Page 30 - Biomarkers for risk stratification and guidance in heart failure

Multimarker risk score in emergency department dyspnea
Prediction of outcome
More than half of the patients (n=347, 58%) were admitted to the hospital
subsequent to their presentation on the ED. Ninety day follow-up was completed in
all patients. After 90 days, 78 patients (13%) had died, 58 (74%) from cardiovascular
causes. At a median follow-up of 365 days (IQR 266-365), 145 patients (24%) had 2 died, 100 (69%) from cardiovascular causes.
All biomarkers were highly associated with 90-day mortality (table 2) and had comparable AUCs for the prediction of 90-day mortality, ranging from 0.73 to 0.75 (P>0.5 for all comparisons, table 3). Cut off values determined from ROC-curve analyses were 4,500 pg/ml for NT-proBNP, 0.04 μg/L for hs-cTnT, 1.125 mg/L for Cys-C, 25 mg/L for hs-CRP and 25 μg/L for Gal-3. Biomarkers were significantly correlated, with the strongest correlation being present between Cys-C and Gal-3 (r=0.70, P<0.001) followed by NT-proBNP and hs-cTnT (r=0.55, P<0.001, appendix 2).
In categorical multivariable analysis including all laboratory findings, NT- proBNP, hs-cTnT, hs-CRP and Cys-C remained independently associated with 90- day mortality (table 2) and thus formed the final biomarker panel. The combination of these 4 biomarkers (i.e hs-cTnT, hs-CRP, Cys-C and NT-proBNP) reached a high predictive accuracy with an AUC for 90-day mortality of 0.83 (95% confidence interval \[CI\]: 0.78-0.87) and the number of biomarkers elevated (i.e. none, 1, 2, 3 or 4) was strongly associated with increased risk of 90-day mortality (odds ratio \[OR\]: 2.94 per elevated biomarker, 95% CI 2.29-3.78, Wald 71.0, P<0.001) with 90-day mortality rates of 0.7%, 4.5%, 10.4%, 25.2% and 49.1%, respectively (P<0.001). The gradual increase in 90-day mortality rate with increasing number of biomarkers elevated was retained in subgroup analyses of patients with versus without acute decompensated heart failure (ADHF) and renal dysfunction (figure 1.). Each specific combination of either 2 or 3 biomarkers (e.g NT-proBNP + hs-cTnT versus hs-CR + Gal-3) had similar 90-day mortality rates (P>0.1 for all comparisons). The 1-year mortality rate also showed a significant increment with increasing number of biomarkers elevated (6.8%, 11.5%, 23.2%, 45.5%, and 61.8%, P<0.001) which is shown in figure 2A.
Following from categorical multivariable analysis on clinical risk factors, age, gender, HF history, dyspnea New York Heart Association functional class and systolic blood pressure formed the final clinical model (table 2). When correcting the final biomarker panel for the final clinical model, hs-cTnT, hs-CRP and Cys-C remained independent prognostic biomarkers whereas NT-proBNP was dropped (table 2). The final prognostic model therefore consisted of the final clinical model
29