Population-based study of plasma metabolites as mediators of the relationship between socioeconomic status and incident diabetes: A social metabolomic approach to health disparities

Type 2 Diabetes (T2D) is a major cause of morbidity and mortality that disproportionately affects minorities and people experiencing socioeconomic disadvantage (SED). While SED is an established risk factor for T2D, the underlying mechanisms are poorly understood. Studies of comprehensively measured metabolites (metabolomics) can improve understandings of this relationship.The dissertation aims were to (1) identify associations between SED, measured as a dichotomized composite score of five indicators: education, income, financial hardship, employment, and wealth, and 7,522 metabolite peaks via linear regression and (2) evaluate those peaks as mediators of the relationship between SED and 10-diabetes incidence using causal mediation methods. Orthogonal partial least squares discriminant analyses (OPLS-DA) were used to evaluate if the metabolites could distinguish the two SED groups. We also aimed to evaluate metabolic pathways associated with SED that may mediate the SED-T2D relationship. Given the potential for race and sex differences, we conducted the analyses in full and stratified samples.Using data on 3,350 Black and White participants from the U.S.-based prospective Coronary Artery Risk Development in Young Adults (CARDIA) study, we found that nearly 40% of the 7,522 metabolite peaks were significantly associated with SED and identified six significant pathways involved in carbohydrate, amino acid, and lipid metabolism. We found that metabolites could explain between 13-20% of the variance in SED and could be used to distinguish our two SED groups with moderate accuracy (AUC: 70.5) in the full sample. We observed race/sex subgroup differences in the bile acid, steroid, and linoleic acid pathways. Approximately 10% of the aim-2 sample (n=2,554) were incident diabetes cases between CARDIA exam years 20-30. The study found that 122 metabolite peaks, including D-threonic acid and 7-ketocholesterol, significantly mediated the SED-diabetes incidence association. Pathway analysis identified primary bile acid and arginine biosynthesis as significant pathways linking SED and diabetes incidence. Our results highlight the potential of untargeted metabolomics to identify peaks and pathways linking SED to T2D. These studies provide a foundation for future research to investigate these pathways further and develop interventions to reduce the burden of chronic disease in disadvantaged populations.