Investigating the role of lipid metabolism in non-communicable diseases and mortality

Abstract

Background: Globally, cardiovascular disease (CVD) is the leading cause of mortality. Targeting lipids and lipoproteins is essential for CVD prevention. Knowledge gaps exist in the trade-off between benefits and risks of lipid modification, due to a lack of comprehensive understanding of the effects of lipid traits and lipid-lowering therapies, including on coronary artery disease (CAD), mortality, body mass index (BMI), type 2 diabetes, gallstone disease, and immune-related diseases, and their underlying mechanisms. As a result, optimal use of lipid-lowering therapies by ancestry and sex remains unclear. I used Mendelian randomization, an instrumental variable analysis with genetic instruments, to fill these gaps which are difficult to address using randomized controlled trials and to obtain less confounded estimates than observational studies.

Methods: I took advantage of individual-level data from the UK Biobank and summary-level data from the largest relevant publicly available genome-wide association studies to assess genetically proxied exposures on outcomes. I investigated dose-response associations of lipid traits with CAD risk and mortality, sex-specific associations of lipid traits with type 2 diabetes risk, glycaemic traits, and sex hormones, the associations of current low-density lipoprotein (LDL)-lowering therapies with type 2 diabetes risk, glycaemic traits, and any mediation by BMI, the associations of different LDL-lowering therapies or pathways with gallstone disease risk, the associations of statins with the risk of immune-related diseases, and finally the associations of an emerging lipid target, asialoglycoprotein receptor 1 (ASGR1) inhibitors, with all-cause mortality and a wide range of health outcomes. Where possible, I investigated by ancestry and by sex.

Results: Genetically predicted apolipoprotein B (apoB) or equivalently LDL-cholesterol was positively associated with CAD risk, CVD mortality, and all-cause mortality across its whole distribution, with possibly stronger associations in men than women. Genetically predicted apoB and lipoprotein (a) had little association with type 2 diabetes risk; however, genetically predicted triglycerides were positively associated with type 2 diabetes risk and glycaemic traits specifically in women, partly driven by sex hormones. Genetically predicted BMI mediated more than half of the diabetogenic effects of statins, which did not extend to proprotein convertase subtilisin/kexin type 9 (PCSK9) inhibitors or ezetimibe. Unlike atheroprotective effects, genetically mimicked LDL-lowering therapies or pathways had distinct and opposing associations with gallstone disease risk. Genetically mimicked statins had little association with the risk of immune-related diseases but were positively associated with asthma risk in East Asians. Genetically mimicked ASGR1 inhibitors were inversely associated with all-cause mortality, comparing favorably with current LDL-lowering therapies. Beyond lipid lowering, genetically mimicked ASGR1 inhibitors were positively associated with liver enzymes, erythrocyte traits, insulin-like growth factor 1, and C-reactive protein but were inversely associated with albumin and calcium, which did not generally extend to current LDL-lowering therapies.

Conclusions: This thesis provides genetic evidence that lowering apoB or equivalently LDL-cholesterol reduces CVD morbidity and mortality across its whole distribution. Unlike atheroprotective effects, different LDL-lowering therapies may have distinct side-effects, repurposing opportunities, and thereby overall effects, likely due to different mechanistic pathways.