Optimizing Exposome-wide Assessments in Cardiometabolic Risk

Sammanfattning: This thesis is focused on cardiovascular disease (CVD) and type 2 diabetes mellitus (T2D), two concomitantconditions that appear with growing concern. In our work, we aim to improve the identification of individuals at-riskof cardiometabolic disease through the characterization of complex environmental exposures (i.e. diet, physicalactivity), that temporally vary, and the health effects on cardiometabolic traits and disease. Our projects werebased upon the Västerbotten Health Survey (VHU) and the Malmö Diet and Cancer (MDCS) studies, whichincluded extensive data on lifestyle, biological intermediates, and clinical outcomes.In Paper I, we utilized the so-called environmental-wide association approach (EWAS), using longitudinal datafrom > 31,000 adults in VHU study. Under generalized linear models, from ~ 300 candidate exposures, 11modifiable variables were associated with most of the cardiometabolic traits; the prioritised variables belonged tosmoking, coffee intake, physical activity, alcohol intake, and context-specific lifestyle domains.In Paper II, we implemented a machine learning-based model to identify individuals with variable susceptibility tolifestyle risk factors for T2D and CVD. Individuals with sensitivity to blood lipids, and blood pressure associatedpredictors were at higher risk to develop cardiometabolic disease. Furthermore, when pooling across sensitivegroups from the two cohorts, the findings suggest a particular vulnerable subpopulation with different risk profile.In Paper III, a series of causal-inference experiments from VHU and publicly available genome-wide associationstudy (GWAS) summary statistics were used to triangulate evidence of the direct and mediated effects byadiposity and physical activity, of macronutrient intake (fat, carbohydrates, protein and sugar) and cardiometabolicdisease. Using structural equation modelling, the mediation analyses enhanced with Mendelian randomizationanalysis, showed a likely causal putative association between carbohydrate intake and T2D. In addition, theintegrative genomic analyses suggested a candidate causal variant localized to the established T2D geneTCF7L2.In Paper IV, we conducted a systematic review and metanalysis of observational studies, complemented byMendelian randomization analysis using GWAS summary statistics, investigating causal associations ofindividuals with high, yet normal, glycaemia associated with cardiovascular complications. Prediabetes was likelycausally associated with coronary heart disease; suggesting higher, but not diabetic levels of blood glucose confera risk, thus, effective preventive strategies may prove successful in prediabetes.