Increased breakthrough bleeding during prophylaxis with B-domain deleted factor VIII--a robust meta-analytic finding.

Meta-analyses of observational studies have become increasingly common to support evidence-based clinical decisions. We analyzed currently available clinical studies of full-length factor VIII (FL-FVIII) vs. B-domain deleted recombinant factor VIII (BDD-rFVIII) using a random effects model to investigate possible differences in clinical efficacy in patients treated during prophylaxis. Some studies reported breakthrough bleeding incidence as mean annual total bleeds, whereas others reported median bleeds. In accord with the consensus recommendations by the Meta-analysis of Observational Studies in Epidemiology (MOOSE) group, all available studies where included. For analysis, data were combined by converting median to mean annual total bleeds using a conversion factor of 2.6, based on clinical data previously compiled by the Universal Data Collection Program of the U.S. Centers for Disease Control and Prevention. To evaluate the sensitivity of our model upon the choice of conversion factor, we re-estimated incidence rate ratios for breakthrough bleeding over a wide range of conversion factors from 1.4-2.6. Even at the lowest extreme conversion factor of 1.4, bleeding incidence was statistically higher in patients treated with BDD-rFVIII compared with FL-FVIII. We also examined the impact of reported patient age on our multivariate model. Exposure to BDD-rFVIII remained an independent predictor of bleeding, regardless of patient age at start or mean age during prophylaxis. These analyses further support the robustness of our meta-analysis and its conclusions.