Longitudinal mapping of the blood proteome to blood metabolome reveals the role of the protein-metabolite axes in metabolic health

The proteomic determinants of blood metabolites have rarely been systematically investigated in population-based studies. Identifying protein-metabolite pairs could help elucidate the role of the protein-metabolite axis in modulating metabolic health. This study was based on a longitudinal cohort of 485 middle-aged and elderly adults (age: 56.9 ± 4.5) with repeated-measured serum proteome data from 3 cohort visits ( N proteome =1455; 411 proteins) and repeated-measured serum metabolome data from 4 cohort visits over 12.3 years ( N metabolome =1940; 196 metabolites). The participants were divided into discovery and validation sets ( N = 402 and 83, respectively). Longitudinal prospective associations between each pairwise combination of proteins and metabolites were examined by a linear mixed-effects model. We used a linear mixed-effects model and logistic regression to investigate the associations of proteins and metabolites in the identified protein-metabolite pairs with 13 metabolic traits and 4 metabolic diseases (type 2 diabetes, metabolic syndrome, hypertension, and obesity), respectively. Mediation analysis was used to explore the mediating effects of metabolites on the protein-metabolic trait/disease associations. We identified 53 longitudinal prospective associations between 28 proteins and 34 metabolites. We then identified significant associations of these proteins and metabolites with metabolic traits and diseases, including 91 protein-metabolic trait associations, 44 protein-metabolic disease associations, 104 metabolite-metabolic trait associations, and 7 metabolite-metabolic disease associations. We further identified 17 protein-metabolite-metabolic trait/disease pathways. We present the first population-based study systematically investigating the longitudinal prospective associations between serum proteins and metabolites. The identified protein-metabolite axes may serve as potential targets for intervention to enhance metabolic health.