Dear fellow NMusers,
I'm trying to develop an integrated parent-metabolite PK model based on the
plasma concentrations obtained from 40 subjects after sequential iv and oral
administrations of the parent drug. The iv bolus was injected at time = 0
while the oral dose was administered at time = 4h. The most stable model so
far is a two-compartment model for the parent drug (with first-order
absorption) linked to a one-compartment metabolite model. However, in order to
obtain reasonable parameter estimates and adequate goodness-of-fit plots, I
needed to specify a "switch" in rate constant values and scaling factors at
time = 4h (oral administration time), as defined in the control stream below.
Without, this "switch", there was a very poor match between the measured and
predicted metabolite concentration values.
May I ask if my code below is correct for modelling the data at hand (assuming
the abovementioned structural model)? If not, would you be able to suggest
possible areas of model improvement?
Thank you and best wishes,
Kok-Yong SENG, PhD
DSO National Laboratories
Singapore
______________________
$PROB RUN# 1 parent-met data (nmol/L vs h)
$INPUT ID NTIM TIME DV AMT ORAL CMT EVID MDV
$DATA DATA.CSV IGNORE=C
$SUBROUTINE ADVAN5
$MODEL
NCOMP=4
COMP=(DEPOT)
COMP=(CENTRAL, DEFDOSE)
COMP=(METAB1)
COMP=(PERIPM)
$PK
IIVCL = ETA(1)
IIVF1 = ETA(2)
IIVKA = ETA(3)
IIVCL1 = ETA(4)
IIVV3 = ETA(5)
KA = THETA(1)*EXP(IIVKA)
CL = THETA(2)*EXP(IIVCL)
V2 = THETA(3)
Q = THETA(4)
V4 = THETA(5)
F1 = THETA(6)*EXP(IIVF1)
FMET = 0.6
CL1 = THETA(7)*EXP(IIVCL1)
V3 = THETA(8)*EXP(IIVV3)
IF (TIME.LE.4) THEN
S2 = V2
S3 = V3
K12 = KA
K20 = (1-FMET)*(CL/V2)
K23 = FMET*(CL/V2)
K24 = Q/V2
K42 = Q/V4
K30 = CL1/V3
ELSE
S2 = V2/F1
S3 = V3
K12 = KA
K20 = (1-FMET)*((CL/F1)/(V2/F1))
K23 = FMET*((CL/F1)/(V2/F1))
K24 = (Q/F1)/(V2/F1)
K42 = (Q/F1)/(V4/F1)
K30 = CL1/V3
ENDIF
$ERROR ;ONLY OBSERVATIONS)
IF (AMT.GT.0) TDOS=TIME
TAD=TIME-TDOS
IF (F.EQ.0) THEN
IPRED=0
ELSE
IPRED=LOG(F)
ENDIF
IRES=DV-IPRED
IF(CMT.EQ.2) W2 = SQRT(THETA(9)**2 )
IF(CMT.EQ.3) W3 = SQRT(THETA(10)**2 )
IF(CMT.EQ.2) IWRES=IRES/W2
IF(CMT.EQ.3) IWRES=IRES/W3
IF(CMT.EQ.2) Y = IPRED + (W2)*EPS(1) ;
IF(CMT.EQ.3) Y = IPRED + (W3)*EPS(2)
$THETA
$OMEGA
$SIGMA
1 FIX
1 FIX
$EST METHOD=1 INTER MAXEVAL=9999 PRINT=1 POSTHOC NOABORT MSFO=1.MSF SIGL=9
NSIG=3
$COV UNCONDITIONAL PRINT=E
:::::DATA.csv::::: (for first two subjects)
C Data Desc: parent-met data (dose in nmoles; time in hr; conc in nmol/L)
CID NTIM TIME DV AMT ORAL CMT EVID MDV
101 0 0 0 2302.308 . 2 1 1
101 0 0 0 2302.308 . 2 1 1
101 0.25 0.25 5.430537181 . . 2 0 0
101 0.25 0.25 0.741358053 . . 3 0 0
101 0.5 0.5 4.799111134 . . 2 0 0
101 0.5 0.5 0.854669023 . . 3 0 0
101 0.75 0.75 4.55617397 . . 2 0 0
101 0.75 0.75 0.691490788 . . 3 0 0
101 1 1 4.106244674 . . 2 0 0
101 1 1 0.589508916 . . 3 0 0
101 1.5 1.5 3.822383616 . . 2 0 0
101 1.5 1.5 0.409972825 . . 3 0 0
101 2 2 3.652067835 . . 2 0 0
101 2 2 0.284763877 . . 3 0 0
101 2.5 2.5 3.346291176 . . 2 0 0
101 2.5 2.5 0.08938465 . . 3 0 0
101 3 3 3.121642304 . . 2 0 0
101 3 3 -0.090114893 . . 3 0 0
101 3.5 3.5 3.278754426 . . 2 0 0
101 3.5 3.5 -0.108362666 . . 3 0 0
101 4 4 . 4604.619 1 1 1 1
101 4 4 . 4604.619 1 1 1 1
101 4 4 3.44923847 . 1 2 0 0
101 4 4 -0.071713115 . 1 3 0 0
101 4.25 4.25 3.597385022 . 1 2 0 0
101 4.25 4.25 -0.048409056 . 1 3 0 0
101 4.5 4.5 3.716095668 . 1 2 0 0
101 4.5 4.5 1.175372119 . 1 3 0 0
101 4.75 4.75 4.154635181 . 1 2 0 0
101 4.75 4.75 2.212447576 . 1 3 0 0
101 5 5 4.08855346 . 1 2 0 0
101 5 5 2.253719503 . 1 3 0 0
101 5.5 5.5 3.726503911 . 1 2 0 0
101 5.5 5.5 1.86297201 . 1 3 0 0
101 6 6 3.52409438 . 1 2 0 0
101 6 6 1.48545147 . 1 3 0 0
101 7 7 -0.791943487 . 1 3 0 0
101 8 8 2.861295756 . 1 2 0 0
101 8 8 0.195365359 . 1 3 0 0
101 10 10 2.960551062 . 1 2 0 0
101 10 10 -0.709552271 . 1 3 0 0
101 12 12 2.164593204 . 1 2 0 0
101 12 12 -0.857712813 . 1 3 0 0
integrated parent-metabolite PK model from sequential iv and oral doses of parent drug
Dear Kok-Yong Seng,
Is the drug undergoing presystemic metabolism after oral administration? This
might explain some differences in metabolic rate constants before and after
oral administration. If your drug undergoes presystemic metabolism, metabolite
concentrations may appear earlier in the systemic circulation than you would
have expected based on your current PK model. A work-around is to use a more
mechanistic-based model that incorporates both systemic and presystemic
metabolism, e.g. as described by Levi et al in J Pharmacokinet Pharmacodyn.
2007 Feb;34(1):5-21..
Cheers,
Rob ter Heine
---
R. ter Heine, PhD, PharmD
Hospital Pharmacist / Clinical Scientist / Clinical Pharmacologist
Meander Medical Center, Amersfoort, The Netherlands
T: +31-33-8502335
E: [email protected]
-----Oorspronkelijk bericht-----
Quoted reply history
Van: [email protected] [mailto:[email protected]] Namens
Seng Kok Yong
Verzonden: maandag 17 februari 2014 9:22
Aan: [email protected]
Onderwerp: [NMusers] integrated parent-metabolite PK model from sequential iv
and oral doses of parent drug
Dear fellow NMusers,
I'm trying to develop an integrated parent-metabolite PK model based on the
plasma concentrations obtained from 40 subjects after sequential iv and oral
administrations of the parent drug. The iv bolus was injected at time = 0
while the oral dose was administered at time = 4h. The most stable model so
far is a two-compartment model for the parent drug (with first-order
absorption) linked to a one-compartment metabolite model. However, in order to
obtain reasonable parameter estimates and adequate goodness-of-fit plots, I
needed to specify a "switch" in rate constant values and scaling factors at
time = 4h (oral administration time), as defined in the control stream below.
Without, this "switch", there was a very poor match between the measured and
predicted metabolite concentration values.
May I ask if my code below is correct for modelling the data at hand (assuming
the abovementioned structural model)? If not, would you be able to suggest
possible areas of model improvement?
Thank you and best wishes,
Kok-Yong SENG, PhD
DSO National Laboratories
Singapore
______________________
$PROB RUN# 1 parent-met data (nmol/L vs h) $INPUT ID NTIM TIME DV AMT ORAL CMT
EVID MDV $DATA DATA.CSV IGNORE=C
$SUBROUTINE ADVAN5
$MODEL
NCOMP=4
COMP=(DEPOT)
COMP=(CENTRAL, DEFDOSE)
COMP=(METAB1)
COMP=(PERIPM)
$PK
IIVCL = ETA(1)
IIVF1 = ETA(2)
IIVKA = ETA(3)
IIVCL1 = ETA(4)
IIVV3 = ETA(5)
KA = THETA(1)*EXP(IIVKA)
CL = THETA(2)*EXP(IIVCL)
V2 = THETA(3)
Q = THETA(4)
V4 = THETA(5)
F1 = THETA(6)*EXP(IIVF1)
FMET = 0.6
CL1 = THETA(7)*EXP(IIVCL1)
V3 = THETA(8)*EXP(IIVV3)
IF (TIME.LE.4) THEN
S2 = V2
S3 = V3
K12 = KA
K20 = (1-FMET)*(CL/V2)
K23 = FMET*(CL/V2)
K24 = Q/V2
K42 = Q/V4
K30 = CL1/V3
ELSE
S2 = V2/F1
S3 = V3
K12 = KA
K20 = (1-FMET)*((CL/F1)/(V2/F1))
K23 = FMET*((CL/F1)/(V2/F1))
K24 = (Q/F1)/(V2/F1)
K42 = (Q/F1)/(V4/F1)
K30 = CL1/V3
ENDIF
$ERROR ;ONLY OBSERVATIONS)
IF (AMT.GT.0) TDOS=TIME
TAD=TIME-TDOS
IF (F.EQ.0) THEN
IPRED=0
ELSE
IPRED=LOG(F)
ENDIF
IRES=DV-IPRED
IF(CMT.EQ.2) W2 = SQRT(THETA(9)**2 )
IF(CMT.EQ.3) W3 = SQRT(THETA(10)**2 )
IF(CMT.EQ.2) IWRES=IRES/W2
IF(CMT.EQ.3) IWRES=IRES/W3
IF(CMT.EQ.2) Y = IPRED + (W2)*EPS(1) ;
IF(CMT.EQ.3) Y = IPRED + (W3)*EPS(2)
$THETA
$OMEGA
$SIGMA
1 FIX
1 FIX
$EST METHOD=1 INTER MAXEVAL=9999 PRINT=1 POSTHOC NOABORT MSFO=1.MSF SIGL=9
NSIG=3 $COV UNCONDITIONAL PRINT=E
:::::DATA.csv::::: (for first two subjects)
C Data Desc: parent-met data (dose in nmoles; time in hr; conc in nmol/L)
CID NTIM TIME DV AMT ORAL CMT EVID MDV
101 0 0 0 2302.308 . 2 1 1
101 0 0 0 2302.308 . 2 1 1
101 0.25 0.25 5.430537181 . . 2 0 0
101 0.25 0.25 0.741358053 . . 3 0 0
101 0.5 0.5 4.799111134 . . 2 0 0
101 0.5 0.5 0.854669023 . . 3 0 0
101 0.75 0.75 4.55617397 . . 2 0 0
101 0.75 0.75 0.691490788 . . 3 0 0
101 1 1 4.106244674 . . 2 0 0
101 1 1 0.589508916 . . 3 0 0
101 1.5 1.5 3.822383616 . . 2 0 0
101 1.5 1.5 0.409972825 . . 3 0 0
101 2 2 3.652067835 . . 2 0 0
101 2 2 0.284763877 . . 3 0 0
101 2.5 2.5 3.346291176 . . 2 0 0
101 2.5 2.5 0.08938465 . . 3 0 0
101 3 3 3.121642304 . . 2 0 0
101 3 3 -0.090114893 . . 3 0 0
101 3.5 3.5 3.278754426 . . 2 0 0
101 3.5 3.5 -0.108362666 . . 3 0 0
101 4 4 . 4604.619 1 1 1 1
101 4 4 . 4604.619 1 1 1 1
101 4 4 3.44923847 . 1 2 0 0
101 4 4 -0.071713115 . 1 3 0 0
101 4.25 4.25 3.597385022 . 1 2 0 0
101 4.25 4.25 -0.048409056 . 1 3 0 0
101 4.5 4.5 3.716095668 . 1 2 0 0
101 4.5 4.5 1.175372119 . 1 3 0 0
101 4.75 4.75 4.154635181 . 1 2 0 0
101 4.75 4.75 2.212447576 . 1 3 0 0
101 5 5 4.08855346 . 1 2 0 0
101 5 5 2.253719503 . 1 3 0 0
101 5.5 5.5 3.726503911 . 1 2 0 0
101 5.5 5.5 1.86297201 . 1 3 0 0
101 6 6 3.52409438 . 1 2 0 0
101 6 6 1.48545147 . 1 3 0 0
101 7 7 -0.791943487 . 1 3 0 0
101 8 8 2.861295756 . 1 2 0 0
101 8 8 0.195365359 . 1 3 0 0
101 10 10 2.960551062 . 1 2 0 0
101 10 10 -0.709552271 . 1 3 0 0
101 12 12 2.164593204 . 1 2 0 0
101 12 12 -0.857712813 . 1 3 0 0
***************************DISCLAIMER****************************
De informatie in dit e-mail bericht is uitsluitend bestemd
voor de geadresseerde. Verstrekking aan en gebruik door
anderen is niet toegestaan. Door de elektronische verzending
van het bericht kunnen er geen rechten worden ontleend aan de
informatie.
Kok-Yong
The code is too complicated (just a technical comment) and in fact is equivalent to
.....
F1 = THETA(6)*EXP(IIVF1)
.....
S2 = V2
...
IF (TIME.GT.4) THEN S2 = V2/F1
...
as all the other parameters are identical in both parts of the IF() block.
As is, the code allows F1 > 1 that would be unusual. If you would like to keep individual F1 values below 1 it is better to code it as
F1 = 1/(1+EXP(THETA(6)*EXP(IIVF1)))
(where THETA(6) is allowed to be both positive and negative)
Also, why do you have two dose records for each dose? Why FMET is fixed to 0.6, was it estimated?
The main problem is S2 = V2/F1 part that is not mechanistic at all. What exactly was wrong without this part, were predicted metabolite concentrations too low or to high?
Have you seen this work:
http://tucson2008.go-acop.org/pdfs/51_gieschke.pdf
that deals with a very similar problem?
Note that the model in the poster above allows some part of the oral dose to bypass the parent compartment, and also it has an intermediate compartment (delay between parent and metabolite) which also can be helpful in fitting the data.
Regards,
Leonid
--------------------------------------
Leonid Gibiansky, Ph.D.
President, QuantPharm LLC
web: www.quantpharm.com
e-mail: LGibiansky at quantpharm.com
tel: (301) 767 5566
Quoted reply history
On 2/17/2014 3:22 AM, Seng Kok Yong wrote:
> Dear fellow NMusers,
>
> I'm trying to develop an integrated parent-metabolite PK model based on the plasma concentrations
> obtained from 40 subjects after sequential iv and oral administrations of the parent drug. The iv
> bolus was injected at time = 0 while the oral dose was administered at time = 4h. The most stable
> model so far is a two-compartment model for the parent drug (with first-order absorption) linked to
> a one-compartment metabolite model. However, in order to obtain reasonable parameter estimates and
> adequate goodness-of-fit plots, I needed to specify a "switch" in rate constant values
> and scaling factors at time = 4h (oral administration time), as defined in the control stream
> below. Without, this "switch", there was a very poor match between the measured and
> predicted metabolite concentration values.
>
> May I ask if my code below is correct for modelling the data at hand (assuming
> the abovementioned structural model)? If not, would you be able to suggest
> possible areas of model improvement?
>
> Thank you and best wishes,
> Kok-Yong SENG, PhD
> DSO National Laboratories
> Singapore
> ______________________
>
> $PROB RUN# 1 parent-met data (nmol/L vs h)
> $INPUT ID NTIM TIME DV AMT ORAL CMT EVID MDV
> $DATA DATA.CSV IGNORE=C
>
> $SUBROUTINE ADVAN5
>
> $MODEL
> NCOMP=4
> COMP=(DEPOT)
> COMP=(CENTRAL, DEFDOSE)
> COMP=(METAB1)
> COMP=(PERIPM)
>
> $PK
> IIVCL = ETA(1)
> IIVF1 = ETA(2)
> IIVKA = ETA(3)
> IIVCL1 = ETA(4)
> IIVV3 = ETA(5)
> KA = THETA(1)*EXP(IIVKA)
> CL = THETA(2)*EXP(IIVCL)
> V2 = THETA(3)
> Q = THETA(4)
> V4 = THETA(5)
> F1 = THETA(6)*EXP(IIVF1)
>
> FMET = 0.6
>
> CL1 = THETA(7)*EXP(IIVCL1)
> V3 = THETA(8)*EXP(IIVV3)
>
> IF (TIME.LE.4) THEN
>
> S2 = V2
> S3 = V3
> K12 = KA
> K20 = (1-FMET)*(CL/V2)
> K23 = FMET*(CL/V2)
> K24 = Q/V2
> K42 = Q/V4
> K30 = CL1/V3
>
> ELSE
>
> S2 = V2/F1
> S3 = V3
> K12 = KA
> K20 = (1-FMET)*((CL/F1)/(V2/F1))
> K23 = FMET*((CL/F1)/(V2/F1))
> K24 = (Q/F1)/(V2/F1)
> K42 = (Q/F1)/(V4/F1)
> K30 = CL1/V3
>
> ENDIF
>
> $ERROR ;ONLY OBSERVATIONS)
>
> IF (AMT.GT.0) TDOS=TIME
> TAD=TIME-TDOS
>
> IF (F.EQ.0) THEN
> IPRED=0
> ELSE
> IPRED=LOG(F)
> ENDIF
> IRES=DV-IPRED
>
> IF(CMT.EQ.2) W2 = SQRT(THETA(9)**2 )
> IF(CMT.EQ.3) W3 = SQRT(THETA(10)**2 )
> IF(CMT.EQ.2) IWRES=IRES/W2
> IF(CMT.EQ.3) IWRES=IRES/W3
> IF(CMT.EQ.2) Y = IPRED + (W2)*EPS(1) ;
> IF(CMT.EQ.3) Y = IPRED + (W3)*EPS(2)
>
> $THETA
> $OMEGA
> $SIGMA
> 1 FIX
> 1 FIX
>
> $EST METHOD=1 INTER MAXEVAL=9999 PRINT=1 POSTHOC NOABORT MSFO=1.MSF SIGL=9
> NSIG=3
> $COV UNCONDITIONAL PRINT=E
>
> :::::DATA.csv::::: (for first two subjects)
> C Data Desc: parent-met data (dose in nmoles; time in hr; conc in nmol/L)
>
> CID NTIM TIME DV AMT ORAL CMT EVID MDV
> 101 0 0 0 2302.308 . 2 1 1
> 101 0 0 0 2302.308 . 2 1 1
> 101 0.25 0.25 5.430537181 . . 2 0 0
> 101 0.25 0.25 0.741358053 . . 3 0 0
> 101 0.5 0.5 4.799111134 . . 2 0 0
> 101 0.5 0.5 0.854669023 . . 3 0 0
> 101 0.75 0.75 4.55617397 . . 2 0 0
> 101 0.75 0.75 0.691490788 . . 3 0 0
> 101 1 1 4.106244674 . . 2 0 0
> 101 1 1 0.589508916 . . 3 0 0
> 101 1.5 1.5 3.822383616 . . 2 0 0
> 101 1.5 1.5 0.409972825 . . 3 0 0
> 101 2 2 3.652067835 . . 2 0 0
> 101 2 2 0.284763877 . . 3 0 0
> 101 2.5 2.5 3.346291176 . . 2 0 0
> 101 2.5 2.5 0.08938465 . . 3 0 0
> 101 3 3 3.121642304 . . 2 0 0
> 101 3 3 -0.090114893 . . 3 0 0
> 101 3.5 3.5 3.278754426 . . 2 0 0
> 101 3.5 3.5 -0.108362666 . . 3 0 0
> 101 4 4 . 4604.619 1 1 1 1
> 101 4 4 . 4604.619 1 1 1 1
> 101 4 4 3.44923847 . 1 2 0 0
> 101 4 4 -0.071713115 . 1 3 0 0
> 101 4.25 4.25 3.597385022 . 1 2 0 0
> 101 4.25 4.25 -0.048409056 . 1 3 0 0
> 101 4.5 4.5 3.716095668 . 1 2 0 0
> 101 4.5 4.5 1.175372119 . 1 3 0 0
> 101 4.75 4.75 4.154635181 . 1 2 0 0
> 101 4.75 4.75 2.212447576 . 1 3 0 0
> 101 5 5 4.08855346 . 1 2 0 0
> 101 5 5 2.253719503 . 1 3 0 0
> 101 5.5 5.5 3.726503911 . 1 2 0 0
> 101 5.5 5.5 1.86297201 . 1 3 0 0
> 101 6 6 3.52409438 . 1 2 0 0
> 101 6 6 1.48545147 . 1 3 0 0
> 101 7 7 -0.791943487 . 1 3 0 0
> 101 8 8 2.861295756 . 1 2 0 0
> 101 8 8 0.195365359 . 1 3 0 0
> 101 10 10 2.960551062 . 1 2 0 0
> 101 10 10 -0.709552271 . 1 3 0 0
> 101 12 12 2.164593204 . 1 2 0 0
> 101 12 12 -0.857712813 . 1 3 0 0