From: bvatul@ufl.edu

Subject: DRUG AND METABOLITE

Date: Mon, 28 May 2001 09:39:06 -0400

 

Hello All

 

Could somebody help me in this?

 

I am analysing the sparse data of a drug and its metabolite after intravenous infusion. I first analysed the data of drug alone using ADVAN3 TRAN4. Using these parameters I then tried to fit the drug and metabolite simulatneously in ADVAN6 TRAN1. In the $DES I used only k10 (which is a sum of elimination of the parent drug and formation rate of its metabolite). I first simulated to get a good idea of starting parameters. I am getting better IPRED when I analyse the parent drug alone. But when I am analysing the drug and metabolite the IPRED of parent drug goes very bad. The variability in the data is also very high. My queries are as follows:

 

1. Does this show that I should fix the predetermined parameters of drug (ADVAN3)and float the metabolite parameters?

2. How will transformation of data (DV) help me?

3. In such cases which is the best method (FO or FOCE or FOCE with interaction).

4. Is it possible that similar distribution models do not apply for a drug and its metabolite in cases of high variability? How to recognise this?

$PROBLEM ANALYSIS OF POP DATA

$INPUT ...

$DATA DM.CSV IGNORE=C;

$SUBROUTINES ADVAN6 TRAN 1 TOL=3

 

$MODEL

COMP=(1); CENTRAL) 1

COMP=(2); PERI) 2

COMP=(3); METAB) 3

COMP=(4); METAB 4

 

$PK

TVQ=THETA(1)

TVCL1=THETA(2)

TVV1=THETA(3)

TVV2=THETA(4)

TVK13=THETA(5)

TVK30=THETA(6)

TVK34=THETA(7)

TVK43=THETA(8)

 

Q=TVQ*EXP(ETA(1))

CL1=TVCL1*EXP(ETA(2))

V1=TVV1*EXP(ETA(3))

V2=TVV2*EXP(ETA(4))

K13=TVK13*EXP(ETA(5))

K30=TVK30*EXP(ETA(6))

K34=TVK34*EXP(ETA(7))

K43=TVK43*EXP(ETA(8))

 

K12=Q/V1

K10=CL1/V1; K10=K10'+K13

K21=Q/V2

 

S1=V1/1000

S3=1

$DES

DADT(1)=-K10*A(1)+K21*A(2)-K12*A(1)

DADT(2)=K12*A(1)-K21*A(2)

DADT(3)=K13*A(1)-K30*A(3)-K34*A(3)+K43*A(4)

DADT(4)=K34*A(3)-K43*A(4)

 

$THETA

(0.00001, 47) ; Q

(0.00001, 19.1) ; CL1

(0.00001, 11) ; V1

(57, FIX) ; V2

(0.00001, 2.96) ; K13

(0.00001, 0.18); k30

(0.00001, 0.3); k34

(0.00001, 0.1); K43

$OMEGA

(0.181)

(0.176)

(0.5, FIX)

(0.5)

(0.5)

(0.5)

(0.5)

(0.5)

$ERROR

Q1=0

IF(CMT.EQ.1)Q1=1

Q2=0

IF(CMT.EQ.3)Q2=1

 

Y1=F*(1+ERR(1))+ERR(3)

Y2=F*(1+ERR(2))+ERR(4)

 

Y=Q1*Y1+Q2*Y2

IPRED=F

IRES=DV-IPRED

$SIGMA

(0.087)

(0.1)

(15.2)

(10)

 

$EST METHOD=0 MAXEVAL=9999 PRINT=5 NOABORT POSTHOC

$TABLE ....

 

Thanks in advance.

 

Atul

 

*****

 

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

Subject: RE: DRUG AND METABOLITE

Date: Tue, 29 May 2001 08:19:12 -0400

 

Atul,

you have an inconsistency in your differential equations: the term -k13*A(1) should appear in the equation for the central compartment. Scaling parameter S3 should not be 1, since you measure metabolite concentrations. Also, for the linear system like yours you do not need to use $DES, you can use general linear ADVAN (ADVAN5 or ADVAN7).

 

Katya

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

Ekaterina Gibiansky, PhD

Senior Scientist

 

GloboMax LLC

7250 Parkway Drive, Suite 430

Hanover, MD 21076

Voice (410) 782-2234

FAX (410) 712-0737

E-mail: gibianskye@globomax.com

 

*****

 

From: "Perez Ruixo, Juan Jose [JanBe]" <JPEREZRU@janbe.jnj.com>

Subject: RE: DRUG AND METABOLITE

Date: Tue, 29 May 2001 14:19:20 +0200

 

Hello all,

 

When we fit simultaneously parent drug and metabolite concentrations it is not possible to estimate the distribution volume of the metabolite (for this reason, S3=1) and fraction of parent drug dose that is converted to metabolite (you assume F3 = 0). If K10 is the total elimination rate constant of parent drug, and we assume that all parent drug is converted into the metabolite, and then K10 is the formation rate constant of the metabolite (this is K10 = K13). You say 'In the $DES I used only K10 (which is a sum of elimination of the parent drug and formation rate of its metabolite)'. It is OK, but your initial estimates of K10 and K13 are not properly constrained to avoid K13 > K10.

 

Possible solutions:

1. Using the constraint K10 >= K13 the estimation step could work better.

2. Modifying actual control file as follows:

 

CL1=THETA(.)

CL2=THETA(.)

V1=THETA(.)

KEL=CL1/V1

K13=CL2/V1

K10=KEL+K13

$DES

DADT(1)=-K10*A(1)+K21*A(2)-K12*A(1)

DADT(2)=K12*A(1)-K21*A(2)

DADT(3)=K13*A(1)-K30*A(3)-K34*A(3)+K43*A(4)

DADT(4)=K34*A(3)-K43*A(4)

 

Now the elimination rate constant of parent drug (KEL+K13) is always greater than K13 (if lower boundaries are 0).

 

Moreover, usually the actual half life of the metabolite is shorter than that of the parent drug and, consequently, the apparent metabolite elimination is controlled by the parent drug elimination. It could be useful thus to restrict K30 =< K13. This is something like a flip-flop model. In your case, however, I think it is not needed because initial estimate of K30 is 0.18 (lower than 2.96).

 

Best regards.

 

Juan Jose Perez Ruixo.

Janssen Research Foundation.