Abstract
AIMS: There is an increasing proportion of hospitalized heart failure (HF) patients classified as HF with preserved ejection fraction (HFpEF) around the world. Growth differentiation factor 15 (GDF-15) is a promising biomarker in HFpEF prognostication; however, the majority of the existing data has been derived from the research on undifferentiated HF, whereas the studies focusing on HFpEF are still limited. This study aimed to determine the prognostic power of GDF-15 in the hospitalized patients with HFpEF in a Chinese cohort. METHODS AND RESULTS: We analysed the levels of serum GDF-15 in 380 patients hospitalized for acute onset of HFpEF measured by heart ultrasound at admission in a prospective cohort. The associations of GDF-15 with 1 year risk of all-cause death and 1 year HF readmission were assessed by the Cox proportional hazards model. Area under the receiver operating characteristic curves was used to compare predictive accuracy. GDF-15 was strongly correlated with 1 year HF readmission and 1 year all-cause death, with event rates of 24.8%, 40.0%, and 50.0% for 1 year HF readmission (P < 0.001), respectively, and with 11.2%, 13.6%, and 24.6% for 1 year all-cause death (P = 0.004) in the corresponding tertile, respectively. In the multivariate linear regression model, GDF-15 had a significantly negative correlation with haemoglobin (P = 0.01) and a positive correlation with creatinine (P = 0.01), alanine transaminase (P = 0.001), and therapy of aldosterone antagonist (P = 0.018). The univariate Cox regression model of GDF-15 showed that c-statistic was 0.632 for 1 year HF readmission and 0.644 for 1 year all-cause death, which were superior to the N-terminal pro-brain natriuretic peptide (NT-proBNP) model with c-statistics of 0.595 and 0.610, respectively. In the multivariable Cox regression model, GDF-15 tertiles independently predicted 1 year HF readmission (hazard ratio 2.25, 95% confidence interval: 1.43-3.54, P < 0.001) after adjusting for baseline Acute Study of Clinical Effectiveness of Nesiritide in Decompensated Heart Failure (ASCEND-HF) risk score, history of HF, NT-proBNP, and high-sensitivity cardiac troponin T. Compared with the model including all the adjusted variables, the model with the addition of GDF-15 improved predictive power, with c-statistic increasing from 0.643 to 0.657 for 1 year HF readmission and from 0.638 to 0.660 for 1 year all-cause death. CONCLUSIONS: In hospitalized patients with HFpEF, GDF-15 measured within 48 h of admission is a strong independent biomarker for 1 year HF readmission and even better than NT-proBNP. GDF-15 combined with the traditional indicators provided incremental prognostic value in this population.