Bioequivalence and Bioavailability Forum

Hi velupharm!

❝ [..] If the pilot study fails because of the formulation and still we got

❝ the clue for the change to be made to formulation...

Up to here everything's fine.

❝ ...and we plan for a direct pivotal study.

Oops.
How can you be so sure that your change will be the right one, or in other words, you already have been confident that your formulation would perform well in the first run--but you failed, demonstrating that your expectations may have been unjustified...
...The other possibility is just pure chance - how 'big' was your pilot compared to the CV? For moderate CVs of ~20% a pilot study of sample size 6 is futile (see this post).

❝ In this scenario, can we take the intrasubject CV from this failed pilot

❝ study for calculation of sample size for pivotal study?

If you have the magic technology of altering the amount absorbed without changing the variability, yes.
Just give it a try: take your pilot studie's data and multiply all subjects's 'test' results by the inverse of your point estimate (i.e, if your GMR was 0.7, multiply by 1.43). Feed your piece of software, and what do we get? Point estimate 100%, CV unchanged...

OK, I would dare doing a pivotal if the CV is very low (let's say 5-7%), which justifies the main study with a sample size of 12. But further up the scale?!