Abstract
OBJECTIVES: Metabolomic markers may provide deeper insight into the underlying biology of frailty. Yet, little data exist characterizing metabolomic correlates of frailty. The objective of this cross-sectional study was to determine the association of serum metabolites with frailty. METHODS: We evaluated 2187 participants from the Framingham Offspring Study (mean age; 62y; 53% women) with metabolomic and frailty data. Metabolites (n = 217) were quantified using liquid chromatography/tandem mass spectrometry (1990–95). Frailty was assessed using Fried's frailty phenotype (1998–01). Multivariable logistic regression models were used to relate individual metabolites with odds of frailty. In 1,815 participants, we created a metabolomic signature: sum of (serum metabolite concentration*β coefficient). Metabolites were included based on the strength of their individual association with frailty (FDR ≤ 0.05). Logistic regression was used to relate the metabolomic signature with odds of frailty. C-statistics determined if the metabolomic signature improved the prediction of frailty beyond a standard model. Models were adjusted for age, sex, BMI, and smoking. RESULTS: Adjusting for age and sex, 24 metabolites were significantly associated with frailty (22 positive and 2 negative; all FDR ≤ 0.05). A majority of metabolites were lipids (n = 19) of the phosphatidylcholine (n = 7) and triacylglycerol (n = 10) sub-classes with high degree of saturation (n = 13 with <2 carbon double-bonds). Additional metabolites have roles in amino acid and cellular metabolism (lactate, asparagine, phosphocreatine, glycerol, and isocitrate). In an age and sex-adjusted model, a weighted signature of the 24 metabolites was associated with increased odds of frailty (OR: 2.0, CI: 1.5, 2.7). C-statistics to predict frailty improved beyond a standard model by the addition of the weighted metabolomic signature (C-statistics: 0.70 vs 0.61). Results were similar with further adjustment for BMI and smoking. CONCLUSIONS: Frailty was associated with 24 distinct metabolites and with a metabolomic signature in this study of older adults. These findings may inform future investigations on the progression of frailty and warrant replication in independent populations. FUNDING SOURCES: AHA and Boston Pepper Center OAIC