Page 37 - Biomarkers for risk stratification and guidance in heart failure

Chapter 2
T 10, even in patients with normal plasma creatinine41. Cystatin C concentrations are not only indicative of renal function, but may also be elevated in response to inflammation and underlying heart disease 42. Natriuretic peptides (NP’s), mainly b-type natriuretic peptide (BNP) and NT-proBNP, are markers that characterize cardiac wall stress and are established biomarkers for diagnosis and prognosis of acute HF43 and for prognosis of other causes of dyspnea27, 44-46. Nonetheless, the prognostic value of NT-proBNP is very limited for short-term risk stratification9, 14, 15. Several other biomarkers have shown to be superior for short-term risk stratification in acute HF and ED dyspnea13, 15.
We found that all investigated biomarkers were predictive of 90-day mortality and had incremental value on top of the clinical risk model (table 3). Gal-3, however, was dropped from the final biomarker panel. This can at least partially be explained by the existence of significant correlations between Gal- 3 and other biomarkers47. In line with our findings, Gal-3 was not a significant predictor after the inclusion of other predictors in a study of ambulatory heart failure patients 38. Also, hs-CRP and Cys-C partially cover the pathophysiological background of Gal-3 which may have caused the exclusion of Gal-3. Furthermore, the prognostic value of creatinine and BUN was attenuated in the presence of Cys-C, which was previously also reported41. Multi-marker assessment thus revealed 4 markers as independent predictors: hs-CRP, hs-cTnT, Cys-C and NT- proBNP. Importantly, the number of elevated markers was highly predictive for 90-day mortality independent of the specific combination of markers. This indicates that all 4 biomarkers in our study are truly additive to each other and confirms our hypothesis on the independent value of biomarkers from different pathophysiological pathways, which remained true in subgroup analyses of renal function and acute HF diagnosis. When further correcting the final biomarker panel for clinical risk factors, NT-proBNP was excluded from the final prediction model. Previous findings about the inferior predictive value of NT-proBNP for short-term risk prediction as discussed previously support the exclusion of NT- proBNP from our final predictive model, although correlation with other markers probably also plays a role here. The final prediction model and the MARKED-risk score thus included the biomarkers hs-CRP, hs-cTnT and Cys-C in addition to 5 clinical variables.
The MARKED-risk score is the first multimarker score assessing short-term prognosis in ED patients with dyspnea. So far, no experience existed in the combined use of hs-CRP, hs-cTnT and Cys-C with regard to short-term risk stratification in an
36