From: "Bonate, Peter"
Subject: [NMusers] MM elimination vs dose
Date: Fri, 30 Sep 2005 12:25:03 -0400

Dear all,
I am having trouble with something and I was hoping that someone would have
some brilliant solution for me.  I am fitting a 3-compartment model IV in 95
patients with rich data using FO-approximation.  My base model has BSA and
Age on CL as power functions.  When I include dose on clearance as a power
model, the OFV decreases from 13620.35 to 13516.61.  The power term (SE) was
-0.341 (0.152).  Further, if I plot AUC(0-infinity) vs dose there is
evidence of nonlinearity based on a power model.  A 2-fold increase in dose
results in a ~2.3 fold increase in AUC.

The next model I tried was a Michaelis-Menten model where BSA and Age was
included on Vmax.  The OFV was 13619.24.  No change from the base model.  I
also tried different starting values for Vmax and Km and get around the same
solution each time, so local minima I don't apply here.

Even when I compare a 3-compartment model with no covariates to a
3-compartment model with MM elimimation the OFV does not change (14138.61
vs. 14138.79, respectively).  But as soon as a put Dose on CL
(OFV=13560.37), BSA on CL (OFV=13651.29), and AGE on CL (OFV = 13942.51),
the OFV decreases from the base model of 14138.61.  Even taking into account
the problems with the distribution of the likelihood ratio test and
FO-approximation, these seem to be real decreases.

So has anyone any ideas for why dose drops the OFV in a linear compartmental
model but a MM elimination model won't work.  All models minimized
successfully without any singularities.  Is the nonlinearity so mild that it
can't be supported by a nonlinear compartmental model?

Thanks for your help,

Pete Bonate

Peter L. Bonate, PhD, FCP
Director, Pharmacokinetics
Genzyme Corporation
4545 Horizon Hill Blvd
San Antonio, TX  78229
phone: 210-949-8662
fax: 210-949-8219


From: Sam Liao
Subject: Re: [NMusers] MM elimination vs dose
Date: Fri, 30 Sep 2005 13:31:15 -0400

Dear Peter:
I am not surprise to hear the problem you had on the MM elimination model.  Based on
my expereience, in order to estimate the Vmax and Km reliably, you will need a
cross-over study with subjects received at least two different doses.

Sam Liao
Pharmax Research 

From: Nick Holford
Subject: Re: [NMusers] MM elimination vs dose
Date: Sat, 01 Oct 2005 05:43:09 +1200


Non-linear PK can influence other processes apart from elimination. Have you
looked for non-linear changes in volume of distribution? You may also want to
consider changes in bioavailability with dose -- there could be an issue with
the IV formulation as doses increase. I doubt if your drug is eliminated via
the skin so why not look for a biologically based allometric scaling function
for clearance rather than the discredited BSA scaling method?

Finally, IMHO FO is ok for quick and dirty teaching projects but not for serious work...

Nick Holford, Dept Pharmacology & Clinical Pharmacology
University of Auckland, 85 Park Rd, Private Bag 92019, Auckland, New Zealand tel:+64(9)373-7599x86730 fax:373-7556


From: "Gobburu, Jogarao V"
Subject: Re: [NMusers] MM elimination vs dose
Date: Fri, 30 Sep 2005 15:34:16 -0400

I agree with Nick. In fact, I would strongly advocate use of simple NCA
analysis to understand the nature of the nonlinearity, before embarking on
mixed effects modeling.



From: "Janet R. Wade"
Subject: Re: [NMusers] MM elimination vs dose
Date: Sat, 1 Oct 2005 09:13:25 +0200

Dear Pete

Do you have data from more than one dose level per subject?  If so, then
your 'issue' reminds me of something I once encountered and where the PK
also appeared to be non-linear with respect to dose (to my surprise since
there was no previous indication of this).  In my case it eventually turned
out that the dose effect was no longer significant once I incorporated IOV
on CL.

Kind regards


Janet R Wade

Exprimo NV