Metabolomic Analyses Reveal Mechanistic Links Between a Mediterranean-Style Diet and Incident Type 2 Diabetes in the Multi-ethnic Study of Atherosclerosis (MESA)

Objectives: To identify metabolites that link intake of a Mediterranean-style diet (MSD) with protection from type 2 diabetes (T2D). Methods: Longitudinal data were included from ∼2,700 men and women (50% female), who were ages 45-90 at baseline, from up to six exams conducted at ∼18-24 month intervals. Global metabolomic profiles, comprising 2,761 compounds after quality control procedures, were quantified in stored plasma collected at baseline (2000-2002), at exam 5 (2010-2011), and at exam 6 (2016-2018). Habitual diet over the past 12 months was assessed using a self-report food frequency questionnaire at baseline and at exam 5, and used to construct a summary score representing adherence to an MSD. Incident T2D (n=386 cases), defined using ADA 2003 criteria, was collected alongside covariates at each exam (exams 1-6). Multi-level linear regression models examined associations between metabolites and MSD over time. Cox-proportional hazard models modelled the association of metabolites and incident T2D, with data right censored at the first T2D presentation. All models controlled for age, income level, and physical activity (time-dependent), and highest education level, gender, and race/ethnicity (time-invariant). Models with MSD additionally controlled for daily caloric intake (time dependent). Significance was set using a Bonferroni correction. Results: MSD and incident T2D were both associated with 129 compounds (P< 1.87*10-5), of which 9 had annotations: 7 were inversely associated with MSD and with increased incidence of T2D: Aspartic acid (HR: 1.34; 1.21-.49), D-gluconic acid (1.43; 1.30-1.57), glutamic acid (1.36; 1.25-1.49), indoxyl-sulfate (1.23; 1.12-1.36), lactoylphenylalanine (1.42; 1.29-1.57), N-lactoyl-tyrosine (1.27; 1.17-1.38) and pyruvic acid (1.27; 1.17-1.38); while 2 were positively associated with MSD and with protection from T2D: Keto-methylvalerate (0.80; 0.73-0.88) and succinic acid (0.67; 0.59-0.77). Conclusions: The shared metabolites identified in this study indicate that the protection from T2D associated with an MSD arises, in part, via metabolic pathways related to shifts in energy metabolism from protein to amino acid metabolism, and behavioral pathways related to appetite control. Future analyses should seek to replicate these observational associations in designs that test for causality. Funding Sources: NIH.