Meta-analytic evidence of increased breakthrough bleeding during prophylaxis with B-domain deleted factor VIII.

A recent meta-analysis of 13 observational studies suggested reduced haemostatic efficacy during prophylaxis and shortened half-life of B-domain deleted factor VIII (BDD-rFVIII) as compared with full-length factor VIII (FL-FVIII). The meta-analysis included a multivariate model that took both dose and age into account. In addition, several assumptions and interpretations were required in order to conduct the meta-analysis. It is important to test the impact of such assumptions and interpretations through sensitivity analysis. In the published meta-analysis, results of several sensitivity analyses were described involving the effect of study design; heterogeneity of subjects in some studies; type of assay used for half-life determinations; and year of publication. In two subsequent brief reports, additional sensitivity analyses addressed choice of median-to-mean conversion factor over a wide range of scenarios and use of age at start of prophylaxis vs mean age during prophylaxis in the multivariate model. A recognized inherent difficulty in the meta-analysis of multiple published reports from similar studies is the possibility of some subject or data overlap. Therefore, the present communication details further sensitivity analyses encompassing assumptions regarding the possibility of subject duplication in a subgroup of subjects from one study and possible duplication of pharmacokinetic data from two smaller studies within the reports of two larger studies. All the sensitivity analyses support the main conclusions of the meta-analysis, namely, that BDD-rFVIII substantially increases the risk of breakthrough bleeding during prophylaxis, possibly at least partly because of shorter half-life than that of FL-FVIII.