From: "Panetta, Carl" <Carl.Panetta@stjude.org>
Subject: Question on Bias est. and the conditional method
Date: Wed, 10 Jan 2001 10:26:02 -0600

I am relatively new to using NONMEM and have several questions about the results I am obtaining and errors I am getting. The data set that I am working with contains 182 individuals who received 5 courses of I.V. MTX (currently I am working with each course separately). There are approximately 4 samples per individual. First, I am just trying to estimate the population PK. Below is attached a copy of the NONMEM file I am using.

1.)Using the first-order method (using the first $ESTIMATION line in the file) everything runs well, but the resulting population PK parameters are between 10 and 20% below the estimates obtained when using a two-step method with Bayesian estimation. Also, the variance (inter-individual variability) estimates are lower by as much as about half. Is this to be expected with the first-order method, or am I coding something wrong in the file?

2.)Using the conditional method (using the second $ESTIMATION line in the file) I am unable to get NONMEM to successfully run. If I use additive etas I get the following error:

______________________________
Starting nonmem execution ...
MONITORING OF SEARCH:

0PRED EXIT CODE = 1

0INDIVIDUAL NO. 23 ID=0.10988500E+06 (WITHIN-INDIVIDUAL) DATA REC NO.
2

THETA=

1.00E+02 9.40E+00 8.00E-02 5.00E-02

OCCURS DURING SEARCH FOR ETA AT A NONZERO VALUE OF ETA

PK PARAMETER FOR K21 IS NON-POSITIVE

1 file(s) copied
______________________________
If I use mult. etas I get the following error:
______________________________

Starting nonmem execution ...
MONITORING OF SEARCH:

0ITERATION NO.: 0 OBJECTIVE VALUE: 0.2418E+04 NO. OF FUNC.
EVALS.: 7

CUMULATIVE NO. OF FUNC. EVALS.: 7
PARAMETER: 0.1000E+00 0.1000E+00 0.1000E+00 0.1000E+00 0.1000E+00
0.1000E
+00 0.1000E+00 0.1000E+00 0.1000E+00
GRADIENT: -0.3924E+04 0.3541E+04 0.3452E+04 -0.3465E+04 0.4125E+03
-0.6664E
+02 -0.4546E+03 -0.1414E+02 0.2731E+04
forrtl: error (72): floating overflow
1 file(s) copied
______________________________
Is there some option that I should be using so that this will run?

______________________________

NONMEM calling file.
______________________________
$PROBLEM T12 POPULATION DATA
$INPUT ID TIME CP=DV DOSE=AMT RATE AGE SEX RACE BSA W H AC TX DAT
$DATA pk_t12_nonmem_mtx501.dat
$SUBROUTINES ADVAN3

$PK
CALLFL=1

TCL=THETA(1)
TV=THETA(2)

;additive eta
CL=TCL+ETA(1)
SC=TV+ETA(2)
K12=THETA(3)+ETA(3)
K21=THETA(4)+ETA(4)

;Mult. eta
;CL=TCL*EXP(ETA(1))
;SC=TV*EXP(ETA(2))
;K12=THETA(3)*EXP(ETA(3))
;K21=THETA(4)*EXP(ETA(4))

K=(3.0/50.0)*CL/SC

$ERROR
Y=F*(1.0+ERR(1))
;Y=F+ERR(1)

$SIGMA 0.1

$THETA (0.07 100.0) (0.9 9.4) (0.0008 0.08) (0.00011 0.05)

;additive eta
$OMEGA 250.0 2.0 0.01 0.01
;mult. eta
;$OMEGA 0.0685 0.083 0.303 0.0013


$ESTIMATION MAXEVAL=9999 NOABORT PRINT=25 SIGDIGITS=3
;$ESTIMATION METHOD=1 NOABORT MAXEVAL=9999 PRINT=25
$COVR
$TABLE ID SC K K12 K21
FIRSTONLY NOAPPEND NOPRINT FILE=t12par.out
$TABLE ID TIME AGE SEX RACE AC ONEHEADER NOPRINT FILE=t12.out
$SCAT RES VS TIME
$SCAT CP VS PRED UNIT
$SCAT AGE VS RES
$SCAT SEX VS RES
$SCAT RACE VS RES
$SCAT AC VS RES
______________________________________________________

J. Carl Panetta, Ph.D.
Biomedical Modeler
Department of Pharmaceutical Sciences
St. Jude Children's Research Hospital
332 North Lauderdale St.
Memphis, TN 38105-2794
Phone: (901) 495-3172
FAX: (901) 525-6869
e-mail: carl.panetta@stjude.org


*****


From: ABoeckmann <alison@c255.ucsf.edu>
Subject: Re: Question on Bias est. and the conditional method
Date: Thu, 11 Jan 2001 11:15:36 -0800 (PST)

Carl,

I don't know why the first order method gives low estimates. However, it is not surprising that conditional estimation runs into difficulties.

The models for K12 and K21 are:
K12=THETA(3)+ETA(3)
K21=THETA(4)+ETA(4)
or
K12=THETA(3)*EXP(ETA(3))
K21=THETA(4)*EXP(ETA(4))

The initial estimates for these thetas are 0.08 and 0.05. With additive eta, it is quite likely that a negative eta will give rise to negative K12 or K21. The mult. eta models are likely to work better.

The overflow is probably due to a large value of eta, leading to a value of EXP(ETA(.)) that causes an overflow.

(It may be that the data for INDIVIDUAL NO. 23 ID=0.10988500E+06 is especially problematic.)

In the help item for abbreviated code, we describe the EXIT statement, that can be used to tell NONMEM to avoid certain vaues of ETA. I would put these statements in the $PK blocK;

$PK
IF (ABS(ETA(1)).GT.10) EXIT 1 100
IF (ABS(ETA(2)).GT.10) EXIT 1 200
IF (ABS(ETA(3)).GT.10) EXIT 1 300
IF (ABS(ETA(4)).GT.10) EXIT 1 400

With the NOABORT option of $EST, NONMEM will attempt to avoid these large etas and will try to continue. The PRDERR file will contain messages describing which condition(s) occured. E.g., if the user error code is 300, the trouble was with eta 3, and so on. Usually the exit conditions will not persist beyond the first few iterations.

- Alison Boeckmann
NONMEM Project Group