Abstract MP45: A Metabolome-wide Association Study of Plant Food Consumption With Blood Pressure

Background: Epidemiological evidence links blood pressure (BP) -lowering effects with fruit, vegetables, pulses, nuts but results are inconsistent and often rely on self-administered dietary assessment methods prone to random and systematic errors and potential under- or overestimation of relationships. Objective: To characterize urinary metabolic signatures of plant food intake (fruit, vegetables, pulses, and nuts) and subsequently to assess their associations with BP, validated across cohorts. Methods: We used untargeted metabolic profiling obtained by proton nuclear magnetic resonance spectroscopy ( 1 H NMR) of two timed repeated 24-hour urine collections derived from 2,032 free-living US participants, and replicated among 449 UK participants, all part of the cross-sectional International Study of Macro- and Micronutrients and Blood Pressure (INTERMAP) study. Associations of plant foods intakes from four 24-hour dietary recalls with 7,100 NMR features were evlauted using univariate linear regression analyses. Significant metabolites were structurally identified using analytical and statistical approaches. We compared extensively adjusted associations of plant food intake and urinary metabolites with BP in the total population (N=2,481). All analyses were extensively adjusted for potential confounders and accounted for multiple testing. Results: We found consistent, reproducible relationships of NMR signatures with plant food intake across repeated urine collections and between populations. Eleven urinary metabolites were identified and two remained unknown (1: δ 2.80 (s); 2: δ 1.16(s), 1.67 (s)). Twelve metabolites were directly correlated with plant foods ranging from 0.15 (P≤1x10 -11 ) to 0.49 (P≤5x10 -119 ), and phenylacetylglutamine was inversely associated (r=-0.18, P≤1x10 -5 ). In adjusted analyses, plant food intake was inversely associated with systolic BP (-1.29 mmHg; 95%CI: -2.38,-0.19), but not with diastolic BP. Metabolites inversely associated with systolic BP were hippurate (-2.67 mmHg;-3.71,-1.63), citrate (-2.50 mmHg; -3.55,-1.44), phenylacetylglutamine (-2.26 mmHg; -3.29,-1.23), unknown1 (-2.05 mmHg; -3.06,-1.04), 4-hydroxyhippurate (-1.45 mmHg; -2.44,-0.45), and unknown2 (-1.09 mmHg; -2.09,-0.09); with diastolic BP were phenylacetylglutamine (-1.12 mmHg; -1.84,-0.40), citrate (-1.02 mmHg; -1.76,-0.28), and unknown1 (-1.18 mmHg; -1.88,-0.47). Plant food-systolic BP associations were attenuated by further adjustment for citrate, hippurate, and unknown1, while hydroxyproline, proline and phenylacetylglutamine were potential effect modifiers. Conclusion: Urinary metabolites were consistently associated with plant food intake. Urinary metabolites that may underlie the plant food-BP association were gut-microbial co-metabolites and tricarboxylic acid cycle intermediates.