SPONSOR: Wayeth Pharmaceutical
TITLE: Antihemophilic Factor VIII (Recombinant)B-Domain Deleted Recombinant Factor VIII (BDDrFVIII, ReFacto AF)[moroctocog alfa], submitted for the treatment of hemophilia A.
DOCUMENT REVIEWED: BLA Supplement
FROM: Boris Zaslavsky, Ph.D, Dr.Sc. (HFM-219)
THROUGH: Ghanshyam Gupta, Ph.D., Chief, TEB (HFM-219)
HFM-215/Chronological File HFM-210/Steven Anderson
Dr. Jain raised a question why the results on the upper limit of inhibitor rate was almost the same for the case of 2 inhibitor formation in 94 subjects as in the case of 2 inhibitor formations in 89 subject.
The excessively simplified Bayes approach where the historical data provided upto 50% of the current data was proposed by Dr. Lachenbruch (Division Director in 2005) and agreed upon by the medical reviewers. As a result of this approach, the decrease in the sample size of the current trial increases the impact of the historical data. Therefore, small changes in the sample size do not influence the result.
The suggested Bayes approach was not agreed by me. I wrote an opposing review on this approach on 3/17/05. I am attaching the review to this document. The date on the document is automatically updated to the date when it is opened, but the time of the file creation is 3/17/07.
COMMENTS FROM MY REVIEW OF 3/17/2005:
The mathematical calculations of the recommended sample size are correct if we assume that the Bayesian model used by the sponsor is correct.
The Bayesian model is simplified to the extent that no statistical methods can be used to evaluate its assumptions. The burden of accepting or rejecting this assumption lies with the medical reviews.