Metabolomics: A window into the mechanism of lifestyle exposures and metabolic health

Abstract

<strong>Background:</strong> Modifiable risk factors, specifically in this work body composition and habitual diet, are of great importance in epidemiology. The metabolome is a rich resource to explore biological pathways from modifiable risk factors to markers of metabolic health and by extension to health conditions. In three studies, we aimed to identify associations between the metabolome of adolescents and young adults and body composition (study one) or habitual diet (study two). Additionally, we aimed to replicate previous results on associations between the metabolome and markers of metabolic health in adolescents and young adults and test the newly discovered associations as mediators in the relationship of modifiable risk factors and markers of metabolic health (study three). <br /> <strong>Methods:</strong> All three studies were performed in a sub-sample of the DONALD open cohort study using untargeted metabolome measurements of the urine (all studies) and blood metabolome (studies two and three). We used linear and random forest regression (study one), as well as partial least squares regression (study two) to discover new associations between the modifiable risk factors and the metabolome. We utilized linear regression to replicate metabolites associated in two independent studies from our systematic literature review and causal mediation analysis to search for mediators (study three). Where applicable, we held the false discovery rate at 5% to correct for multiple testing. We used imputation to address missing values and stratified all our analyses by sex. <br /> <strong>Results:</strong> We identified 30 metabolites associated with body composition (study one), ten of which were associated with both body mass index (BMI) and body fat percentage (BF). We identified six metabolites as putative biomarkers of habitual dietary intake and replicated one additional marker previously reported by others (study two). We replicated ten metabolites associated with markers of metabolic health and identified no potential mediators (study three). We observed few overlaps between the sexes suggesting strong sex differences in all three studies. <br /> <strong>Conclusion:</strong> Our work demonstrated the great potential of exploring the metabolome as a rich resource and tool in epidemiology. We additionally proposed that sex-specific investigations should be considered in future studies.