Background: Treatment of patients with atrial fibrillation (AF) with oral anticoagulants is essential to prevent thromboembolic events, mainly stroke, and mortality. However, this therapy is also associated with significant risk of major bleeding events. Biomarkers have recently been recognized to provide improved risk prediction models in AF populations.
Purpose: To further improve the assessment of bleeding risk and understand the pathophysiological mechanisms for bleeding events in patients with AF. The multiplex analytical technology Proximity Extension Assay (PEA) was utilized to screen hundreds of plasma biomarkers simultaneously in small amounts of plasma.
Method: In an unstratified case-control cohort of the ARISTOTLE trial a total of 204 cases with ISTH major bleeding were identified and a random sample of 3,996 controls were followed for a median of 1.7 years. Plasma samples obtained at randomization were analysed by conventional immunoassays or using PEA panels for cardiovascular disease and inflammation, which measured 255 protein biomarkers in multiplex. The association of biomarkers and outcome was evaluated simultaneously by Random Forest and individually by Cox-regression analyses adjusted for clinical characteristics and renal function. The significance level was adjusted for multiple testing.
Results: Six biomarkers were identified as most strongly associated with major bleeding according to both Random Forest and the adjusted Cox-regression analyses. The hazard ratio (95% confidence intervals) per interquartile range was 1.18 (1.04–1.33) for Ephrin type-B receptor 4 (EPHB4), 1.31 (1.14–1.52) for Fibroblast growth factor 23 (FGF-23), 1.35 (1.09–1.68) for Growth differentiation factor 15 (GDF-15), 1.55 (1.27–1.90) for Osteopontin (OPN), 1.45 (1.12–1.87) for Tumor necrosis factor receptor 1 (TNF-R1), 1.36 (1.15–1.60) for Trefoil factor 3 (TFF3).
Conclusions: In patients with AF treated with oral anticoagulants, out of a large number of biomarkers, Osteopontin and Fibroblast growth factor 23 were the strongest predictors of major bleeding events. The association of these biomarkers with bleeding events in this setting is novel and the pathophysiological mechanisms behind these findings warrant validation and further in-depth studies.