Plasma biomarkers to predict or rule out early post-discharge events after hospitalization for acute heart failure.
Improved prediction of early post-discharge death or rehospitalization after admission for acute heart failure is a major unmet need. We evaluated the value of biomarkers to predict either low or high risk for early post-discharge events.
METHODS AND RESULTS:
A total of 1653 patients enrolled in the PROTECT trial who were discharged alive and with available blood samples were included. Forty-seven biomarkers were serially evaluated in these patients. Measurement closest to discharge was used to evaluate the predictive value of biomarkers for low and high post-discharge risk. Patients were classified as ‘low risk’ if post-discharge 30-day risk of death or heart failure rehospitalization was <5% while risk >20% was used to define ‘high risk’. Cut-off values that yielded a 95% negative predictive value and a 20% positive predictive value were identified for each biomarker. Partial area under the receiver operating characteristic curve (pAUC) in the high-sensitivity and high-specificity regions was calculated to compare low-risk and high-risk predictive values. Of patients analysed, 193 (11.7%) patients reached the 30-day death or heart failure rehospitalization outcome. We found marked differences between low-risk and high-risk predictors. Cardiac-specific troponin I was the strongest biomarker for low-risk prediction (pAUC = 0.552, 95% confidence interval 0.52-0.58) while endothelin-1 showed better performance for high-risk prediction (pAUC = 0.560, 95% confidence interval 0.53-0.59). Several biomarkers (individually and in combination) provided added predictive value, on top of a clinical model, in both low-risk and high-risk regions.
Different biomarkers predicted low risk vs. high risk of early post-discharge death or heart failure readmission in patients hospitalized for acute heart failure.
© 2017 The Authors. European Journal of Heart Failure © 2017 European Society of Cardiology.
Acute heart failure; Biomarker; High risk; Low risk; Predictive value; Risk stratification