Bioequivalence and Bioavailability Forum

Hi Osama,

❝ Although the GCC GL is adopted from the EMA GL, but the upper cap of scaling is about only 39% and not 50%!!!!

library(PowerTOST)
noquote(sprintf("%.2f%%", 100*U2CVwR(U = 1/0.75)))
[1] 39.25%

I wouldn’t call that scaling. The GL calls for fixed limits of 75.00–133.33% (based on a “clinically not relevant Δ” of 25%) for any CV_wR >30% (there is no upper cap and the widened limits are fixed).
That’s the approach mentioned in the EMA’s Q&A-document of July 2006:

BTW, with fixed limits there are no issues with inflation of the type I error* like in all reference-scaling methods (EMA, Health Canada, FDA). When discussing the EMA’s draft, sponsors complained that ABEL is more restrictive at CV_wR 30–39.25% than the “old” approach…

❝ At the moment, I think you could have a good chance under the conditions that your BE study is demonstrated in a replicate design and that the high within-subject variability for Cmax not caused by outliers.

That’s interesting!

---

• Wrong! See this thread.