From: Yu_Dale 
Subject: [NMusers] Statistical power for AUC and Cmax
Date:Mon, 24 Jun 2002 11:25:56 -0700

This is not a strictly NONMEM question but may be someone could give me some input here. 

In designing a bioequivalence study, the traditional approach is to power the study based on Cmax since it is
typically more variable than AUC. The idea is that if Cmax passess the BE criterion, AUC will always follow.
Is anyone aware of any simulation or publication that verifies this postulate?  On the other hand, is the
idea of picking one of the two metrics incorrect? Should one treat AUC and Cmax as two separate outcomes and
therefore one should consider the joint probability of coming up with 80% power?

Thanks for your help in advance. 

Dale Yu


From: "Masson, …ric" 
Subject:RE : [NMusers] Statistical powerfor AUC and Cmax
Date:Sat, 29 Jun 2002 22:10:42 -0400

BE are typically powered based on the intra-subject CV of Cmax, and AUC.  In most cases, Cmax exhibits
greater variability than AUC since it is affected by not only F and Clearance, but also V and Ka;  this is
true for most drugs but there is exception.  Thus, typically BE studies usually fails on Cmax.
For US, since the criteria (90% CI within 80-125%) is based on both Cmax, and AUC, the study should be
powered on the most variable parameters, usually Cmax.

For regulatory approval, Cmax and AUC are considered as two separate outcomes.  Cmax not only provide a
matrix comparing rate of absorption but also maximum exposure, whereas AUC reflects total exposure.

√?ric Masson, Pharm.D.
Senior Director, Scientific and Regulatory Affairs
Anapharm Inc.