AbstractGenetic and epigenetic data provide the opportunity to robustly appraise the causal effect of an exposure on an outcome of interest, improve understanding of risk and prognostic pathways, and predict the status of a risk factor, prognostic factor, or outcome. In the case of oropharyngeal cancer (OPC), genetic and epigenetic data have rarely been applied in these contexts.
Using large population-based OPC cohorts alongside bioinformatic, genetic and epigenetic resources, I have applied a series of methodologies which aim to improve understanding of the causal risk factor pathways associated with this disease, beyond the limited degree of inference afforded by conventional observational studies. To this end, throughout this thesis I have employed: enriched literature object mining, genome-wide association studies (GWAS), epigenome-wide association studies (EWAS), two-sample Mendelian randomization (MR), MR-phenome-wide association studies (MR-PheWAS), two-step MR and epigenetic prediction scores.
The OPC cohorts forming the core data resources in this thesis are the Head and Neck 5000 study (HN5000) and the head and neck cancer OncoArray study: HN5000 contains genetic, epigenetic and mortality data for 448 individuals with oropharyngeal cancer, whilst OncoArray contains genetic data on 2,641 cases and 6,585 controls.
Methods applied in this thesis have provided evidence for enrichment in literature of 4 risk factors for OPC, the association of 16 phenotypes with OPC incidence, novel whole-blood-based CpG sites associated with HPV, alcohol, smoking and oropharyngeal cancer survival, evidence for a causal effect of smoking-related methylation at the SPEG gene with OPC survival, and finally, evidence for the value of blood-based methylation signatures in predicting mortality in those with OPC. These findings highlight the merits of using genetic and epigenetic data to improve on conventional observational analyses, with the caveat that replication and triangulation of these findings from a range of methodological approaches is the optimal route to ensure their robustness and validity.
|Date of Award||24 Mar 2020|
|Supervisor||Caroline L Relton (Supervisor), Steve J Thomas (Supervisor), Rebecca Richmond (Supervisor) & Hannah R Elliott (Supervisor)|