Elevated on-treatment D-dimer level is associated with major bleeding during extended anticoagulation for venous thromboembolism

Assessing optimal anticoagulant duration in venous thromboembolism (VTE) requires balancing the risks of recurrent VTE and anticoagulant-associated bleeding with fixed-duration or indefinite anticoagulation. In unprovoked VTE, elevated on-treatment D-dimer level is a predictor of recurrent thromboembolism after anticoagulant cessation, but its impact on anticoagulant-associated bleeding is unknown. We performed a post-hoc pooled analysis of two prospective cohorts (BLEEDRISK and REVERSE-2 studies) that enrolled unprovoked VTE patients who continued anticoagulation after ≥3 months of initial treatment. We calculated incidence rates of major bleeding, intracranial hemorrhage and fatal bleeding by D-dimer quartiles, then determined hazard ratios with the lowest D-dimer quartile as reference. Among 3,337 participants, incidence rates of anticoagulant-associated major bleeding from the lowest to highest D-dimer quartile were 0.45, 0.91, 1.43 and 2.20 events per 100 person-years respectively. Accounting for age, sex and anticoagulant type, individuals in the highest D-dimer quartile had a 2.7-fold higher risk of anticoagulant-associated major bleeding compared to the lowest quartile (adjusted hazard ratio [aHR] 2.70, 95% CI 1.35 - 5.41, p=0.005). This association remained robust after additional adjustment for hemoglobin, creatinine and concurrent anti-platelet use in the BLEEDRISK cohort. We did not observe differences in risks of intracranial hemorrhage or fatal bleeding by D-dimer quartiles. Elevated on-treatment D-dimer levels predicted higher risks of major bleeding during indefinite anticoagulation. Given the established relationship between elevated on-treatment D-dimer levels and recurrent VTE after time-limited anticoagulation, our findings underscore the need for additional risk stratification tools in this population to personalize anticoagulant treatment that maximizes its net clinical benefit.