Protein signatures associated with methylation patterns of heart failure phenotypes

Abstract Introduction The heart failure (HF) syndrome is highly susceptible to the exposome. The interplay between DNA methylation (DNAm), surrogate for exposomal influences, and the plasma proteome in the progression and development of phenotypes of heart failure is unclear. We investigated the relation between methylation patterns of HF phenotypes and proteins signatures. Methods Individuals with symptomatic HF (stage C/D with HFpEF, HFmrEF and HFrEF) of the MyoVasc cohort (N=3,289) were investigated. EDTA-Plasma and DNA were extracted from peripheral blood and stored in -80°C. DNAm was measured using the Infinium MethylationEPIC v2.0 methylation array. Concentrations for 538 proteins were measured using an immuno-qPCR assay in EDTA plasma. Multivariable regression models adjusted for age and sex were used to assess the relationship between DNAm signatures and individual proteins in each phenotype. Ridge-regularized linear regression was used to generate a phenotype-specific protein-signature in relation to each methylation signature, linear regression to estimate links between protein signatures and cardiovascular traits and Cox regression relate protein signatures to clinical outcome. Results Data from 1,317 individuals with symptomatic HF were analyzed. After FDR adjustment, 310 proteins were significantly related to HFpEF methylation signature, 314 to HFmrEF and 266 to HFrEF. Of associated proteins, 49.9% (198) were found in all phenotypes, 8.82% (35) only in HFpEF, 12.1% (48) in HFmrEF and 4.79% (19) in HFrEF. Tumor necrosis, arterial hypertension and T-cell regulation pathways were enriched in all three phenotypes. NT-proBNP had the strongest association with HFrEF (β=0.95 [0.82-1.08], p<0.0001) followed by presence of plaques in carotid artery and C-reactive protein, while left ventricular mass (LVM) had stronger association with HFpEF (β=0.22 [0.11-0.33], p=0.0001). All three signatures were associated with worsening of HF (HFpEF: HR=1.93 [1.61-2.30], p<0.0001; HFmrEF: HR=1.63 [1.29-2.04], p<0.0001; HFrEF: HR=1.64 [1.38-1.95], p<0.0001) and all-cause death in HF (HFpEF: HR=1.93 [1.65-2.26], p<0.0001; HFmrEF: HR=1.95 [1.58-2.42], p<0.0001; HFrEF: HR=2.24 [1.86-2.69], p<0.0001). Methylation-associated protein signatures had stronger associations with clinical outcome than the methylation signatures themselves (p < 0.01). Conclusion This study revealed protein signatures associated with methylation patterns of HF phenotypes that shared common and unique features for HF phenotypes. Important common pathways were enriched in tumour necrosis, arterial hypertension and regulation of T cells. However, important differences between phenotypes were also revealed, e.g. NT-proBNP was strongly associated with the HFrEF protein signature and LVM with the HFpEF signature. The methylation-associated protein signatures emphasise the possible exposomal influence on HF syndrome.