Abstract MP03: Metabolite Signature of Ultra-processed Foods Intake and Cardiovascular Morbidity and Mortality Among Low-income Black and White Americans

Background: High ultra-processed foods (UPF) intake may increase risks of CVD morbidity and mortality, but the mechanisms are unclear, and there is no biomarker for UPF intake. We aimed to identify UPF-related circulating metabolite signature (UPFsig) and evaluate its associations with incident coronary heart disease (CHD) and CVD mortality in a cohort of mostly low-income Black and White Americans. Methods: Untargeted profiling of baseline plasma metabolites was conducted in two nested case-control studies of incident CHD (n=1023) and incident prostate cancer (n=665). An 89-item food frequency questionnaire was administered at baseline (2002-2009). UPF intake (classified by Nova) was calculated as % of weight per day. The associations of UPF intake with 1,100 metabolites were evaluated by linear regression, adjusting for fasting status, batch, age, sex, race, education, income, smoking, alcohol drinking, physical activity, sitting hours, daily calories, BMI, and history of diabetes, hypertension, and hypercholesterolemia. Metabolites with P <0.05 were selected by elastic net with repeated 10-fold cross-validation to construct the UPFsig. Conditional logistic and Cox regression were used to examine the associations of UPFsig with incident CHD and CVD mortality, respectively. Results: Of 1,688 participants, 70% were men, 61% were Black, and the mean baseline age was 56.5 (SD:12.0) years. The mean intake of UPF was 41.1% (SD:15.3). A total of 126 metabolites were selected to construct UPFsig (correlation r=0.51, P <0.0001). The UPFsig was associated with incident CHD (adjusted OR [95%CI] per SD increase =1.43 [1.20-1.69]), particularly among Black (1.73 [1.35-2.22]) compared to White participants (1.17 [0.93-1.47], P interaction =0.01). Additionally, during a mean 13.4-year follow-up, 285 CVD deaths were noted. The UPFsig was associated with CVD mortality (adjusted HR [95%CI] per SD increase =1.22 [1.07-1.39]), even after adjusting for UPF score (1.17 [1.01-1.36]). The adverse associations were largely consistent across participants of varied sociodemographics, lifestyles, and metabolic disease status. Conclusions: We identified a metabolite signature of UPF and demonstrated its significant associations with CVD morbidity and mortality. The UPFsig may serve as a biomarker and provide insight into mechanisms of UPF intake on CVD. The racial difference in the association of UPFsig with incident CHD warrants further research.