From: Gautam Baheti 
Subject: [NMusers] Two formulation of the same Drug
Date: Tue, 30 Apr 2002 07:08:55 -0700 (PDT)

 Hello NONMEM users, 

I have been using NONMEM to determine the apparent clearance (CL/F) of two
different formulations (X and Y) of a drug in a population model. It has been
established from previous studies that formulation X has a higher bioavailability
hence a lower apparent clearance (30% higher) than formulation Y. Twenty seven
patients are on formulation X and 21 patients on formulation Y.

I am trying to use formulation type as a covariate and have been unsuccessful
to get it to run when I use both the formulations together, but I get the expected
value when I break them into two separate data sets. I have tried different estimates
for Theta (5) the error I am getting is ? The estimates of theta are near the boundary.?
or at times I get Minimization is successful but I am not getting and estimates.

My control file is as follows

$PROB ANALYSIS2

$INPUT ID TIME AMT DV SS DI=II TAD DOSE

$DATA DOSE.PRN IGNORE=C

$SUBROUTINES ADVAN2 TRANS2

$PK

 TVCL=THETA (1)+THETA (5)*DOSE

 CL=TVCL*EXP (ETA (1))

 TVV=THETA (2)

 V= TVV*EXP (ETA (2))

 K= CL/V

        IF (DOSE.EQ.1) TVKA=THETA (3)

        KA=TVKA

        IF (DOSE.EQ.0) TVKA=THETA (4)

        KA=TVKA

S2=V

$THETA (0,105,500) (1,1000, 5000) (0.25 FIXED) (1.35 FIXED)(5,20,100)

$ERROR

Y=F*(1+ERR (1))

$OMEGA (0.3)

$SIGMA (0.3)

$EST MAXEVAL=1000 PRINT=5 

$COV 


Any help in this regard will be appreciated.

 Thank you,

 Gautam Baheti



*******

From: "Gobburu, Jogarao V" 
Subject: RE: [NMusers] Two formulation of the same Drug
Date: Tue, 30 Apr 2002 10:24:23 -0400

Dear Gautam,
 
1. Firstly, as you pointed out, you should probably test for the difference in
the rate of absorption and/or relative bioavailability of X and Y. CL and V are
fundamental properties of the drug and not necessarily of the formuation. So,
please try the following: 

 TVCL=THETA (1)

 CL=TVCL*EXP (ETA (1))

 TVV=THETA (2)

 V= TVV*EXP (ETA (2))

 K= CL/V

TVKA=THETA(3)*DOSE + THETA(4)*(1-DOSE)

KA = TVKA 

F1 = 1*DOSE + THETA(5)*(1-DOSE)  ; Theta(5) is the bioavailability
of DOSE=0 formulation relative to DOSE=1
 
2. Further, since you have two occasions of treatments try to test if inclusion
of inter-occasion variability is important on CL and V.
 
Regards, 
Joga Gobburu, 
Team Leader, Pharmacometrics,
CDER, FDA.


*******

From: "Bachman, William" 
Subject: RE: [NMusers] Two formulation of the same Drug
Date: Tue, 30 Apr 2002 10:27:41 -0400

A couple of things off the top of my head:
 
1. CL (CL/F as you are modeling it) confounds the CL (which is probably
the same for both formulations) and the F (which is different), so you might
try separating out the F by modeling F1.  (and applying the formulation covariate
to that parameter instead of CL).  The confounding might explain why you can
model them separately but not together. 

2. THETA(5) is constrained (5,20,100).  Make sure it's really in that range.  I try
not to over-constrain parameter if I don't have to.  I just generally constrain to
be positive (if appropriate.)  might explain the at boundary situation.

3. Does either formulation exhibit a lag? if so, model it.

4. Is there significant variability in KA? if so, model it.  Fixing the one KA may also be 
over-constraint.  Not knowing your data, its' hard to tell.
 
good luck,
Bill
 

*******