Bioequivalence and Bioavailability Forum

Dear all

There is a wide experience that for many drugs the pharmacokinetics in women differ from those in men and these differences may be explained for example by differences in body weight. Since individualization of drug dosage should be based on therapeutic response gender differences in some pharmacokinetic parameters have no implications for therapy. For testing generic drug products these gender differences do not matter since the objective of a bioequivalence study is, however, not the investigation of the pharmacokinetics of the drug entities itself, but the comparison of the pharmacokinetic profiles of galenic preparations. Bioequivalence is generally recommended to be examined in subjects of both sexes. Based on the assumption that an interaction between galenics and sex of the subject is highly unlikely, the statistical evaluation of a BE study include the whole study population and does not distinguish between different subpopulations (i.e. males and females). There is a wide experience that two formulations that were bioequivalent in one study population will also be bioequivalent in other populations [Rhodes CT: Generic substitution: Does interchangeability mean equality of all functional relevant attributes? Clin Res Reg Affairs (1995), 12(4): 267-272]. Usually a cross-over design is used which is the most powerful design for a bioavailability study since it removes the inter-subject variability (gender differences in pharmacokinetics) from the comparison of average bioavailability between the formulations. This was the stand of my knowledge until yesterday.
We performed a BE study with borderline results in Cmax. The study was well powered and the ratio between males and females was almost equal so we decided to evaluate both sexes separately (with a power of ca. 80% each). The result was striking: whereas the comparison Test vs Reference in females clearly demonstrated bioequivalence within the acceptance range of 80-125%, the 90% CI in men was in great parts (including the point estimator) outside the acceptance range. There was only a very small overlap in CI between men and women. In conclusion there is a clear interaction between galenics and sex of the subject. Do you agree? Have you ever seen such a phenomenon and what could be the explanation?

I am looking forward to your reply.
Kind regards
Dan

---

Edit: Category changed. [Helmut]