Bioequivalence and Bioavailability Forum

Dear Rahul,

THX for the !

❝ (ANDA application No is 75133. Information is available on CDER)

Can you come up with a link?

❝ The study conducted was crossover, two treatment, two period in which volunteers (N =30) were divided in two groups with subject 1 to 20 in group I and 21 to 30 in group II.

❝ The dosing was done in a cross over fashion to one group first and then to second group.

Though I have not seen the application, from the details you gave I would say this was not an add-on study, but a study with a fixed sample size of 30 subjects, where due to logistic reasons two groups were dosed. This is acceptable with the FDA (see FDA's 2001 Guidance, Section VII.A.).

❝ As add on dsign is not acceptable in US, I am confused with this type of design.

Oh no, an add-on design is acceptable. FDA was actually actively involved in the work published by Potvin et al. 2008 (see this post for the reference and this post for a flow-chart). Donald Schuirmann of the FDA is one of the co-authors and according to a conversation I had in May this year with Barbara Davit (Acting Director of the Division of Bioequivalence II, Office of Generic Drugs, CDER/FDA) the method is not only acceptable but supported by the FDA.

❝ In add on study design, if the result of the first group is passing, then is it necessary to carry out the dosing for the second group?

If power in the interim is >80% and the study passes 80%-125% at alpha 0.05 (=90% confidence interval), stop. If power <80% and the study passes at alpha 0.0294 (=94.12% confidence interval), stop - otherwise go for the second stage. In the pooled analysis (stage 1 + stage 2), calculate the 94.12% confidence interval. For details see one of my lectures (slides 52-62).

❝ Can we divide the number of volunteers in 2:1 proportion in add on design?

If you fail in the first stage at alpha 0.0294 perform a sample size calculation based on the CV from the first part. No fixed sample size of 1:anything.