Diabetes is associated with a distinct proteomic profile in acute heart failure patients

Introduction

Diabetes mellitus is common among patients with heart failure (HF) and is strongly associated with a worsened prognosis. While diabetes is known to influence multiple biological pathways, the circulating proteomic profile associated with diabetes in acute HF, independent of renal function and HF severity, remains incompletely characterised.

Aims

To identify circulating proteins associated with diabetes in patients with acute HF using targeted proteomics.

Methods

Analyses were carried out in a hospitalized HF cohort consisting of both de-novo HF and acute-on-chronic HF regardless of ejection fraction, with available targeted proteomics, consisting of 92 proteins associated with cardiovascular disease, inflammation and metabolism. Analyses were restricted to individuals with complete proteomic measurements and non-missing covariates. Protein concentrations were analysed on the NPX scale (log2). For each protein, linear regression models were fitted with protein NPX as the dependent variable and diabetes as the main exposure, adjusted for age, sex, body mass index (BMI), estimated glomerular filtration rate (eGFR), and ln-transformed BNP. Missing values in BMI, eGFR, and BNP were handled using single median imputation within the study population. False discovery rate (FDR 10%) correction was applied across proteins in the fully adjusted model. Effect sizes are reported as adjusted differences in log2 NPX for diabetes, corresponding to fold changes of 2^β.

Results

The analysed population consisted of 315 patients, of whom 117 (37.1%) had diabetes. Median age was 77 years (interquartile range (IQR) 67.5–83), and 96 patients (30.5%) were women. Median BMI was 26.9 kg/m² (IQR 23.9–31.0), median eGFR was 45 mL/min/1.73 m² (IQR 32–59), and median NT-proBNP was 4274 pg/mL (IQR 2278–8887).

In the fully adjusted, FDR-controlled analysis, 21 proteins were associated with diabetes at q < 0.10 (Figure 1). Of these, 20 proteins showed higher circulating levels in patients with diabetes, while one protein (CPA1) showed lower levels. The strongest associations, based on the lowest q-values, included Gal4, GDF15, IL1RT1, TRAP, and TNFRSF14. Across significant proteins, adjusted differences ranged from +0.105 to +0.381 log2 NPX (approximately 7.5% to 30% higher levels in diabetes), whereas CPA1 was lower in diabetes (β −0.333 log2 NPX; approximately 21% lower) (Figure 2).

Conclusion

In hospitalized patients with heart failure, diabetes was associated with a distinct circulating proteomic profile independent of age, sex, BMI, renal function, and NT-proBNP. The diabetes-associated signal was dominated by higher levels of markers related to inflammatory, vascular, and TNF receptor-associated pathways. These findings support diabetes as a major systemic modifier of circulating biology in acute HF and motivate further investigation of how this proteomic signature relates to HF phenotype and prognosis.Fig 1.Volcano plotFor image description, please refer to the figure legend and surrounding text.Fig 2.Forest plotFor image description, please refer to the figure legend and surrounding text.