Predictive value of factor VIII levels for recurrent venous thrombosis: results from the MEGA follow‐up study

Background Prediction of recurrent venous thrombosis remains a challenge in the clinic. Objective To investigate the predictive value of coagulation factor VIII (FVIII) levels for recurrent venous thrombosis. Patients/methods Patients, aged 18–70 years with a first venous thrombosis, were followed from discontinuation of anticoagulant treatment (1999–2010 MEGA follow‐up study). The levels of FVIII activity, FVIII antigen and von Willebrand factor (VWF) antigen were measured at least 3 months after cessation of anticoagulant treatment. Results Of 2242 patients followed for a median of 6.9 years, 343 developed recurrent thrombosis (incidence rate 2.7/100 patient‐years; 95% confidence interval [CI] 2.5–3.1). Recurrence rates steadily increased with higher FVIII activity levels, from 1.4 (95% CI 1.0–1.9), 2.3 (95% CI 1.8–2.9), 3.0 (95% CI 2.4–3.7), 3.2 (95% CI 2.5–4.1), 3.9 (95% CI 2.8–5.3) to 5.1 (95% CI 3.8–6.8) per 100 patient‐years, for levels ranging from < 100 IU dL−1 to > 200 IU dL−1. Patients in the highest category of FVIII (> 200 IU dL−1) had a three‐fold higher recurrence rate than patients in the lowest category (≤ 100 IU dL−1) (hazard ratio 3.4; 95% CI 2.2–5.3). Results were similar for FVIII antigen and VWF antigen levels, in several sensitivity analyses, and FVIII predicted recurrence rates over a long time period. Within subgroups of patients currently assumed to have low recurrence risks, a high level of FVIII was still predictive for recurrences. Adding FVIII to an existing prediction model (DASH score) improved its predictive value, and, after replacement of D‐dimer with FVIII, the model performed equally well, if not better. Conclusions FVIII predicted recurrence in a dose–response fashion, overall and in several subgroups, and is a strong candidate component of recurrence prediction tools.