Background: In recent years, several biomarkers have been investigated for predicting clinical outcome of patients suffering from heart failure (HF). Previous studies exploring N-terminal pro-brain natriuretic peptide (NT-proBNP) and other biomarkers such as mid-regional pro-adrenomedullin (MR-proADM), carboxy-terminal pro-arginine-vasopressin (CT-proAVP, copeptin) and mid-regional pro-atrial natriuretic peptide (MR-proANP) have provided heterogeneous results and been limited to specific population. A comprehensive comparison of the predictive value of these biomarkers in a large cohort of patients with chronic HF, particularly with regard to different phenotypes of HF is currently missing.

Methods: Data from the MyoVasc study (NCT04064450; N=3,289), a prospective cohort study on chronic HF, were analyzed. Study participants received a highly standardized 5-hour examination in a dedicated study center with deep clinical phenotyping and biosampling.  Information on clinical outcome was derived from structured follow-up with subsequent validation via source data and independent adjudication of clinical endpoints. Humoral vascular and cardiac biomarkers, i.e. proAVP, MR-proADM, and MR-proANP (BRAHMS Kryptor, ThermoFisher Scientific, Germany) and NT-proBNP (Elecsys, Roche, Germany) were measured at baseline. Cox linear regression analyses with adjustment for potential confounders were performed to analyze the predictive value of the biomarkers in patients with symptomatic HF (ie, HF Stage C/D).

Results: The analysis sample comprised 1,741 individuals from  the MyoVasc study with HF stage C/D (mean age: 67.7±9.9 years; female sex: 34.2%). A total of 721 individuals were classified as HF with preserved ejection fraction (HFpEF), 401 with HF with mildly reduced ejection fraction (HFmrEF), and 343 with HF with reduced ejection fraction (HFrEF). In multivariable Cox regression analysis with adjustment for age, sex, clinical profile and medication, the predictive value for worsening of heart failure was highest for NT-proBNP (hazard ratio per standard deviation (HR_SD) 2.16 (95% confidence interval (CI) 1.91/2.35; P<0.01), followed by MR-proANP,  MR-proADM , and CT-proAVP. For all-cause death a similar pattern was observed in the total sample. In sensitivity analyses within HF phenotypes, NT-proBNP showed the highest estimates for predicting worsening of HF as well as mortality in subjects with HFrEF. In contrast, MR-proADM was identified as best predictor for all-cause mortality in both individuals with HFmrEF (HR_SD 2.0 (95%CI 1.53/2.63); P<0.01) and subjects with HFpEF (HR_SD 2.14 (95%CI 1.76/2.61); P<0.01) in multivariable regression analysis compared with gold standard NT-proBNP (HFpEF: HR_SD 1.74 (95%CI 1.37/2.19); P<0.01; HFmrEF: HR_SD 1.80 (95%CI 1.31/2.46); P<0.01), MR-proANP, and proAVP. Further analysis demonstrated that the high predictive value of MR-proADM for all-cause death was driven by its performance in predicting non-cardiac death in individuals with HFpEF and HFmrEF, respectively. For cardiac death, NT-proBNP was the best predictor as compared to the other biomarkers across all HF phenotypes.

Conclusion: The present study demonstrated a high predictive value of MR-proADM for all-cause death in subjects with HFpEF and HFmrEF, which exceeded the performance of NT-proBNP as gold standard. These results might be relevant for the development of individually tailored approaches to risk stratification in chronic HF.