Bioequivalence and Bioavailability Forum

Hi Rahul,

❝ Dear Helmut,

    ▲▲▲▲▲▲▲   Not interested in opinions of other members? I’m not the only guru here.

❝ Thank you for allowing me to join forum.

No allowance needed. Welcome to the club.

❝ We conducted two BE study for one of the ANDA, in one study we got 90% CI (77-108) and another study we got (106-125.8) …

Since the CI in the second study was substantially narrower than in the first, I guess its sample size was larger. Details?

❝ … for same test and RLD lot.

That’s rather unusual. Since you are obviously aiming at ABE (80.00–125.00%) and the drug is not highly variable (correct?), such a behavior of the PE is strange (91.2% vs 115.5%).
I know only a few cases (e.g., dasatinib), where the reference is a lousy product (extreme batch-to-batch variability) and it’s very difficult to hit such a moving target in a repeated study. Seems not to be in your case. Confusing. Randomization, data entry, coding checked? Anything different? Clinical performance, bioanalytics?

❝ Will it possible to combine both study two justify product is bio-equivalent based on observe variability for regulatory submissions.

Extremely unlikely for the FDA, nothing stated in any guidance. An R-t-R almost guaranteed.
The EMA is pretty clear in this respect:

[…] a combined analysis of all studies can be provided in addition to the individual study analyses. It is not acceptable to pool together studies which fail to demonstrate bioequivalence in the absence of a study that does.

(my emphasis)

Hi RG628,

❝ We conducted two BE study for one of the ANDA, in one study we got 90% CI (77-108) and another study we got (106-125.8) for same test and RLD lot. Will it possible to combine both study two justify product is bio-equivalent based on observe variability for regulatory submissions.

Very unfortunate.
I would check first if T and R were by coincidence switched somehow in the evaluation. I.e. if you had a 90% CI for T/R of 77-108 in one study, and then got a CI of 106-126 for R/T (rather than T/R) in the next. This would be consistent.
Such an issue could for example be related to treatment coding and alphabetical order etc. If T and R were called T and R in one study, but A and B in the other, and if both datasets were evaluated with the same script then strange things could happen depending on how the scripts were written.

I know this is a long shot from my side and I hope it does not offend you that I am mentioning it as a potential cause for a surprising observation. I can only confess that this is the type of mistake I could so easily make myself.