Research Questions:
1. How do genetic determinants of lipid metabolites interact with lipoproteins to influence CVD risk?
2. Can genetic clustering of lipid-lipoprotein interactions provide novel insights into lipid-mediated disease mechanisms?
3. Do polygenic risk scores (PRS) derived from lipid-lipoprotein clusters predict intermediate atherosclerotic phenotypes and CVD outcomes?
Objectives:
1. Develop partitioned polygenic risk scores (PPRS) reflecting genetically determined lipid-lipoprotein interactions using genome-wide association study (GWAS) data.
2. Validate the generated PPRS against measured lipid metabolites in cohorts with available lipidomic data.
3. Assess the predictive value of these PPRS for incident CVD events in the UK Biobank (UKB).
Scientific Rationale:
Lipidomics studies have identified numerous lipid metabolites associated with CVD, yet their effects are often confounded by their lipoprotein carriers. Most lipidomics analyses measure total plasma lipid levels without considering the class-specific origin of these metabolites. Since different lipoproteins (e.g., LDL, VLDL, HDL) exhibit distinct lipid compositions and atherogenic properties, it is crucial to disentangle the role of lipoprotein-specific lipids in CVD risk.
Genetic clustering provides a powerful tool to define biologically relevant lipid-lipoprotein interactions. By leveraging UKB data, this project will apply machine-learning-based genetic clustering to classify variants associated with both lipid metabolites and conventional lipoprotein measures (LDL-C, triglycerides, Lp(a), HDL-C). The resulting genetic clusters will be used to construct PPRS, which will be tested for associations with imaging-based markers of atherosclerosis, metabolic dysfunction, and incident CVD events.