Growth differentiation factor-15 and bleeding risk in patients with venous thromboembolism

Growth differentiation factor-15 (GDF-15) has been proposed as a bleeding biomarker.

To evaluate the predictive value of GDF-15 for clinically relevant bleeding (CRB, major bleeding and clinically relevant non-major bleeding) during anticoagulation for venous thromboembolism (VTE).

This study was part of BACH-VTE, a cohort study including patients with acute, symptomatic pulmonary embolism or deep vein thrombosis with 2 years of follow-up. Patients with active cancer and pregnant/peripartum were excluded. GDF-15 was measured at baseline and 3 months. CRB cumulative incidences and regression models were adjusted for competing risks (death/anticoagulation stop). The performance of the VTE-PREDICT bleeding risk score with and without GDF-15 and the ABC-bleeding risk score, which includes GDF-15, was analyzed by assessing discrimination, calibration, and overall performance.

Among 383 patients (39.2% female, median age 55 years, 48.0% with pulmonary embolism), 49 first CRB occurred (median follow-up: 178 days), corresponding to a 1-year cumulative incidence of 11.9% (95% confidence interval [95% CI]: 8.7-15.7). Median GDF-15 levels were 968 (interquartile range, 645-1554) pg/mL at baseline and 836 (597-1292) pg/mL at 3 months. Baseline GDF-15 was not associated with CRB in the univariable or multivariable model (subdistribution hazard ratio 1.05 [95% CI: 0.76-1.46] and 0.96 [95% CI: 0.70-1.30] per doubling, adjusted for bleeding history, age, and sex). Accordingly, GDF-15 did not meaningfully improve the predictive performance of VTE-PREDICT. ABC-bleeding showed modest discrimination for CRB in this cohort.

In patients receiving anticoagulation for acute VTE, GDF-15 was not associated with CRB and did not improve bleeding prediction.