Dietary Patterns, Metabolomics and Frailty in a Large Cohort of 120 000 Participants.

Objective: To examine the associations of dietary patterns with frailty and whether metabolic signatures (MSs) mediate these associations. Methods: We used UK Biobank data to examine (1) the associations of four dietary patterns (i.e., alternate Mediterranean diet [aMED], Recommended Food Score [RFS], Dietary Approaches to Stop Hypertension [DASH] and Mediterranean-DASH Intervention for Neurodegenerative Delay [MIND] diet) with frailty (measured by the frailty phenotype and the frailty index) using multivariable logistic regression (analytic sample 1: N = 124 261; mean age = 57.7 years), and (2) the mediating role of MSs (weighted sums of the metabolites selected from 168 plasma metabolites using the LASSO algorithm) in the above associations via mediation analysis (analytic sample 2: N = 26 270; mean age = 57.7 years). Results: Four dietary patterns were independently associated with frailty (all P < 0.001). For instance, compared to participants in the lowest tertile for RFS, those in the intermediate (odds ratio [OR]: 0.81; 95% confidence interval [CI]: 0.74, 0.89) and highest (OR: 0.62; 95% CI: 0.56, 0.68) tertiles had a lower risk of frailty. We found that 98, 68, 123 and 75 metabolites were associated with aMED, RFS, DASH and MIND, respectively, including 16 common metabolites (e.g., fatty acids, lipoproteins, acetate and glycoprotein acetyls). The MSs based on these metabolites partially mediated the association of the four dietary patterns with frailty, with the mediation proportion ranging from 26.52% to 45.83%. The results were robust when using another frailty measure, the frailty index. Conclusions: The four dietary patterns were associated with frailty, and these associations were partially mediated by MSs. Adherence to healthy dietary patterns may potentially reduce frailty development by modulating metabolites.