A plasma metabolite profile predicts worsening of heart failure and reflects distinct inflammatory mechanisms in symptomatic heart failure
E Yapici, S Zeid, R Martens, S Nguyen, T Koeck, V Ten Cate, T Schmidlin, P WildAbstract
Background
Systemic inflammation drives heart failure (HF) development and progression, with mounting evidence indicating that metabolites act as active signalling molecules that modulate inflammatory responses and protein expression.
Aim
To identify a plasma metabolite signature linked to worsening of HF (WoHF), to characterise its relationship with systemic inflammation, and to examine its longitudinal effects on inflammatory protein profiles in symptomatic HF.
Methods
In the Myovasc cohort (N=3,289 ), untargeted metabolite profiling by liquid chromatography-mass spectrometry and inflammatory protein profiling by affinity proteomics (Olink Target96 Inflammation panel) were performed. Elastic net Cox regression with 10-fold cross-validation was used to select WoHF-linked metabolites, from which a weighted signature was constructed. Cumulative incidence was assessed by quartiles of the signature, and the prognostic value of the signature was further assessed by Cox regression with interaction terms to account for HF phenotype-specific variation. Associations with C-reactive protein, the lymphocyte/monocyte ratio (LMR), and the neutrophil/lymphocyte ratio (NLR) were assessed to evaluate links with systemic inflammation and immune balance. Longitudinal associations between baseline metabolites and 2-year changes in protein were analysed using linear regression stratified by HF phenotype with Benjamini–Hochberg correction.
Results
Among 1,664 symptomatic HF participants [mean age 68.0 ± 9.9; 32.5% women], 22 metabolites were identified (CV C-index=0.69; 297 WoHF events). Ten metabolites, including N-acetylneuraminate, nudifloramide, and TMAO, conferred increased risk; 12, including L-arginine, nicotinamide, and N-methylarginine, conferred decreased risk of WoHF. Participants with a higher WoHF metabolite signature experienced more WoHF, cardiac death, and mortality (p <0.0001). Cox regression analysis showed that the WoHF metabolite signature was prognostic for worsening of HF independent of age, sex, and cardiovascular risk factors [Hazard ratio 1.99 (95% CI 1.79-2.20), p < 0.0001]. Three metabolites showed phenotype-dependent prognostic value with significantly stronger hazard associations in HFpEF relative to HFrEF [1-methylhistidine (pint = 0.037), 1-methyladenosine (pint = 0.008), and 5-hydroxyindoleacetate (pint = 0.021)]. WoHF metabolites were associated with increased C-reactive protein and NLR, and decreased LMR, whereas protective metabolites showed the opposite pattern (all p < 0.05). Finally, WoHF metabolites predicted longitudinal changes in 40 inflammatory proteins (all FDR p <0.05) over two years of follow-up across HF phenotypes.
Conclusion
A plasma metabolite signature robustly predicts WoHF and reflects underlying inflammatory dysregulation, with effects on longitudinal inflammatory protein trajectories, identifying novel biomarkers and potential therapeutic targets in HF.