Abstract 42: Metabolite Risk Scores of Lifestyle Measures Improve Incident Hypertension Prediction

Abstract

Introduction: Metabolites capture the metabolic environment associated with exposures. Objective: To develop and assess the usefulness of metabolite risk scores (MRSs) as biomarkers representing exposure measures. Methods: We used two separate sets of individuals sampled from the Hispanic Community Health Study/Study of Latinos (HCHS/SOL). Both sets of individuals had metabolite data measures on the Metabolon HD4 platform. We applied the least absolute selection and shrinkage operator (LASSO) to develop MRSs for current smoking, physical activity, Mediterranean diet, and anxiety in set 1 (n=2,178), cross-sectionally, and evaluated them for replication in set 2 individuals (n=4,002). We defined hypertension as SBP140, DBP90, or hypertension medication use, and estimated MRS associations with incident hypertension longitudinally (6 years after the baseline exam, on average) in set 2 individuals free of hypertension at baseline (426 cases and 1,556 controls). The association models were adjusted for age, sex, and BMI. We compared the estimated association with those of self-reported phenotypes. Results: All MRSs were developed in set 1 individuals have replicated associations in set 2 individuals. The figure provides the estimated association of MRSs for diet, smoking, physical activity, and anxiety with incident hypertension and compares them to the associations of self-reported Mediterranean diet, anxiety measured using the Spielberger trait anxiety scale, pack-years of smoking, and self-reported physical activity in MET-min/day. All associations are per 1 SD increase in the exposure measure. The figure shows that the MRS associations are often stronger than those of the self-reported measure. Conclusions: MRSs have the potential to serve as accurate biomarkers of self-reported exposures.