Abstract TH879: Baseline Metabolomic Profile Predicts 5-Year Multimorbidity Trajectories and Incident Disease Risk in a Community-Dwelling Japanese Cohort

Abstract

Background: The accumulation of multiple chronic diseases (multimorbidity) is a major public health challenge. However, the patterns of disease accumulation over time and their underlying biological drivers remain poorly understood. We aimed to identify distinct multimorbidity trajectories and their associated baseline metabolomic signatures. Methods: We conducted a prospective analysis of 7,441 community-dwelling Japanese adults aged ≥40 years, free of cardiovascular disease and cancer at baseline. Using claims data from 2015 to 2021, we calculated the monthly cumulative Charlson Comorbidity Index (CCI) for each participant. We applied k-means clustering to 5-year CCI data to identify distinct multimorbidity trajectories. Baseline plasma metabolomic profiles were analyzed using ordinal logistic regression to identify metabolites associated with trajectory progression. Subsequently, we used adjusted Cox proportional hazards models to assess the association between these progression-associated metabolites and the risk of future incident diseases over a 5-year follow-up. Results: We identified 3 distinct multimorbidity trajectories among 7,256 participants with complete follow-up: Stable Low (n=6,367), Gradual Increase (n=620), and Rapid Increase (n=269). The Rapid Increase group was characterized by older age and higher baseline CCI. Ordinal logistic regression identified 38 baseline metabolites significantly associated with a higher odds of being in a more progressive trajectory (FDR < 0.05). These metabolites were primarily involved in urea cycle/arginine metabolism and fatty acid metabolism. In disease-specific analyses, several progression-associated metabolites predicted incident disease. For instance, higher baseline levels of Phenylalanine, a top predictor of trajectory progression, were associated with an increased risk of both incident congestive heart failure (HR: 1.30; 95% CI: 1.11-1.52) and chronic pulmonary disease (HR: 1.26; 95% CI: 1.09-1.45). Conversely, higher levels of the medium-chain fatty acid Hexanoate were associated with a reduced risk of congestive heart failure (HR: 0.76; 95% CI: 0.67–0.87). Conclusion: Distinct data-driven multimorbidity trajectories exist and can be predicted by a baseline metabolomic signature. This signature is also linked to the future risk of specific cardiometabolic diseases, suggesting a potential role for metabolomics in the early risk stratification and prevention of multimorbidity.