From: "Sreenivasa Rao Vanapalli" <svanapal@blue.weeg.uiowa.edu>

Date: Mon, 14 May 2001 17:33:55 -0500

I'm trying to fit multiple oral dose data with autoinduction. The data fits well to a two compartment model with first order absorption. But when I'm trying to fit the data to a two compartment with hypothetical enzyme compartment, I'm getting error message 'floating-point error:divide by zero'

The control file I'm using is,

$PROBLEM MOXIDECTIN POOLED DATA

$INPUT ID DOSE=AMT TIME CP=DV WT CMT SEX EVID

$MODEL NCOMPARTMENTS=4 COMP=(DEPOT,INITIALOFF,DEFDOSE)

COMP=(CENTRAL,DEFOBS,NOOFF) COMP=(ENZ) COMP=(PERIPH)

DADT(3)=KENZ-KENZ*A(3)*(1-(CP/(CP+IC50)))

$OMEGA .002 .0001 .0001 .0001 .001 .001 .001

$ESTIMATION MAXEVAL=9999 POSTHOC

Any help would be appreciated.

Sreenivasa Rao Vanapalli, Ph.D,

Janssen Postdoctoral Research Scholar,

S411 PHAR, College of Pharmacy,

University of Iowa, Iowa City, IA-52242

From: "Piotrovskij, Vladimir [JanBe]" <VPIOTROV@janbe.jnj.com>

Date: Tue, 15 May 2001 09:00:19 +0200

Very short look at your code: you have 3 diff eqs in the $DES block, but 4 compartments are declaired in the $MODEL block.

From: "Ziad Hussein" <zhussein@medeval.com>

Date: Tue, 15 May 2001 09:32:58 +0100

Few years ago I have used a model with autoinduction for lamotrigine without the use of differential equation British Journal of Clinical Pharmacology 43: 457-465, 1997. The equation to describe autoinduction is:

TVCL=Theta(1)-Theta(2)*exp(-Theta(3)*Time)

in our case lamotrigine was dosed over 48 weeks. I hope that your data covers at least 2 weeks of dosing to able to model the autoinduction.

Tel: xx-44-161-226 6525 (Ext 295)

From: "Gibiansky, Ekaterina" <gibianskye@globomax.com>

Date: Tue, 15 May 2001 13:24:43 -0400

another short look. You should not use observed values (DV, CP) in the $DES block, differential equations should be written in terms of predicted amounts in the compartments (A(2)/V2 in your case). Also, equation 3 does not seem right.

--------------------------------------------------

E-mail: gibianskye@globomax.com

From: "Sreenivasa Rao Vanapalli" <svanapal@blue.weeg.uiowa.edu>

Date: Tue, 15 May 2001 19:24:51 -0500

I'm trying to fit multiple oral data to a two compartment model. This drug has very long half life (20 days). Plasma samples were collected for 28 days for single dose and every 28 days for multiple dose for six doses and for twelve doses. Declined Plasma concentrations were observed after administration of second dose, suggesting autoinduction. I'm wondering if my control stream to model this data is right or wrong. Can someone help me out. I'm getting floating point error messages.

$INPUT ID MDV DOSE=AMT TIME DV WT CMT SEX EVID

$MODEL NCOMPARTMENTS=4 COMP=(DEPOT,INITIALOFF,DEFDOSE)

COMP=(CENTRAL,DEFOBS,NOOFF) COMP=(PERIPH) COMP=(ENZ)

DADT(4)=KENZ-KENZ*A(4)*(1-(CP/(CP+IC50)))

$THETA (0,.3,10) ;CLIN apparent drug clearance

(0,.001,100) ;KENZ rate constant for enzyme degradation/inactivation

(0,200,1000) ;IC50 Drug conc at 50% of the max inhibition of enzyme

$OMEGA .002 .0001 .0001 .0001 .001 .001 .001

$ESTIMATION MAXEVAL=9999 POSTHOC

Sreenivasa Rao Vanapalli, Ph.D,

Janssen Postdoctoral Research Scholar,

S411 PHAR, College of Pharmacy,

University of Iowa, Iowa City, IA-52242

From: "Piotrovskij, Vladimir [JanBe]" <VPIOTROV@janbe.jnj.com>

Date: Wed, 16 May 2001 08:48:37 +0200

You modified already your control compared to the yesterday version, particularly, you excluded DV from $DES block. There are still a few issues that may cause problems:

1. An enzyme submodel. I prezume the enzyme should have some steady-state level which is altered by the drug. There should be a formation rate for the enzyme not equal to KENZ, say KIN. The baseline amount of enzyme is then KIN/KENZ.

2. Initialization of compartment 4. I think you do not need $AESINITIAL. You have to introduce a unit amount into COMP 4 at TIME=0 using an extra dose record in your data set.

3. In your model, the enzyme has no impact on the drug elimination. In the framework of the autoinduction model CL should depend on the enzyme amount.

Have a look at the following recent publications:

1. Hassan M. Svensson USH. Ljungman P. Bjorkstrand B. Olsson H. Bielenstein M. Abdel-Rehim M. Nilsson C. Johansson M. Karlsson MO. A mechanism-based pharmacokinetic-enzyme model for cyclophosphamide autoinduction in breast cancer patients. British Journal of Clinical Pharmacology. 1999; 48(5):669-677.

2. Kerbusch T. Huitema ADR. Ouwerkerk J. Keizer HJ. Mathot RAA. Schellens JHM. Beijnen JH. Evaluation of the autoinduction of ifosfamide metabolism by a population pharmacokinetic approach using NONMEM. British Journal of Clinical Pharmacology. 2000; 49(6):555-561.