Growth differentiation factor-15 predicts major bleeding, major adverse cardiac events and mortality in patients with end-stage kidney disease on haemodialysis: findings from the VIVALDI study.

Abstract

Patients with end-stage kidney disease (ESKD) are at high risk of cardiovascular events and bleeding. Optimizing risk assessment of ESKD patients regarding the risk of thromboembolism and bleeding complications in comorbid conditions, including atrial fibrillation and coronary heart disease, is challenging. To improve risk prediction we investigated growth differentiation factor-15 (GDF-15), a promising cardiovascular biomarker, and its relation to adverse outcomes. In this prospective, multicentre, population-based cohort study, GDF-15 was measured in 594 ESKD patients on haemodialysis (median age 66years, 38% female), who were followed up for a median of 3.5years. The association of GDF-15 with major bleeding, arterial thromboembolism, major adverse cardiac events (MACE) and death was analysed within a competing risk framework. Further, we evaluated the additive predictive value of GDF-15 to cardiovascular and death risk assessment. GDF-15 levels were in median 5475ng/l (25th-75th percentile 3964-7533) and independently associated with major bleeding {subdistribution hazard ratio [SHR] 1.31 per double increase [95% confidence interval (CI) 1.00-1.71]}, MACE [SHR 1.47 (95% CI 1.11-1.94)] and all-cause mortality [SHR 1.58 (95% CI 1.28-1.95)] but not arterial thromboembolism [SHR 0.91 (95% CI 0.61-1.36)]. The addition of GDF-15 to the HAS-BLED score significantly improved discrimination and calibration for predicting major bleeding [C-statistics increased from 0.61 (95% CI 0.52-0.70) to 0.68 (95% CI 0.61-0.78)]. Furthermore, we established an additive predictive value of GDF-15 beyond current risk models for predicting MACE and death. GDF-15 predicts the risk of major bleeding, cardiovascular events and death in ESKD patients on haemodialysis and might be a valuable marker to guide treatment decisions in this challenging patient population.