Multimarker Analysis of Serially Measured GDF-15, NT-proBNP, ST2, GAL-3, cTnI, Creatinine, and Prognosis in Acute Heart Failure

BACKGROUND: Studies on serially measured GDF-15 (growth differentiation factor 15) in acute heart failure (HF) are limited. Moreover, several pathophysiological pathways contribute to HF. Therefore, we aimed to explore the (additional) prognostic value of serially measured GDF-15 using a multi-marker approach to more accurately predict HF risk. METHODS: TRIUMPH (Translational Initiative on Unique and Novel Strategies for Management of Patients With Heart Failure) is a prospective cohort of 496 patients with acute HF who were enrolled in 14 hospitals in the Netherlands between 2009 and 2014. Blood sampling was scheduled at 7 moments during 1-year follow-up. GDF-15, NT-proBNP (N-terminal pro-B-type natriuretic peptide), ST2 (suppression of tumorigenicity 2), galectin-3, troponin I, and creatinine were measured in a central laboratory. We associated repeated measurements of these biomarkers with the composite primary end point of all-cause mortality and HF rehospitalization, using multivariable joint modeling. RESULTS: Median age was 74 years, and 37% were women. Median baseline GDF-15 was 4632 pg/mL. The primary end point was reached in 188 (40%) patients. The average estimated GDF-15 level increased weeks before the primary end point was reached. The hazard ratio per 1 SD difference in log-GDF-15 was 2.14 (95% CI, 1.78-2.57) unadjusted, 1.96 (1.49-2.53) after adjustment for clinical confounders and 1.44 (1.05-1.91) when jointly modeled with all biomarkers. The adjusted HRs for NT-proBNP were 2.38 (1.78-3.33) and 1.52 (1.15-2.08), respectively. The multimarker model combining GDF-15, NT-proBNP, and troponin I provided a favorable risk discrimination (area under the curve=0.785). CONCLUSIONS: Sequentially measured GDF-15 independently and dynamically predicts risk of adverse outcomes during 1-year follow-up after index admission for acute HF. NT-proBNP remains a robust predictor among potential candidates. Multiple biomarkers should be considered for stratification in clinical practice. REGISTRATION: URL: https://www.trialregister.nl/trial/1783; Unique Identifier: NTR1893. (The trial can be found temporarily at https://trialsearch.who.int/Trial2.aspx?TrialID=NTR1893.).