| dc.description.abstracteng | Early detection of significant cardiac dysfunction in patients at high cardiovascular risk is a relevant clinical problem. Biomarkers that are elevated by cardiac dysfunction may simplify diagnostic algorithms. Within the ongoing Diast-CHF trial, 1590 outpatients with a risk factor for or history of heart failure underwent cardiovascular examination, comprehensive echocardiography and blood sampling for the measurement of seven biomarkers (MR-proANP, NT-proBNP, CT-proAVP, hsCRP, CT-proET-1, MR-proADM and PIIINP). Patients had relevant systolic dysfunction (SD) when left ventricular ejection fraction (LVEF) was < 50 % or relevant diastolic dysfunction (DD) when LVEF was > 50 % and echocardiographic signs of elevated filling pressures were present. Compared to a control group of healthy subjects (KG) biomarker concentrations were significantly elevated in subgroups with cardiac dysfunction (SD, DD). The diagnostic ability of the single biomarkers was analyzed by measuring the area under the receiver operator characteristics curve (AUC). Of all seven biomarkers, natriuretic peptides (MR-proANP, NT-proBNP) had the highest AUC when discriminating between subgroups (SD/DD vs. KG, SD vs. KG, DD vs. KG). Biomarkers were less accurate to diagnose DD than SD. To understand the additive value of novel biomarkers, we constructed a model by multivariate binary logistic regression, using variables indicative of a risk for the development or presence of heart failure consisting of medical history, physical examination plus conventional laboratory testing. Regarding the diagnostic ability of the single biomarkers we only added NT-proBNP and MR-proANP. Presence of SD/DD was predicted by the clinical model with reasonable accuracy (0,704). Adding biomarkers, NT-proBNP alone improved the statistical model to a final AUC of 0,738, MR-proANP provided no further increase in AUC. The AUC for the prediction of SD was 0,810 for established predictors, adding NT-pro-BNP raised the AUC to 0,840. For the detection of DD AUC for conventional predictors was 0,715, adding MR-proANP and NT-proANP yielded a total AUC of 0,741. In conclusion, combining multiple biomarkers was only useful for the detection of DD. Although discrimination was improved, the information gained was little.
In our study a multimarker approach yielded limited improvement for the detection of SD/DD over established predictors of cardiac dysfunction. | de |