Importance of Biomarkers for Long-Term Mortality Prediction in Acutely Dyspneic Patients

Abstract

BACKGROUND Although numerous biomarkers may be prognostically meaningful in patients with acute dyspnea, few comparative analyses have addressed possible associations between a wide range of candidate biomarkers and clinical variables. METHODS Vital status was obtained for 517 acutely dyspneic patients at 4 years after emergency department presentation. A wide array of biomarkers was measured in this cohort, including natriuretic peptides, necrosis markers, inflammatory markers, hematologic markers, and renal markers. We performed statistical evaluation by using minimization of the Bayesian information criterion to evaluate predictors of 4-year mortality. Cox proportional hazards analysis was used to confirm results from the Bayesian information criterion. A final risk model was derived, and this model was then validated by applying it to patients from a separate cohort of acutely dyspneic patients. RESULTS By 4 years, there were 186 deaths (36%). In addition to several clinical variables, several biomarkers were significant predictors of death, including log-transformed concentrations of hemoglobin (hazard ratio=0.77; P < 0.001), soluble ST2 (hazard ratio=1.38; P < 0.001), and amino-terminal pro-B–type natriuretic peptide (hazard ratio=1.19; P < 0.001). Risk models that used these significant variables were accurate in predicting 4-year mortality in both the training and validation sets. CONCLUSIONS When added to traditional clinical variables, selected biomarkers added significant value for long-term prognostication in acute dyspnea.