Untargeted Metabolomics and Cognitive Function in the Boston Puerto Rican Health Study (OR32-03-19)

Abstract

ObjectivesWe sought to determine the metabolic signature associated with poor vs good cognition in a cohort of Boston-area Puerto Ricans. MethodsWe studied 638 participants the Boston Puerto Rican Health Study (BRPHS), a prospective Boston-area cohort study of Puerto Ricans, with comprehensive metabolite profiling (>700 metabolites) at baseline and a concurrent assessment of global cognitive function. Cognition was assessed at baseline using a comprehensive cognitive battery from which a global cognitive score was derived.Our analysis focused on 638 annotated metabolites that had reported values in ≥80% of the participants. We performed univariate linear regression analyses to identify metabolites significantly associated with global cognitive score. To examine whether these metabolites could then discriminate between the low from (<=1SD below the mean, N = 85) the high (>=1SD above the mean, N = 99) cognition groups, and to identify metabolites that could best discriminate between low and high cognition groups, we conducted partial least square discrimination analysis (PLS-DA). We used separation distance as the test statistic and plotted the scores of the first two partial least square components to inspect the separation of cases from controls. We also identified metabolites contributing most strongly to the separation (VIP > 1). Results113 metabolites were significantly associated with cognitive function in univariate analyses. In PLS-DA analyses using these 113 metabolites. when plotting the first two PLS components, there was a clear separation between the low and high cognition groups and controls (Figure 1) (permutation test based on separation distance p-value 0.012) (Figure 1). 23 metabolites had a Variable Importance in Projection (VIP) scores > 1. The metabolites with the highest VIP scores and elevated in the low relative to the high cognition group included cystathionine (amino acid and downstream metabolite of homocysteine), quinate (xenobiotic/food component, component medication used to treat malaria) and p-cresol-glucuronide (amino acid). (Table 1). ConclusionsIn this cross sectional study of Boston Area Puerto Ricans, we observed a separation between high and low cognitive function groups in the Boston Puerto Rican study based on metabolomic profiles. Funding SourcesThis work was supported by the National Institutes of Health. Supporting Tables, Images and/or Graphs▪▪