|
Abstract:
|
Human monoclonal antibodies (mAb) targeting C. difficile toxins A and B were evaluated in earlier Phase 2 trials. Results suggest that A alone is inadequate, but A+B reduced C. difficile infection (CDI) recurrence relative to placebo. Since human data for the individual mAbs were limited, an adaptive approach was considered to assess the 2 individual mAbs, as well as the combination and placebo (to satisfy FDA combination product rules). The Phase 3 design included a planned interim analysis to evaluate individual mAb therapies relative to combined therapy (but virtually all Type I error was allocated to the final analysis based on FDA recommendations). A 4-stage multiplicity strategy provides control of Type 1 error across multiple comparisons at two analysis times. Simulations confirmed that Type 1 error is controlled at planned levels, and there is high power to select an effective regimen. But the probability of dropping an individual mAb arm is only 36%, even for a substantial difference versus the combination. This raises the question of whether more Type I error should have been allocated to the interim analysis, even in the context of evaluating a combination product.
|
