Sarcopenia and Its Associated Metabolic Profile Predict Incident Heart Failure: A Prospective Cohort Study of 267 335 Adults in the UK Biobank

Background Sarcopenia is prevalent in heart failure (HF), but its role in incident HF and underlying metabolic mechanisms remains unclear. We examined the interplay between sarcopenia phenotypes, circulating metabolic profiles, and incident HF. Methods We analyzed UK Biobank participants without baseline HF. Associations between sarcopenia phenotypes and HF incidence were assessed using Cox regression. Nuclear magnetic resonance metabolomics was used to characterize sarcopenia‐related profiles. Cox regression was applied to test metabolite–HF associations, and least absolute shrinkage and selection operator regression was further used to refine predictors. Incremental predictive value beyond clinical risk factors was evaluated using discrimination and reclassification metrics. Results During a median 15.3 years, 10 233 of 267 335 participants (mean age 56.5 ± 8.1 years; 44.6% men) developed HF. Groups with confirmed sarcopenia (n=1993) and low handgrip strength (normalized to body mass index) only (n=18 796) were associated with higher HF risk (hazard ratio [HR], 1.63 [95% CI, 1.44–1.85]; HR, 1.76 [95% CI, 1.66–1.85]) compared with the reference group, with stronger effects observed in younger adults and women. Metabolomic profiling revealed sarcopenia‐related alterations (higher glycoprotein acetyls, glucose–lactate, phenylalanine, tyrosine, 3‐hydroxybutyrate; lower omega‐3 fatty acids, docosahexaenoic acid, glycine, glutamine, histidine), which also predicted higher HF risk (Bonferroni‐adjusted P < 0.05). Selected metabolites are significant mediators and modestly improved HF prediction (15‐year net reclassification improvement 14%–15%; integrated discrimination improvement 0.9%–1.0%). Conclusions Sarcopenia, particularly reduced handgrip strength, was a strong predictor for incident HF. Lipid‐, amino acid‐, and energy metabolism–related alterations modestly improved risk prediction and partially mediated the association.