Metabolomics biomarkers for diet and adiposity

Abstract

Diet is an important determinant of health and well-being. Epidemiologic studies link higher habitual intake of sweetened beverages (SBs), sweet and fatty snacks, and the broad range of ultra-processed foods (UPF) with an increased risk of obesity. Poor nutritional profiles, higher caloric intake, and energy imbalance are some of the proposed mechanisms, but other biological pathways underlying diet-related weight gain and regulation are not fully defined. Human metabolome provides a rich resource for understanding metabolic alterations associated with diet. Here, we conduct a literature review on biomarkers of SBs (study 1); investigate the metabolomic signatures of SBs and added sugar intake in children, adolescents, and young adults and their association with adiposity measures (study 2); investigate the reproducibility of urine biomarkers of sweet and fatty snacks across two independent cohorts (study 3); and investigate the metabolomic profiles of UPF intake in adolescents and young adults and their association with adiposity (study 4). <br/>

In study 1, we conducted a systematic review of the literature on biomarkers of SBs and their levels of validity. In study 2, we used 3 data sets across 3 age groups: children (3.0–10.3 y), adolescents (14.9–18.4 y), and young adults (18.0–21.9 y), from the DONALD cohort study. In study 3, we included the previously defined sample of children and children from an external cohort, the IDEFICS/I.Family cohort. In study 4, we included the adolescent and young adult analytic samples defined in study 2. We used untargeted metabolomics in urine and plasma across all studies and additionally conducted lipidomics on plasma. We applied multiple machine learning methods because of the high-dimensional data: the random forest, partial least squares, and LASSO for joint metabolite selection (study 2 and 3); particle swarm optimization and extreme gradient boosting for investigating metabolite data missing mechanisms (study 4); and robust sparse PCA for deriving metabolite patterns (study 4). We used linear and mixed effects for covariate adjustments (study 2-4). <br/>

We identified metabolomic signatures of SBs, added sugar, sweet and fatty snacks, and UPF intake in young individuals. Some of these metabolomic changes were related to adiposity measures and may be important research targets for better understanding of the mechanisms through which these foods contribute to weight gain and adiposity.