Multiple Biomarkers for Risk Prediction in Chronic Heart Failure

Study Questions:

In patients with heart failure (HF), do biomarkers assist in risk prediction above that of currently available clinical risk prediction tools?

Methods:

This was a multicenter cohort of 1,513 patients with chronic systolic HF. Measured biomarkers included uric acid, serum creatinine, high-sensitivity C-reactive protein (hs-CRP), myeloperoxidase (MPO), troponin I (TnI), B-type natriuretic peptide (BNP), soluble fms-like tyrosine kinase receptor-1 (SFlt-1), and soluble toll-like receptor-2 (ST2). From these biomarkers, a risk score was formulated. The primary outcome of interest was survival free of death, transplant, or ventricular assist device (VAD) placement based on the biomarker risk score versus the Seattle Heart Failure Model (SHFM) score. Areas under the receiver operating characteristic curves (AUCs) were compared for the two scores.

Results:

The mean ± standard deviation patient age was 56 ± 15 years, 34% were female, and 22% were African American. Diabetes was present in 28%, New York Heart Association class III or IV symptoms were present in 29% and 8%, respectively, and mean left ventricular ejection fraction (LVEF) was 34 ± 17%. Over a median [interquartile range] 2.7 [1.4-4.0] years of follow-up, 317 patients (21%) had a primary outcome event: 187 died, 99 were transplanted, and 31 received VADs. Of the biomarkers studied, BNP, hs-CRP, ST2, SFlt-1, TnI, uric acid, and serum creatinine were selected for risk score development. Compared with patients in the low biomarker tertile, those in the medium and high biomarker score tertiles had 4.7- and 14-fold increased risk of having a primary outcome event during follow-up. The biomarker score AUC was 0.80 (0.76-0.83), which was superior to that of the SHFM (AUC 0.80, p = 0.003 for comparison). If the biomarker score was used in combination with the SHFM, the net reclassication index was 24% (p < 0.001) and reclassification occurred most in patients who had had events. When the biomarker score was added to the SHFM, the AUC of the SHFM improved to 0.81 (p = 0.019).

Conclusions:

The authors concluded that biomarkers may add prognostication in patients with HF.

Perspective:

This study suggests that the addition of biomarkers measuring burden of inflammation, wall stress, myocyte injury, and renal function may improve in risk stratification of patients with HF. Application of this risk prediction tool will be limited by the nonavailability in clinical laboratories of many of the biomarkers examined. However, like what evolved with the availability of BNP and NT-BNP testing, studies validating the utility and accuracy of such biomarkers may improve access to such testing. Additional concerns are raised with the complexity of the composite model (SHFM added to biomarker model). Models with more components tend to be more challenging to validate in other cohorts, especially when other noncardiac conditions (e.g., arthritis, renal failure) can impact marker accuracy.

Clinical Topics: Anticoagulation Management, Cardiac Surgery, Heart Failure and Cardiomyopathies, Cardiac Surgery and Heart Failure, Acute Heart Failure, Heart Failure and Cardiac Biomarkers, Mechanical Circulatory Support

Keywords: Follow-Up Studies, Chronic Disease, Heart-Assist Devices, Vascular Endothelial Growth Factor Receptor-1, Creatinine, Toll-Like Receptor 2, Natriuretic Peptides, Biological Markers, Troponin I, Stroke Volume, ROC Curve, United States, Inflammation, Area Under Curve, Uric Acid, Systole, Dexamethasone, Heart Diseases, Prognosis, Renal Insufficiency, C-Reactive Protein, Heart Failure, Peroxidase, Peptide Fragments, Diabetes Mellitus


< Back to Listings