From: "Sam T" <samt4@hotmail.com>
Subject: Data Construct / ADVAN selection
Date: Tue, 03 Apr 2001 13:11:17 -0000

Dear NONMEM users,

I would like to use NONMEM to do some population modeling and I have a couple of questions. First, is the data item specified correctly for the following single-dose infusion type data (Inf. time=2)? Second what kind of structural model would be appropriate for these data, and is it reasonable to assume that the ADVAN1 or 3 /Trans1 or 4 could give an adequate model for such data?

For about 75 subjects, we have a single-dose infusion (Time of infusion, IDUR=2 hrs.) applied daily. I treat these data as a zero order infusion with idur=2, and I used the folowing data item for one subject:

DV is concentration in ng/mL, time is in hour, and AMT(dose) is in mg.


I tried 2 little control streams as follow:

Sub.

Dosemg

Time-hrs.

DV-ng/ml

RATE

IDUR

MDV

EVID

1

150

0

0

-1

2

1

0

1

.

0

0

.

.

0

0

1

.

0.1

.

.

.

1

0

1

.

1

1.328

.

.

0

0

1

.

2

1.68

.

.

0

0

1

.

2.083

1.28

.

.

0

0

1

.

2.167

1.2

.

.

0

0

1

.

2.33

1.024

.

.

0

0

1

.

2.5

0.936

.

.

0

0

1

.

3

0.752

.

.

0

0

1

.

4

0.472

.

.

0

0

1

.

5

0.38

.

.

0

0

1

.

8

0.18

.

.

0

0

1

.

14

0.12

.

.

0

0

2

340

0

0

-1

2

1

1

2

.

0

0

.

.

0

0

2

.

0.1

.

.

.

1

0

2

.

1

3.34

.

.

0

0

2

.

2

3.52

.

.

0

0

2

.

2.083

2.9

.

.

0

0

2

.

2.167

2.58

.

.

0

0

2

.

2.33

2.06

.

.

0

0

2

.

2.5

1.86

.

.

0

0

2

.

3

1.4

.

.

0

0

2

.

4

0.45

.

.

0

0

2

.

5

0.53

.

.

0

0

2

.

8

0.315

.

.

0

0

2

.

14

0.167

.

.

0

0

One Comp.:

 

$PROB Population PK 1 comp.

$INPUT ID AMT TIME DV RATE IDUR MDV EVID

$DATA Data_Inf

$SUBS ADVAN1 TRANS2

$PK

V=THETA(1)+ETA(1)

CL=THETA(2)+ETA(2)

 

S1=V/1000 ;where Dose is in mg and DV is in ng/ml

IF(AMT.GT.0) NRT=AMT/IDUR ;NRT is nominal rate of infusion

R1=NRT

 

$ERROR

Y=F+ERR(1)

$THETA

(0.0001,150,10000)

(0.0001,50,10000)

 

$OMEGA 0.25 0.25

$SIGMA 0.04

$EST NOABORT SIGDIGITS=4 MAXEVAL=900 POSTHOC PRINT=5

$COV

$TABLE ...

 

Two Comp.:

 

$PROB Population PK 2 comp.

$INPUT ID AMT TIME DV RATE IDUR MDV EVID

$DATA Data_Inf

$SUBS ADVAN3 TRANS4

$PK

CL=THETA(1)+ETA(1)

V1=THETA(2)+ETA(2)

Q=THETA(3)+ETA(3)

V2=THETA(4)+ETA(4)

 

S1=V1/1000 ; where Dose is in mg and DV is in ng/ml.

 

IF(AMT.GT.0) NRT=AMT/IDUR ; NRT is the nominal rate of infusion.

R1=NRT

 

$ERROR

Y=F+ERR(1)

$THETA

(0.00001,25,100000)

(0.00001,30,100000)

(0.00001,30,100000)

(0.00001,100,100000)

 

 

$OMEGA 0.04 0.04 0.04 0.04

$SIGMA 0.2

$EST NOABORT SIGDIGITS=4 MAXEVAL=900 POSTHOC PRINT=5

$COV

$TABLE ...

 

I tried some of the basic models but I am not getting expected estimations as I obtained by STS (standard two stage) or the program is terminated by round-off error, so I question whether the way the data step specified is correct or some changes have to be made on control stream(s). If someone has any advice, comments, on the above specification, or suggestions for how to model such data, I would really appreciate it.

 

Thanks in advance.

Sam T.


 

*****


From: LSheiner <lewis@c255.ucsf.edu>

Subject: Re: Data Construct / ADVAN selection

Date: Tue, 03 Apr 2001 09:34:36 -0700

 

Off the top of my head ...

 

1. Your very first record should have EVID=1, not 0. But, why don't you just drop the EVID column completely; it's not needed with this simple data (e.g. use $INPUT ID AMT TIME DV RATE IDUR MDV DROP.

 

2. Why are you modeling the infusion duration if you know it is 2 hours? More logically, for individual 1, for example, the first record should read

ID AMT TIME DV RATE IDUR MDV EVID

1 150 0 0 75 2 1 1


Note the '75' rate means the infusion will last 2 hours to deliver the total dose (AMT) of 150. For the second individual, the AMT = 340, and the RATE = 170, etc.


With this set-up you can drop the last 3 columns (IDUR MDV EVID),

$INPUT ID AMT TIME DV RATE DROP DROP DROP

as none of them are needed, and also your code simplifies to.


$PK

V=THETA(1)+ETA(1)

CL=THETA(2)+ETA(2)

S1=V/1000 ;where Dose is in mg and DV is in ng/ml

 

3. The additive etas may give you trouble in the posthoc step; it's probably safer to use


$PK

V=THETA(1)*EXP(ETA(1))

CL=THETA(2)*EXP(ETA(2))

 

4. You should probably start with a fuill OMEGA, nota diagonal one; you can then see if a diagonal one su7ffices. Thus you might start with

$OMEGA BLOCK(2) 0.25 .1 0.25

 

LBS.

 

--

_/ _/ _/_/ _/_/_/ _/_/_/ Lewis B Sheiner, MD (lewis@c255.ucsf.edu)

_/ _/ _/ _/_ _/_/ Professor: Lab. Med., Bioph. Sci., Med.

_/ _/ _/ _/ _/ Box 0626, UCSF, SF, CA, 94143-0626

_/_/ _/_/ _/_/_/ _/ 415-476-1965 (v), 415-476-2796 (fax)



*****

From: Amir Abbas Tahami Monfared <tahamima@MAGELLAN.UMontreal.CA>
Subject: Re: Data Construct / ADVAN selection
Date: Tue, 10 Apr 2001 14:11:47 -0400

Dear NONMEM users,

I used NONMEM to do the population modeling and I have a couple of questions. The data items are specified for the following single-dose infusion type (Inf. time=2). I used one and two-cmt models. The results and the goodness of fit satisfies the use of two-cmt model. However, I noticed that the estimate of VSS (VSS=theta(4)+eta(4)) for all individual is identical after POSTHOC estimation.

 

zero order infusion with inf.time=2, data for one individual:

 

Sub.

Dose

Time

DV

RATE

1

100

0

0

50

1

.

0

0

.

1

.

1

3.22

.

1

.

2

3.55

.

1

.

2.1

2.72

.

1

.

2.3

2.42

.

1

.

2.5

1.86

.

1

.

3

1.40

.

1

.

4

0.45

.

1

.

5

0.53

.

1

.

8

0.32

.

 

I tried this control stream:

 

$PROB Population PK ...

$INPUT ID AMT TIME DV RATE

$DATA Data

$SUBS ADVAN3 TRANS3

$PK

TVCL=THETA(1)

CL=TVCL+ETA(1)

TVV=THETA(2)

V=TVV+ETA(2)

TVQ=THETA(3)

Q=TVQ+ETA(3)

TVVSS=THETA(4)

VSS=TVVSS+ETA(4)

 

S1=V

 

$ERROR

Y=F+ERR(1)

$THETA

(0.1,30)

(0.1,25)

(0.1,50)

(0.1,100)

 

$OMEGA 0.25 0.25 0.25 0.25

$SIGMA 0.04

$EST SIGDIGITS=4 MAXEVAL=999 POSTHOC PRINT=5

$COV

$TABLE ID TIME CL V Q VSS

$SCAT TIME VS DV

$SCAT ...

 

I would also like to add that the estimates for volume of distribution in central cmt. and VSS are half of the mean estimate obtained by STS method. If someone has any advice, comments, on the above specification, or suggestions for how to model such data, I would really appreciate it.

 

Thanks in advance.

Amir A. Tahami



*****

From: Matthew Riggs <riggsmm@yahoo.com>

Subject: Re: Data Construct / ADVAN selection

Date: Wed, 11 Apr 2001 05:58:38 -0700 (PDT)

 

Amir,

I would start by referring you to recent answer provided by Lewis Sheiner to a similar question (copied below)... "

 

3. The additive etas may give you trouble in the posthoc step; it's probably safer to use


$PK

V=THETA(1)*EXP(ETA(1))

CL=THETA(2)*EXP(ETA(2))

 

4. You should probably start with a full OMEGA, not a diagonal one; you can then see if a diagonal one suffices. Thus you might start with

 

$OMEGA BLOCK(2) 0.25 .1 0.25

 

LBS."

 

Additionally, with this relatively rich data set you may want to try the first-order conditional estimation method (Method=1) with interaction versus the default first-order method. In your case, you may change the default using:

 

$EST SIGDIGITS=4 MAXEVAL=999 METHOD=1 INTERACTION

PRINT=5

 

My final suggestion would be to define your typical value (poulation mean) VSS as TVSS, rather than TVVSS. NONMEM will only read 4 letters of a user defined variable. This may help your Bayesian estimated (POSTHOC) individual VSS estimates. Another diagnostic: Was the gradient for ETA4 zero? Or was the final estimate for ETA4 zero or the same as the initial estimate?

 

Hope this helps.

Matthew Riggs