Abstract P498: Different Metabolomic Associations With Mediterranean Style Diet (MSD) Scores Between Individuals With Normoglycemia vs. Impaired Fasting Glucose (IFG) in the Multi-Ethnic Study of Atherosclerosis Transomic Precision Medicine Initiative (MESA-TOPMed) Data

Abstract

Diet-metabolite associations for foods high in monounsaturated fatty acids (MUFAs) differ between those with normo- vs. with dys- glycemia. As metabolites represent the physiological effects of food after it has been digested, processed, and absorbed, this suggests that dysglycemia alters how foods are metabolized and thus their potential health impacts. Our goal was to examine whether metabolites associated with adherence to a MSD, a diet high in MUFA, differed by glycemia status (normoglycemia vs. IFG [125<FG≥110 mg/dL]). We leveraged the longitudinal data on diet and plasma metabolites (N=2,816) in MESA-TOPMed, collected at two exams ~10y apart (baseline (BL): 2000-2002, and follow-up (FU): 2010-2011), on ~3,700 multi-ethnic US adults, ages 44-84y at baseline. Between exams, the proportion of the population with IFG increased from 25% to 40% (P<1.0*10 -16 ), providing a natural experiment to examine how MSD-metabolite associations change alongside population-level glycemia changes. We interpreted linear models which specified metabolites values as the outcome, and MSD score as the predictor, and adjusted for demographic (age, gender, race/ethnicity, income level, highest education level) and health behavior (physical activity, energy intake, smoking status, body mass index [BMI]) information, with longitudinal models also including time as a fixed effect (days since BL). Longitudinal random-intercept multi-level models identified 789 compounds associated with MSD at metabolome wide levels of significance (P<1.8*10 -5 ). Significant diet-by-time interactions (P<.05) indicated that the 241 metabolites showed associated with MSD that were time-specific, changing as population dysglycemia increased (alongside changes in other characteristics), and we observed different effect sizes for MSD-metabolite associations in those with IFG vs those with normoglycemia (t = 3.4, df = 145, P = 9.0*10 -4 ). 548 metabolites showed associations with an MSD that were time-invariant, and when analyses were stratified by glycemia status, more of these more ‘universally associated’ metabolites were significantly associated with an MSD in those with IFG and normoglycemia, vs. in just one dysglycemia group (45% vs. 15%; X 2 =117, df=2, P<2.0*10 -16 ). Our longitudinal analyses demonstrated that population-level metabolite associations with an MSD change as the population dysglycemia changes. Of course, other characteristics changed as well, and only some of these (e.g., BMI and age) were controlled for in our analyses. However, our cross-sectional analyses provided additional evidence that MSD-metabolite associations differ between those with normoglycemia vs. IFG. We are now seeking replication in other TOPMed cohorts, to support that notion that tailoring dietary advice to T2D prevention vs. management would be an important direction for future precision nutrition efforts.