Objective: In clinical studies of patients with severe thrombocytopenia, rescue treatments are used to prevent or stop bleeding. Estimating risk reductions of bleeding for clinical study medications can be challenging. This study evaluated a new and possibly more accurate way of assessing the effects of a treatment intervention on bleeding-related outcomes. We developed a composite endpoint, termed bleeding-related episodes (BRE).
Research design and methods: BREs were assessed in a post-hoc analysis of patients with chronic immune thrombocytopenia (ITP) who participated in two romiplostim, phase 3, placebo-controlled studies. Patients received romiplostim or placebo once weekly for 24 weeks. A BRE was defined as an actual bleeding event and/or the use of rescue medication. In total, 125 patients (41 placebo, 84 romiplostim) with platelet counts <30 K were enrolled.
Clinical trial registration: NCT00102323/NCT00102336.
Results: The rate of all BREs across all studies was reduced by 56% in patients receiving romiplostim compared with placebo. The rate of BREs using immunoglobulin (IVIg or anti-D Ig) was reduced by 89% in patients receiving romiplostim compared with placebo. BREs were more frequent in both groups at platelet counts <50 × 10(9)/L. Results were similar between splenectomized and nonsplenectomized patients. We believe that prior to the development of this tool, bleeding events were underdiagnosed. The BRE tool allowed the identification of multiple interventions within bleeding episodes, which may have required separate interventions and were therefore considered to be additional BREs.
Conclusions: In this study, the composite endpoint of a bleeding event and the use of rescue medication within close proximity of the bleeding event appears to be feasible and informative. The BRE tool allows for more precise understanding of the effect of rescue therapies in ITP and has broader applications to future clinical trials where assessment of bleeding risk can be complicated or masked by rescue interventions.
Limitations: This was a post hoc analysis. The assignment of platelet counts to a BRE was based on the platelet count on the first day of a BRE, which may not reflect the platelet count during the entire episode, and the assignment of platelet counts was based on the estimation required for events that occurred between weekly measurements.