutagenic exposures seen in OAC. Most importantly I have shown that the majority of OAC drivers are mutated early in the development of disease and do not predict risk of progression to invasive cancer.