From: onkpharm <>
Subject: problem with AUC
Date: Tue, 17 Apr 2001 12:48:40 +0200

Hello everybody,

I am a new NONMEM-user and trying to model data from 20 children receiving high-dose busulfan orally over 4 days every 6 hours. I'm using a one compartment-model (ADVAN2 TRANS2) with IOV (interoccasional variability) on Cl; furthermore, I included ETAS on Cl, V and ka and I use BSA as a covariate for Cl and V. My questions are as follows:

1. How can I determine the AUC (Area under Curve from 0 to infinity) when the bioavailability is unknown? Does anybody know a possibility?

2. There are 349 plasma samples from 147 administrations; I thought this would be enough data to use Method=1 as estimation method. But I get a smaller objective function when using Method=0. Is this normal with these many data points?

Thanks in advance for your help,

Brunhild Schiltmeyer, R. Ph.
Department of Pediatric Oncology
University of Muenster, Germany



From: "Rik Schoemaker" <>
Subject: Re: problem with AUC
Date: Thu, 19 Apr 2001 12:42:31 +0200

Dear Brunhild,

If you only have oral data then the parameter you estimate is Cl/F and AUC is simply given by AUC=3DDose/(Cl/F); the problem is not in estimating AUC without knowing bioavailability but rather in finding out what Cl is...

As far as comparing objective functions between method 0 and 1 is concerned: you should never do it. It provides no indication whether one is preferable over the other; it is like comparing apples to pears because they provide different approximations to the model you're trying to estimate. This does not mean that in practice they cannot be similar, but this should not guide you. My advice: if method 1 converges then use method 1 (always, at least for your final model).

Kind regards,
Rik Schoemaker

Rik C. Schoemaker, PhD
Biostatistician - pharmacokineticist
Centre for Human Drug Research (CHDR)
Zernikedreef 10, 2333CL Leiden, Netherlands
tel +31 (0) 71 5246417; fax +31 (0) 71 5246499