Bioequivalence and Bioavailability Forum

Dear Ashwini!

❝ Can we use 2x2 crossover studies results for the prediction of 90% C.I. of AUC and Cmax for replicated design (e.g. 3 period, 4 period etc)

Not easily. You may use results from a 2×2 study to perform sample size planning. It's a good idea to add a safety margin on CV_intra (i.e., calculate an upper confidence interval) as well.
Then you run the replicate study and wait. Remember that any particular study is just a sample of many (theoretically an infinite number of) potential studies. If you really want a prediction with some given confidence, you should:

• try to model the PK seen in the 2×2 study (preferrably by PopPK),
  
• simulate a large number of replicate BE studies in silico (Monte Carlo; not less than 10 000 studies),
  
• evaluate the CI obtained in these simulations.

❝ Is there any method in winnonlin or SAS.

1. You have to run a PopPK first (NONMEM, WinBugs, WinNonMix, Kinetica).
   
2. No (although there's some other software by Pharsight called Trial Simulator; trial simulation will be an integrated part of the PopPK module of version 6 'Phoenix' of WinNonlin).
   
3. Everything is possible in
   .