Bioequivalence and Bioavailability Forum

Dear Ohlbe,

❝ ❝ than an increase in the analytical variability can only decrease chances to demonstrate bioequivalence.

❝

❝ Still, I'm not fond of the "I'm just as bad for the reference as for the test so there is no issue" reasoning, sorry.

“To be as bad as the reference” is a common challenge in generic BE.
Joking aside, taking all of Dan’s “ands” for granted higher residual variance increases the producer’s risk. The patient’s risk is fixed. If a study passes despite a high CV, fine. But you are right that we might run into an ethical issue, e.g., if the study was planned as such and better methods would have required fewer subjects. On the other hand, the effect of analytical variability (as compared to physiologic variability) on the residual variance is generally overrated.^*

Do you remember this example (slides 46–49)? The LC/MS-MS method didn’t employ a stable isotope IS and was hit by a massive matrix effect. These were the good ol’ days of plausibility reviews – we stopped the analysis after 12 subjects and went for GC/MS. If you compare the profiles on slide 48 it’s clear that the LC/MS-MS method was awful. But have a look at the CVs and PEs on slide 49. Did it really matter? No. Actually the CV of C_max of the GC/MS method was higher.

---

• Gaffney M. Variance Components in Comparative Bioavailability Studies. J Pharm Sci 1992;81(4):315–7. doi:10.1002/jps.2600810402.