Pathology - Theses
Permanent URI for this collection
Search Results
1 - 1 of 1
-
ItemNetwork approaches to understanding biomarker biology( 2017)Common non-communicable diseases such as cardiovascular diseases, chronic respiratory diseases, and diabetes are the leading causes of premature mortality and ill-health worldwide. These are complex diseases that take multiple decades to manifest with myriad genetic, lifestyle, and environmental risk factors. The last decade has seen rapid technological advancements and falling costs to high-throughput omic profiling platforms enabling large-scale studies into the molecular differences between healthy individuals and those that will later develop disease. Epidemiological studies of omic data in population cohorts have identified many new biomarkers of future disease risk and mortality. The end-goal of biomarker research is not only to identify individuals at increased risk of disease, but also to find ways to intervene to reduce that risk. Identification and characterisation of the biological processes these biomarkers participate in or reflect is a fundamental step in the process of clinical translation. This thesis explores the use of network-based approaches for identification and characterisation of aberrant biological processes associated with biomarkers using population cohort multi-omic data. A scalable and efficient method for robust statistical assessment of network module preservation and reproducibility is developed: NetRep. Multi-omic data from three large population biobanks are analysed to identify and characterise biological processes associated with elevated GlycA levels; a heterogeneous NMR signal that has been of recent interest as a biomarker for long-term risk of cardiovascular disease, type II diabetes, and premature mortality. I found that elevated GlycA levels corresponded with the presence of sub-clinical inflammation and increased coordinated expression of a reproducible gene coexpression network module indicative of neutrophil activity. Accordingly, analysis of a population cohort with linked electronic health records showed that elevated GlycA levels had long-term consequences for increased risk of severe infections up to 14-years in the future. To fine-map the GlycA biomarker, I developed accurate imputation models for predicting concentrations of three of the five glycoproteins contributing to the GlycA signal from population-based serum NMR data: alpha-1-acid glycoprotein (AGP), alpha-1 antitrypsin (A1AT), and haptoglobin (HP). Imputation of these three glycoproteins in two large population cohorts with linked electronic health records revealed elevated A1AT had the most severe long-term ramifications for future disease and mortality risk over an 8-year follow-up period. In total, this thesis shows the utility of leveraging population-based omic data for elucidating biomarker biology and provides a useful framework to guide future studies of new and established biomarkers for future disease risk.