Re: [NMusers] NONMEM/PsN benchmark for SGE expansion

[Paul Matthias Diderichsen]

Hi Julia,

On 3/25/2011 4:42 PM, Ivashina, Julia wrote:

* What type of analysis is the most sensitive to an increase in
  number of work nodes?


As far as I know, the version of nonmem that you have installed does not 
support parallel processing. Legend has it that this feature will show 
up in the next version of nonmem (a beta version is said to exist... ;-) ).



* What should be the expected gain from increasing the number in
  -threads 50 times?


As you've experienced this depends on a number of factors; your 
distributed computing facility will bring you the largest benefits with 
very CPU intensive jobs. This could be models with very large datasets 
or PKPD models relying on numerical integration (ADVAN6, ADVAN8, ADVAN13).


Smaller runs will spend a comparatively higher proportion of the runtime 
compiling nonmem and pushing data across the network. The way you've set 
up the grid file system plays a role here too. I'm not too surprised 
that you don't see much benefit when runtimes are as short as 5-10 seconds.



* What parts of NONMEM/PsN are the most optimized for parallel
  execution?


Bootstrap and VPC/NPC are the scripts that are most suited for running 
on a grid. Use the command line parameters -samples and -threads with 
the boostrap command, and -samples and -n_simulation_models (and 
possibly -threads) with VPC/NPC.


See: http://psn.sourceforge.net/pdfdocs/npc_vpc_userguide.pdf and 
http://psn.sourceforge.net/pdfdocs/bootstrap_userguide.pdf



* What are the scenarios where gain from parallelization is the biggest?


In my experience, your gains will be biggest when executing many 
long-running jobs (eg a complete model history, a bootstrap, or a 
NPC/VPC with the -n_simulation_models parameters)


Kind regards,
--
Paul Matthias Diderichsen, PhD
Quantitative Solutions B.V.
+31 624 330 706

RE: [NMusers] Time-to-event analysis with $DES

[Andrew Hooker]

Hi All

 

I would just like to point out that in the Weibull hazard that Nick
proposes:

 

HAZNOW=LAMD*ALPH*(TIME+DEL2)**(ALPH-1)

 

The units of LAMD would have to be TIME**(-ALPH).  That is, as soon as ALPH
changes then the units of LAMD change.  This seems strange and unnecessary
to include this type of correlation in the parameters.  If we instead
reparameterize the model to:

 

  HAZNOW=LAMD*ALPH*(LAMD*(TIME+DEL2))**(ALPH-1)

 

Then LAMD has units of 1/TIME no matter the value of ALPH.  I have tested
the two parameterizations in one simple simulation example and found that
the relative standard errors were significantly lower with this second
parameterization.

 

Best regards

Andy

 

 

Andrew Hooker, Ph.D.

Associate Professor of Pharmacometrics

Dept. of Pharmaceutical Biosciences

Uppsala University

Box 591, 751 24, Uppsala, Sweden

Phone: +46 18 471 4355

 
www.farmbio.uu.se/research/Research+groups/Pharmacometrics/ 

 

From: owner-nmus...@globomaxnm.com [mailto:owner-nmus...@globomaxnm.com] On
Behalf Of Nick Holford
Sent: den 24 mars 2011 07:54
To: nmusers
Subject: Re: [NMusers] Time-to-event analysis with $DES

 

Hyewon,

Sorry -- I just realized that EFF is not the drug effect but 1-drug effect
so that it has a value of 1 when AUC is zero. Please ignore my remarks about
EFF being zero when AUC is zero in my comments below and consider this code
to describe the effect of AUC on hazard:

 HAZNOW=LAMD*ALPH*(TIME+DEL2)**(ALPH-1)*(1-BETAE*AUC/(EC+AUC)) 

I think it is a very strong assumption that the drug will reduce the hazard
to zero at infinite AUC which you can test by estimating a parameter (such
as BETAE). If BETAE is significantly less than 1 then this means the
assumption is not supported.

Nick

On 24/03/2011 8:17 a.m., Nick Holford wrote: 

Hyewon,

The most obvious problem with your code is in $DES. You must use the
variable T not the variable TIME when referring to a time varying hazard.
TIME is the time on each data record. It does not change in $DES. T is the
time from the last data record upto the current data record and changes
within $DES.

You are predicting the likelihood of an event at the exact time of the event
observation record with multiple events per subject. As you seem to realize
you need to compute the cumulative hazard (CUMHAZ) either from TIME=0 or
from the TIME of the last non-censored event for each subject.

In my opinion the following code is clearer and less dependent on your data
structure than the method you are using which works only if your data has
just event records after the TIME=0 record. 

IF (MDV.EQ.0.AND.CS.EQ.0) THEN
   OLDCHZ=A(1) ; cum haz upto time of this event
ELSE
  OLDCHZ=OLDCHZ ; need to do this if OLDCHZ is a random variable
ENDIF

Your hazard model looks rather complicated. It seems to be based on the
product of a Weibull  baseline hazard 
LAMD*ALPH*(TIME+DEL2)**(ALPH-1)
then something odd involving the Weibull parameters
*EXP(-LAMD*(TIME+DEL2)**ALPH)
and then forces the hazard to be zero if AUC is zero. 
*EFF
Is this really what you want? Do you know the hazard of event is zero if AUC
is zero? Or is the last right parenthesis in the wrong place?

I would suggest something like this where LAMD and ALPH are the two
parameters of the Weibull baseline hazard (when AUC is zero) and BETAE is a
parameter describing the effect of the drug on the overall hazard.

 HAZNOW=LAMD*ALPH*(TIME+DEL2)**(ALPH-1)*EXP(BETAE*EFF) 

You may also find it easier to develop the model if you do not try to
estimate a random effect on LAMD until you have got reasonable estimates for
the other parameters. 

You may find it helpful to look at this presentation showing how to code
time to event models in NM-TRAN:
http://pkpdrx.com/holford/docs/time-to-event-analysis.pdf

Best wishes,

Nick

On 24/03/2011 3:25 a.m., Hyewon Kim wrote: 

Dear NMuser 

I am trying to analyze time to repeated event data using NONMEM. 
The response were obtained till 24 hours after drug administration. 
Inhibitory Emax model was implemented. 
I am getting unreasonable parameter estimates which is far beyond what data
say. 
If some body can point out what i am doing wrong, it would be very helpful. 
Thank you in advance. 

Hyewon 


Data set (# of observations =62, # of patients=50 ) 
CIDTIME CS MDVAUC 
.  101 0.11.111 
.  101 0.05   00   1.111 
.  101 2   001.111 
.  102 0.1 0 
.  102 24 100 
.  103 0.10.999 
.  103 0.75  . 00.999 
 

Model File 
$PROB RUN# 101 
$INPUT C ID TIME CS MDV AUC 
;CS:0=having event,1=censo

Re: [NMusers] Time-to-event analysis with $DES

[Nick Holford]

Andy,

Thanks for this response. In fact I did not propose that particular 
Weibull parameterization but simply pointed out that it seemed to be 
what the original questioner seemed to be trying to use.


My own preference for parameterizing the Weibull is

  HAZARD=LAMDA0*EXP(BETA*LOG(TIME))


This parameterization makes is easier to understand that the Weibull 
(ln(time)) is just a transformation of the Gompertz (time):



  HAZARD=LAMDA0*EXP(BETA*TIME)


Its then easy to extend hazards to more interesting and possibly more 
useful hazards such as the Gombull (aka the Weipertz):



  HAZARD=LAMDA0*EXP(BETAG*TIME + BETAW*LOG(TIME))


One of the features of the Weibull that I dont like is that the hazard 
is predicted to be zero when time is zero. This seems a particularly 
unlikely biological occurrence and also requires extra coding to deal 
with this special case when time is zero to avoid NaNs.


I'd be interested in knowing if you find any difference in the standard 
errors with the ln(time) parameterization. But whatever you find it 
won't change my behaviour because as you know standard errors are not 
much use for making predictions :-)



Best wishes,


Nick



On 28/03/2011 3:05 p.m., Andrew Hooker wrote:


Hi All

I would just like to point out that in the Weibull hazard that Nick 
proposes:


HAZNOW=LAMD*ALPH*(TIME+DEL2)**(ALPH-1)

The units of LAMD would have to be TIME**(-ALPH).  That is, as soon as 
ALPH changes then the units of LAMD change.  This seems strange and 
unnecessary to include this type of correlation in the parameters.  If 
we instead reparameterize the model to:


  HAZNOW=LAMD*ALPH*(LAMD*(TIME+DEL2))**(ALPH-1)

Then LAMD has units of 1/TIME no matter the value of ALPH.  I have 
tested the two parameterizations in one simple simulation example and 
found that the relative standard errors were significantly lower with 
this second parameterization.


Best regards

Andy

Andrew Hooker, Ph.D.

Associate Professor of Pharmacometrics

Dept. of Pharmaceutical Biosciences

Uppsala University

Box 591, 751 24, Uppsala, Sweden

Phone: +46 18 471 4355

www.farmbio.uu.se/research/Research+groups/Pharmacometrics/ 



*From:*owner-nmus...@globomaxnm.com 
[mailto:owner-nmus...@globomaxnm.com] *On Behalf Of *Nick Holford

*Sent:* den 24 mars 2011 07:54
*To:* nmusers
*Subject:* Re: [NMusers] Time-to-event analysis with $DES

Hyewon,

Sorry -- I just realized that EFF is not the drug effect but 1-drug 
effect so that it has a value of 1 when AUC is zero. Please ignore my 
remarks about EFF being zero when AUC is zero in my comments below and 
consider this code to describe the effect of AUC on hazard:


 HAZNOW=LAMD*ALPH*(TIME+DEL2)**(ALPH-1)*(1-BETAE*AUC/(EC+AUC))

I think it is a very strong assumption that the drug will reduce the 
hazard to zero at infinite AUC which you can test by estimating a 
parameter (such as BETAE). If BETAE is significantly less than 1 then 
this means the assumption is not supported.


Nick

On 24/03/2011 8:17 a.m., Nick Holford wrote:

Hyewon,

The most obvious problem with your code is in $DES. You must use the 
variable T not the variable TIME when referring to a time varying 
hazard. TIME is the time on each data record. It does not change in 
$DES. T is the time from the last data record upto the current data 
record and changes within $DES.


You are predicting the likelihood of an event at the exact time of the 
event observation record with multiple events per subject. As you seem 
to realize you need to compute the cumulative hazard (CUMHAZ) either 
from TIME=0 or from the TIME of the last non-censored event for each 
subject.


In my opinion the following code is clearer and less dependent on your 
data structure than the method you are using which works only if your 
data has just event records after the TIME=0 record.


IF (MDV.EQ.0.AND.CS.EQ.0) THEN
   OLDCHZ=A(1) ; cum haz upto time of this event
ELSE
  OLDCHZ=OLDCHZ ; need to do this if OLDCHZ is a random variable
ENDIF

Your hazard model looks rather complicated. It seems to be based on 
the product of a Weibull  baseline hazard

LAMD*ALPH*(TIME+DEL2)**(ALPH-1)
then something odd involving the Weibull parameters
*EXP(-LAMD*(TIME+DEL2)**ALPH)
and then forces the hazard to be zero if AUC is zero.
*EFF
Is this really what you want? Do you know the hazard of event is zero 
if AUC is zero? Or is the last right parenthesis in the wrong place?


I would suggest something like this where LAMD and ALPH are the two 
parameters of the Weibull baseline hazard (when AUC is zero) and BETAE 
is a parameter describing the effect of the drug on the overall hazard.


 HAZNOW=LAMD*ALPH*(TIME+DEL2)**(ALPH-1)*EXP(BETAE*EFF)

You may also find it easier to develop the model if you do not try to 
estimate a random effect on LAMD until you have got reasonable 
estimates for the other parameters.


You may find it helpful to look at this presentation show

[NMusers] Autoinduction model - An increased clearance(day 1- 14)

[Shankar Lanke]

Dear All,

I am working on a Pop PK data where the patients are treated with HIV drug.
An autoinduction is involved with prolonged administration of the drug. An
increased CL is expected from day 1 to day 14.
We have intense data on day 1 and day 14 with sparse data between. Since a
lag period is involved for the induction I used the equation *CL* = *
CLinduced* -(*CLinduced - CLpre)*exp*(-*kout*(t*-*Tlag*)) described by Johan
Gabrielsson as more appropriate.

Also when I included a lag period for absorption in my earlier model my fits
are better and OBF decreased by 200.

However the final model with or without lag time for absorption + auto
induction model is either terminated or covariance step is being aborted.
I changed the initial estimates several times but still no luck. Though the
Auto induction model aborts the fits are better than the lag time model
however the estimates for Vd are 4 fold less than the expected.

I appreciate your input and suggestions. Here is my code.

$SUBROUTINES ADVAN13 TRANS1 TOL=5   ;(I used ADVAN6 too)
$MODEL
   NPAR=9 NCOMP=4
   COMP=(DEPOT,DEFDOSE)
   COMP=(LAG)
   COMP=(OBSV,DEFOBS)
   COMP=(PERIP)
$PK
   CLP=THETA(1)
   CLI=THETA(6)
   KOUT=THETA(7)
   TLAG=THETA(8)*EXP(ETA(6))

   TVCL=CLI-(CLI-CLP)*EXP(-KOUT*(TIME-TLAG))
   CL=TVCL*EXP(ETA(1))
   TVV2=THETA(2)
   V2=TVV2*EXP(ETA(2))
   TVQ=THETA(3)
   Q=TVQ*EXP(ETA(3))
   TVV3=THETA(4)
   V3=TVV3*EXP(ETA(4))
   TVKA=THETA(5)
   KA=TVKA*EXP(ETA(5))
   TVALAG1=THETA(9)
   ALAG1=TVALAG1*EXP(ETA(7))
   S3=V2
$DES
   K=CL/V2
   K23=Q/V2
   K32=Q/V3
   DADT(1)=-KA*A(1)
   DADT(2)=KA*A(1)-A(2)/ALAG1
   DADT(3)=A(2)/ALAG1-K23*A(3)-K*A(3)+K32*A(4)
   DADT(4)=K23*A(3)-K32*A(4)
$ERROR
 DEL=0
 IF (F.LE.0.0001) DEL=1
 IPRE=F
 W1= 1
 W2= F
 IRES= DV-IPRE
 IWRE=IRES/(W1+W2)
   Y = F + W1*ERR(1) + W2*ERR(2)
   DV2=ABS(V2-TVV2)
$EST METHOD=1 INTERACTION PRINT=5 MAX= SIG=3 MSFO=JLM.MSF
$THETA
  (0, 6);[CLP]
  (0, 90);[V2]
  (0, 19);[Q]
  (0, 200);[V3]
  (0, 0.16);[KA]
  (0, 8);[CLI]
  (0, 0.001);[KOUT]
  (0, 250);[TLAG]
  (0, 0.3);[ALAG1]
$OMEGA
  0.23 ;[CL] omega(1,1)
  0.18;[V2] omega(2,2)
  0 FIXED ;[Q] omega(3,3)
  0.42;[V3] omega(4,4)
  0.19;[KA] omega(5,5)
  0.09;[TLAG for Ka]
  0.1;[ALAG1 for CLI]
$SIGMA
  0.06 ;[P] sigma(1,1)
  0.09 ;[A] sigma(2,2)
$COV MATRIX=S


Regards,
Shankar Lanke Ph.D.
University at Buffalo
Office # 716-645-4853
Fax # 716-645-2886
Cell # 678-232-3567

RE: [NMusers] Autoinduction model - An increased clearance(day 1- 14)

[R.ter.Heine]

Dear Shankar,
 
How rich is your dataset? In other words: do you have enough data
troughout the induction period to estimate the lagtime? You could, for
example try to fix the lagtime to a reasonable time and estimate the
inter-individual variability. Another way of estimating the
autoniduction is more physiologically based with a theoretical enzyme
compartment. For example, see:
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2014348/figure/fig01/
 
Which drug PK are you modelling? Most likely it is a non-nucleoside
reverse transcriptase inhibitor. The cyp3a4 autoinduction with efavirenz
is debatable and less profound than autoinduction with, for example,
nevirapine. 
 
Sincerely,
 
Rob ter Heine



Van: owner-nmus...@globomaxnm.com [mailto:owner-nmus...@globomaxnm.com]
Namens Shankar Lanke
Verzonden: maandag 28 maart 2011 15:53
Aan: nmusers@globomaxnm.com
Onderwerp: [NMusers] Autoinduction model - An increased clearance(day 1-
14)


Dear All, 

I am working on a Pop PK data where the patients are treated with HIV
drug. An autoinduction is involved with prolonged administration of the
drug. An increased CL is expected from day 1 to day 14.
We have intense data on day 1 and day 14 with sparse data between. Since
a lag period is involved for the induction I used the equation CL =
CLinduced -(CLinduced - CLpre)*exp(-kout*(t-Tlag)) described by Johan
Gabrielsson as more appropriate. 


Also when I included a lag period for absorption in my earlier model my
fits are better and OBF decreased by 200. 


However the final model with or without lag time for absorption + auto
induction model is either terminated or covariance step is being
aborted.
I changed the initial estimates several times but still no luck. Though
the Auto induction model aborts the fits are better than the lag time
model however the estimates for Vd are 4 fold less than the expected.


I appreciate your input and suggestions. Here is my code.


$SUBROUTINES ADVAN13 TRANS1 TOL=5   ;(I used ADVAN6 too)
$MODEL
   NPAR=9 NCOMP=4
   COMP=(DEPOT,DEFDOSE)
   COMP=(LAG)
   COMP=(OBSV,DEFOBS)
   COMP=(PERIP)
$PK
   CLP=THETA(1)
   CLI=THETA(6)
   KOUT=THETA(7)
   TLAG=THETA(8)*EXP(ETA(6))
  
   TVCL=CLI-(CLI-CLP)*EXP(-KOUT*(TIME-TLAG))
   CL=TVCL*EXP(ETA(1))
   TVV2=THETA(2)
   V2=TVV2*EXP(ETA(2))
   TVQ=THETA(3)
   Q=TVQ*EXP(ETA(3))
   TVV3=THETA(4)
   V3=TVV3*EXP(ETA(4))
   TVKA=THETA(5)
   KA=TVKA*EXP(ETA(5))
   TVALAG1=THETA(9)
   ALAG1=TVALAG1*EXP(ETA(7))
   S3=V2
$DES
   K=CL/V2
   K23=Q/V2
   K32=Q/V3
   DADT(1)=-KA*A(1) 
   DADT(2)=KA*A(1)-A(2)/ALAG1  
   DADT(3)=A(2)/ALAG1-K23*A(3)-K*A(3)+K32*A(4) 
   DADT(4)=K23*A(3)-K32*A(4)  
$ERROR
 DEL=0
 IF (F.LE.0.0001) DEL=1
 IPRE=F
 W1= 1
 W2= F
 IRES= DV-IPRE
 IWRE=IRES/(W1+W2)
   Y = F + W1*ERR(1) + W2*ERR(2)
   DV2=ABS(V2-TVV2)
$EST METHOD=1 INTERACTION PRINT=5 MAX= SIG=3 MSFO=JLM.MSF 
$THETA 
  (0, 6);[CLP]
  (0, 90);[V2]
  (0, 19);[Q]
  (0, 200);[V3]
  (0, 0.16);[KA]
  (0, 8);[CLI]
  (0, 0.001);[KOUT]
  (0, 250);[TLAG]
  (0, 0.3);[ALAG1]
$OMEGA
  0.23 ;[CL] omega(1,1)
  0.18;[V2] omega(2,2)
  0 FIXED ;[Q] omega(3,3)
  0.42;[V3] omega(4,4)
  0.19;[KA] omega(5,5)
  0.09;[TLAG for Ka]
  0.1;[ALAG1 for CLI]
$SIGMA
  0.06 ;[P] sigma(1,1)
  0.09 ;[A] sigma(2,2)
$COV MATRIX=S


Regards,
Shankar Lanke Ph.D.  
University at Buffalo
Office # 716-645-4853
Fax # 716-645-2886 
Cell # 678-232-3567 


***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.

Re: [NMusers] Autoinduction model - An increased clearance(day 1- 14)

[Luann Phillips]

Shankar,

A couple of thoughts that may help.

(1) If the model is terminating with a note that states something along 
the lines of 'Infinite value of objective function', this indicates that 
the estimated value of TLAG for an individual is occurring at the time 
of a sample.


Potential solutions:
(a) Remove the eta from alag. With only sparse data between Day 1 and 
Day 14, the ability to discern individual differences in this parameter 
is probably limited.
(b) If you do have enough data to see differences in individuals, ignore 
samples that occur very close to the estimated TLAG for an individual.

(c) Try a different distribution for IIV of this parameter

(2) Re-parameterize CLinduced vs. CLpre. There is nothing in the model 
or control stream that prevents CLpre > CLinduced at the typical value 
level or the individual level. This could occur during the search 
causing termination.


Re-parmeterization:
CLinduced = CLpre + ThetaN

TVCL=CLinduced - (CLinduced - CLpre)*exp(-kout(t-tlag) becomes
TVCL=CLpre + thetaN(1-exp(-kout(t-tlag))

For typical value of thetaN use a lower bound of 0

(3) The current model also assumes that all subjects have the amount of 
change in CLinduced vs. CLpre. So you may want to try the following 
which also allows the change in CL to be continuous between TIME steps.

This will allow NONMEM to integrate more smoothly around time=TLAG.

$PK
CLP = THETA(N)*EXP(ETA(A))
CLI = THETA(N+1)*EXP(ETA(B))  ;represents the increase from CLP

$DES
;T in the $DES block represents continuous time
;TIME represents the discrete time values in the input dataset
CL = CLP + CLI*(1-EXP(-KOUT*(T-TLAG))

Good Luck,

Luann Phillips
Director, PK/PD
Cognigen Corporation
(716) 633-3463 ext. 236


Shankar Lanke wrote:

Dear All,

I am working on a Pop PK data where the patients are treated with HIV 
drug. An autoinduction is involved with prolonged administration of the 
drug. An increased CL is expected from day 1 to day 14.
We have intense data on day 1 and day 14 with sparse data between. Since 
a lag period is involved for the induction I used the 
equation /CL/ = /CLinduced/ -(/CLinduced - 
CLpre)*exp/(-/kout*(t/-/Tlag/)) described by Johan Gabrielsson as more 
appropriate. 

Also when I included a lag period for absorption in my earlier model my 
fits are better and OBF decreased by 200. 

However the final model with or without lag time for absorption + auto 
induction model is either terminated or covariance step is being aborted.
I changed the initial estimates several times but still no luck. Though 
the Auto induction model aborts the fits are better than the lag time 
model however the estimates for Vd are 4 fold less than the expected.


I appreciate your input and suggestions. Here is my code.

$SUBROUTINES ADVAN13 TRANS1 TOL=5   ;(I used ADVAN6 too)
$MODEL
   NPAR=9 NCOMP=4
   COMP=(DEPOT,DEFDOSE)
   COMP=(LAG)
   COMP=(OBSV,DEFOBS)
   COMP=(PERIP)
$PK
   CLP=THETA(1)
   CLI=THETA(6)
   KOUT=THETA(7)
   TLAG=THETA(8)*EXP(ETA(6))
  
   TVCL=CLI-(CLI-CLP)*EXP(-KOUT*(TIME-TLAG))

   CL=TVCL*EXP(ETA(1))
   TVV2=THETA(2)
   V2=TVV2*EXP(ETA(2))
   TVQ=THETA(3)
   Q=TVQ*EXP(ETA(3))
   TVV3=THETA(4)
   V3=TVV3*EXP(ETA(4))
   TVKA=THETA(5)
   KA=TVKA*EXP(ETA(5))
   TVALAG1=THETA(9)
   ALAG1=TVALAG1*EXP(ETA(7))
   S3=V2
$DES
   K=CL/V2
   K23=Q/V2
   K32=Q/V3
   DADT(1)=-KA*A(1) 
   DADT(2)=KA*A(1)-A(2)/ALAG1  
   DADT(3)=A(2)/ALAG1-K23*A(3)-K*A(3)+K32*A(4) 
   DADT(4)=K23*A(3)-K32*A(4)  
$ERROR

 DEL=0
 IF (F.LE.0.0001) DEL=1
 IPRE=F
 W1= 1
 W2= F
 IRES= DV-IPRE
 IWRE=IRES/(W1+W2)
   Y = F + W1*ERR(1) + W2*ERR(2)
   DV2=ABS(V2-TVV2)
$EST METHOD=1 INTERACTION PRINT=5 MAX= SIG=3 MSFO=JLM.MSF 
$THETA 
  (0, 6);[CLP]

  (0, 90);[V2]
  (0, 19);[Q]
  (0, 200);[V3]
  (0, 0.16);[KA]
  (0, 8);[CLI]
  (0, 0.001);[KOUT]
  (0, 250);[TLAG]
  (0, 0.3);[ALAG1]
$OMEGA
  0.23 ;[CL] omega(1,1)
  0.18;[V2] omega(2,2)
  0 FIXED ;[Q] omega(3,3)
  0.42;[V3] omega(4,4)
  0.19;[KA] omega(5,5)
  0.09;[TLAG for Ka]
  0.1;[ALAG1 for CLI]
$SIGMA
  0.06 ;[P] sigma(1,1)
  0.09 ;[A] sigma(2,2)
$COV MATRIX=S


Regards,
Shankar Lanke Ph.D. 
University at Buffalo

Office # 716-645-4853
Fax # 716-645-2886
Cell # 678-232-3567 

Re: [NMusers] Autoinduction model - An increased clearance(day 1- 14)

[Shankar Lanke]

Dear Rob ter Heine,

I am working with Efavirenz, I working with 66 patients, 924 data points,
intense on day 1 and 14  and a trough con in between the two weeks.

I looked into the Physiological model presented by Dr. Karlsson earlier but
I did not used it since I dont have any information about ENZYME comp or
precursor.

I used the reasonable estimates based on earlier literature and aslo I tried
NPD approach.

Thank you very much Rob ter Heine, I appreciate your input.


On Mon, Mar 28, 2011 at 10:36 AM,  wrote:

>  Dear Shankar,
>
> How rich is your dataset? In other words: do you have enough data troughout
> the induction period to estimate the lagtime? You could, for example try to
> fix the lagtime to a reasonable time and estimate the inter-individual
> variability. Another way of estimating the autoniduction is more
> physiologically based with a theoretical enzyme compartment. For example,
> see: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2014348/figure/fig01/
>
> Which drug PK are you modelling? Most likely it is a non-nucleoside reverse
> transcriptase inhibitor. The cyp3a4 autoinduction with efavirenz is
> debatable and less profound than autoinduction with, for example,
> nevirapine.
>
> Sincerely,
>
> Rob ter Heine
>
>  --
> *Van:* owner-nmus...@globomaxnm.com [mailto:owner-nmus...@globomaxnm.com]
> *Namens *Shankar Lanke
> *Verzonden:* maandag 28 maart 2011 15:53
> *Aan:* nmusers@globomaxnm.com
> *Onderwerp:* [NMusers] Autoinduction model - An increased clearance(day 1-
> 14)
>
> Dear All,
>
> I am working on a Pop PK data where the patients are treated with HIV drug.
> An autoinduction is involved with prolonged administration of the drug. An
> increased CL is expected from day 1 to day 14.
> We have intense data on day 1 and day 14 with sparse data between. Since a
> lag period is involved for the induction I used the equation *CL* = *
> CLinduced* -(*CLinduced - CLpre)*exp*(-*kout*(t*-*Tlag*)) described by
> Johan Gabrielsson as more appropriate.
>
> Also when I included a lag period for absorption in my earlier model my
> fits are better and OBF decreased by 200.
>
> However the final model with or without lag time for absorption + auto
> induction model is either terminated or covariance step is being aborted.
> I changed the initial estimates several times but still no luck. Though the
> Auto induction model aborts the fits are better than the lag time model
> however the estimates for Vd are 4 fold less than the expected.
>
> I appreciate your input and suggestions. Here is my code.
>
>  $SUBROUTINES ADVAN13 TRANS1 TOL=5   ;(I used ADVAN6 too)
> $MODEL
>NPAR=9 NCOMP=4
>COMP=(DEPOT,DEFDOSE)
>COMP=(LAG)
>COMP=(OBSV,DEFOBS)
>COMP=(PERIP)
> $PK
>CLP=THETA(1)
>CLI=THETA(6)
>KOUT=THETA(7)
>TLAG=THETA(8)*EXP(ETA(6))
>
>TVCL=CLI-(CLI-CLP)*EXP(-KOUT*(TIME-TLAG))
>CL=TVCL*EXP(ETA(1))
>TVV2=THETA(2)
>V2=TVV2*EXP(ETA(2))
>TVQ=THETA(3)
>Q=TVQ*EXP(ETA(3))
>TVV3=THETA(4)
>V3=TVV3*EXP(ETA(4))
>TVKA=THETA(5)
>KA=TVKA*EXP(ETA(5))
>TVALAG1=THETA(9)
>ALAG1=TVALAG1*EXP(ETA(7))
>S3=V2
> $DES
>K=CL/V2
>K23=Q/V2
>K32=Q/V3
>DADT(1)=-KA*A(1)
>DADT(2)=KA*A(1)-A(2)/ALAG1
>DADT(3)=A(2)/ALAG1-K23*A(3)-K*A(3)+K32*A(4)
>DADT(4)=K23*A(3)-K32*A(4)
> $ERROR
>  DEL=0
>  IF (F.LE.0.0001) DEL=1
>  IPRE=F
>  W1= 1
>  W2= F
>  IRES= DV-IPRE
>  IWRE=IRES/(W1+W2)
>Y = F + W1*ERR(1) + W2*ERR(2)
>DV2=ABS(V2-TVV2)
> $EST METHOD=1 INTERACTION PRINT=5 MAX= SIG=3 MSFO=JLM.MSF
> $THETA
>   (0, 6);[CLP]
>   (0, 90);[V2]
>   (0, 19);[Q]
>   (0, 200);[V3]
>   (0, 0.16);[KA]
>   (0, 8);[CLI]
>   (0, 0.001);[KOUT]
>   (0, 250);[TLAG]
>   (0, 0.3);[ALAG1]
> $OMEGA
>   0.23 ;[CL] omega(1,1)
>   0.18;[V2] omega(2,2)
>   0 FIXED ;[Q] omega(3,3)
>   0.42;[V3] omega(4,4)
>   0.19;[KA] omega(5,5)
>   0.09;[TLAG for Ka]
>   0.1;[ALAG1 for CLI]
> $SIGMA
>   0.06 ;[P] sigma(1,1)
>   0.09 ;[A] sigma(2,2)
> $COV MATRIX=S
>
>
> Regards,
> Shankar Lanke Ph.D.
> University at Buffalo
> Office # 716-645-4853
> Fax # 716-645-2886
> Cell # 678-232-3567
>
>  --
> 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.
> --
>



-- 
Regards,
Shankar Lanke Ph.D.
University at Buffalo
Office # 716-645-4853
Fax # 716-645-2886
Cell # 678-232-3567

RE: [NMusers] Autoinduction model - An increased clearance(day 1- 14)

[Ulrika Simonsson]

Dear Shankar,

 

I would guess that it is the term (t-tlag) which is negative in your code as
long as tmailto:owner-nmus...@globomaxnm.com] On
Behalf Of Shankar Lanke
Sent: den 28 mars 2011 15:53
To: nmusers@globomaxnm.com
Subject: [NMusers] Autoinduction model - An increased clearance(day 1- 14)

 

Dear All,

 

I am working on a Pop PK data where the patients are treated with HIV drug.
An autoinduction is involved with prolonged administration of the drug. An
increased CL is expected from day 1 to day 14.

We have intense data on day 1 and day 14 with sparse data between. Since a
lag period is involved for the induction I used the equation CL = CLinduced
-(CLinduced - CLpre)*exp(-kout*(t-Tlag)) described by Johan Gabrielsson as
more appropriate. 

 

Also when I included a lag period for absorption in my earlier model my fits
are better and OBF decreased by 200. 

 

However the final model with or without lag time for absorption + auto
induction model is either terminated or covariance step is being aborted.

I changed the initial estimates several times but still no luck. Though the
Auto induction model aborts the fits are better than the lag time model
however the estimates for Vd are 4 fold less than the expected.

 

I appreciate your input and suggestions. Here is my code.

 

$SUBROUTINES ADVAN13 TRANS1 TOL=5   ;(I used ADVAN6 too)

$MODEL

   NPAR=9 NCOMP=4

   COMP=(DEPOT,DEFDOSE)

   COMP=(LAG)

   COMP=(OBSV,DEFOBS)

   COMP=(PERIP)

$PK

   CLP=THETA(1)

   CLI=THETA(6)

   KOUT=THETA(7)

   TLAG=THETA(8)*EXP(ETA(6))

  

   TVCL=CLI-(CLI-CLP)*EXP(-KOUT*(TIME-TLAG))

   CL=TVCL*EXP(ETA(1))

   TVV2=THETA(2)

   V2=TVV2*EXP(ETA(2))

   TVQ=THETA(3)

   Q=TVQ*EXP(ETA(3))

   TVV3=THETA(4)

   V3=TVV3*EXP(ETA(4))

   TVKA=THETA(5)

   KA=TVKA*EXP(ETA(5))

   TVALAG1=THETA(9)

   ALAG1=TVALAG1*EXP(ETA(7))

   S3=V2

$DES

   K=CL/V2

   K23=Q/V2

   K32=Q/V3

   DADT(1)=-KA*A(1) 

   DADT(2)=KA*A(1)-A(2)/ALAG1  

   DADT(3)=A(2)/ALAG1-K23*A(3)-K*A(3)+K32*A(4) 

   DADT(4)=K23*A(3)-K32*A(4)  

$ERROR

 DEL=0

 IF (F.LE.0.0001) DEL=1

 IPRE=F

 W1= 1

 W2= F

 IRES= DV-IPRE

 IWRE=IRES/(W1+W2)

   Y = F + W1*ERR(1) + W2*ERR(2)

   DV2=ABS(V2-TVV2)

$EST METHOD=1 INTERACTION PRINT=5 MAX= SIG=3 MSFO=JLM.MSF 

$THETA 

  (0, 6);[CLP]

  (0, 90);[V2]

  (0, 19);[Q]

  (0, 200);[V3]

  (0, 0.16);[KA]

  (0, 8);[CLI]

  (0, 0.001);[KOUT]

  (0, 250);[TLAG]

  (0, 0.3);[ALAG1]

$OMEGA

  0.23 ;[CL] omega(1,1)

  0.18;[V2] omega(2,2)

  0 FIXED ;[Q] omega(3,3)

  0.42;[V3] omega(4,4)

  0.19;[KA] omega(5,5)

  0.09;[TLAG for Ka]

  0.1;[ALAG1 for CLI]

$SIGMA

  0.06 ;[P] sigma(1,1)

  0.09 ;[A] sigma(2,2)

$COV MATRIX=S



Regards,
Shankar Lanke Ph.D. 

University at Buffalo
Office # 716-645-4853
Fax # 716-645-2886

Cell # 678-232-3567 

 

Re: [NMusers] Autoinduction model - An increased clearance(day 1- 14)

[Shankar Lanke]

Dear Philips,

Thank you very much for your input, I am re-parameterizing the model now. I
will run this model and update you soon.

Regards,


On Mon, Mar 28, 2011 at 10:48 AM, Luann Phillips <
luann.phill...@cognigencorp.com> wrote:

> Shankar,
>
> A couple of thoughts that may help.
>
> (1) If the model is terminating with a note that states something along the
> lines of 'Infinite value of objective function', this indicates that the
> estimated value of TLAG for an individual is occurring at the time of a
> sample.
>
> Potential solutions:
> (a) Remove the eta from alag. With only sparse data between Day 1 and Day
> 14, the ability to discern individual differences in this parameter is
> probably limited.
> (b) If you do have enough data to see differences in individuals, ignore
> samples that occur very close to the estimated TLAG for an individual.
> (c) Try a different distribution for IIV of this parameter
>
> (2) Re-parameterize CLinduced vs. CLpre. There is nothing in the model or
> control stream that prevents CLpre > CLinduced at the typical value level or
> the individual level. This could occur during the search causing
> termination.
>
> Re-parmeterization:
> CLinduced = CLpre + ThetaN
>
> TVCL=CLinduced - (CLinduced - CLpre)*exp(-kout(t-tlag) becomes
> TVCL=CLpre + thetaN(1-exp(-kout(t-tlag))
>
> For typical value of thetaN use a lower bound of 0
>
> (3) The current model also assumes that all subjects have the amount of
> change in CLinduced vs. CLpre. So you may want to try the following which
> also allows the change in CL to be continuous between TIME steps.
> This will allow NONMEM to integrate more smoothly around time=TLAG.
>
> $PK
> CLP = THETA(N)*EXP(ETA(A))
> CLI = THETA(N+1)*EXP(ETA(B))  ;represents the increase from CLP
>
> $DES
> ;T in the $DES block represents continuous time
> ;TIME represents the discrete time values in the input dataset
> CL = CLP + CLI*(1-EXP(-KOUT*(T-TLAG))
>
> Good Luck,
>
> Luann Phillips
> Director, PK/PD
> Cognigen Corporation
> (716) 633-3463 ext. 236
>
>
>
> Shankar Lanke wrote:
>
>> Dear All,
>>
>> I am working on a Pop PK data where the patients are treated with HIV
>> drug. An autoinduction is involved with prolonged administration of the
>> drug. An increased CL is expected from day 1 to day 14.
>> We have intense data on day 1 and day 14 with sparse data between. Since a
>> lag period is involved for the induction I used the equation /CL/ =
>> /CLinduced/ -(/CLinduced - CLpre)*exp/(-/kout*(t/-/Tlag/)) described by
>> Johan Gabrielsson as more appropriate.
>> Also when I included a lag period for absorption in my earlier model my
>> fits are better and OBF decreased by 200.
>> However the final model with or without lag time for absorption + auto
>> induction model is either terminated or covariance step is being aborted.
>> I changed the initial estimates several times but still no luck. Though
>> the Auto induction model aborts the fits are better than the lag time model
>> however the estimates for Vd are 4 fold less than the expected.
>>
>> I appreciate your input and suggestions. Here is my code.
>>
>> $SUBROUTINES ADVAN13 TRANS1 TOL=5   ;(I used ADVAN6 too)
>> $MODEL
>>   NPAR=9 NCOMP=4
>>   COMP=(DEPOT,DEFDOSE)
>>   COMP=(LAG)
>>   COMP=(OBSV,DEFOBS)
>>   COMP=(PERIP)
>> $PK
>>   CLP=THETA(1)
>>   CLI=THETA(6)
>>   KOUT=THETA(7)
>>   TLAG=THETA(8)*EXP(ETA(6))
>> TVCL=CLI-(CLI-CLP)*EXP(-KOUT*(TIME-TLAG))
>>   CL=TVCL*EXP(ETA(1))
>>   TVV2=THETA(2)
>>   V2=TVV2*EXP(ETA(2))
>>   TVQ=THETA(3)
>>   Q=TVQ*EXP(ETA(3))
>>   TVV3=THETA(4)
>>   V3=TVV3*EXP(ETA(4))
>>   TVKA=THETA(5)
>>   KA=TVKA*EXP(ETA(5))
>>   TVALAG1=THETA(9)
>>   ALAG1=TVALAG1*EXP(ETA(7))
>>   S3=V2
>> $DES
>>   K=CL/V2
>>   K23=Q/V2
>>   K32=Q/V3
>>   DADT(1)=-KA*A(1)DADT(2)=KA*A(1)-A(2)/ALAG1
>> DADT(3)=A(2)/ALAG1-K23*A(3)-K*A(3)+K32*A(4)DADT(4)=K23*A(3)-K32*A(4)
>>  $ERROR
>>  DEL=0
>>  IF (F.LE.0.0001) DEL=1
>>  IPRE=F
>>  W1= 1
>>  W2= F
>>  IRES= DV-IPRE
>>  IWRE=IRES/(W1+W2)
>>   Y = F + W1*ERR(1) + W2*ERR(2)
>>   DV2=ABS(V2-TVV2)
>> $EST METHOD=1 INTERACTION PRINT=5 MAX= SIG=3 MSFO=JLM.MSF $THETA
>> (0, 6);[CLP]
>>  (0, 90);[V2]
>>  (0, 19);[Q]
>>  (0, 200);[V3]
>>  (0, 0.16);[KA]
>>  (0, 8);[CLI]
>>  (0, 0.001);[KOUT]
>>  (0, 250);[TLAG]
>>  (0, 0.3);[ALAG1]
>> $OMEGA
>>  0.23 ;[CL] omega(1,1)
>>  0.18;[V2] omega(2,2)
>>  0 FIXED ;[Q] omega(3,3)
>>  0.42;[V3] omega(4,4)
>>  0.19;[KA] omega(5,5)
>>  0.09;[TLAG for Ka]
>>  0.1;[ALAG1 for CLI]
>> $SIGMA
>>  0.06 ;[P] sigma(1,1)
>>  0.09 ;[A] sigma(2,2)
>> $COV MATRIX=S
>>
>>
>> Regards,
>> Shankar Lanke Ph.D. University at Buffalo
>> Office # 716-645-4853
>> Fax # 716-645-2886
>> Cell # 678-232-3567
>>
>


-- 
Regards,
Shankar Lanke Ph.D.
University at Buffalo
Office # 716-645-4853
Fax # 716-645-2886
Cell # 678-232-3567

Re: [NMusers] Autoinduction model - An increased clearance(day 1- 14)

[Shankar Lanke]

Dear Alice,

I reduced the number of eta's too but I didnt removed on induction. My
S3=V2, Vd is in mg/L so there are no issues with unit transformations.  I
agree with your suggestion, I will removed eta on induction model and I will
fix the induction parameter.

Thank you very much for your time and suggestions.

Its very nice of you all to respond to NM Beneficiary emails despite of your
busy schedules.


Regards,

On Mon, Mar 28, 2011 at 11:17 AM, Nichols, Alice
wrote:

>  When volume is off I always first check that all my unit transforms are
> accurate ie conc ng/ml / dose in mg
>
>
>
> Also for not getting covariance,  you probably need to reduce the number of
> etas you are estimating, its most likely having trouble estimating 1 or more
> of these,
>
>
>
> In particular estimating eta on your induction parameter may not be
> feasible given the minimal data for induction time course,
>
>
>
> I also wonder if you may need to fix the induction parameter  given that
> only sparse sampling between days 1 and 14
>
>
>
> *Alice*
>
>
>
> *Alice I Nichols, PhD*
>
> *Senior Director*
>
> *Clinical Pharmacology, Primary Care*
>
> *(484-865-8741 / 610-265-5372)*
>
>  alice.nich...@pfizer.com
>   --
>
> *From:* owner-nmus...@globomaxnm.com [mailto:owner-nmus...@globomaxnm.com]
> *On Behalf Of *Shankar Lanke
> *Sent:* Monday, March 28, 2011 9:53 AM
> *To:* nmusers@globomaxnm.com
> *Subject:* [NMusers] Autoinduction model - An increased clearance(day 1-
> 14)
>
>
>
> Dear All,
>
>
>
> I am working on a Pop PK data where the patients are treated with HIV drug.
> An autoinduction is involved with prolonged administration of the drug. An
> increased CL is expected from day 1 to day 14.
>
> We have intense data on day 1 and day 14 with sparse data between. Since a
> lag period is involved for the induction I used the equation *CL* = *
> CLinduced* -(*CLinduced - CLpre)*exp*(-*kout*(t*-*Tlag*)) described by
> Johan Gabrielsson as more appropriate.
>
>
>
> Also when I included a lag period for absorption in my earlier model my
> fits are better and OBF decreased by 200.
>
>
>
> However the final model with or without lag time for absorption + auto
> induction model is either terminated or covariance step is being aborted.
>
> I changed the initial estimates several times but still no luck. Though the
> Auto induction model aborts the fits are better than the lag time model
> however the estimates for Vd are 4 fold less than the expected.
>
>
>
> I appreciate your input and suggestions. Here is my code.
>
>
>
> $SUBROUTINES ADVAN13 TRANS1 TOL=5   ;(I used ADVAN6 too)
>
> $MODEL
>
>NPAR=9 NCOMP=4
>
>COMP=(DEPOT,DEFDOSE)
>
>COMP=(LAG)
>
>COMP=(OBSV,DEFOBS)
>
>COMP=(PERIP)
>
> $PK
>
>CLP=THETA(1)
>
>CLI=THETA(6)
>
>KOUT=THETA(7)
>
>TLAG=THETA(8)*EXP(ETA(6))
>
>
>
>TVCL=CLI-(CLI-CLP)*EXP(-KOUT*(TIME-TLAG))
>
>CL=TVCL*EXP(ETA(1))
>
>TVV2=THETA(2)
>
>V2=TVV2*EXP(ETA(2))
>
>TVQ=THETA(3)
>
>Q=TVQ*EXP(ETA(3))
>
>TVV3=THETA(4)
>
>V3=TVV3*EXP(ETA(4))
>
>TVKA=THETA(5)
>
>KA=TVKA*EXP(ETA(5))
>
>TVALAG1=THETA(9)
>
>ALAG1=TVALAG1*EXP(ETA(7))
>
>S3=V2
>
> $DES
>
>K=CL/V2
>
>K23=Q/V2
>
>K32=Q/V3
>
>DADT(1)=-KA*A(1)
>
>DADT(2)=KA*A(1)-A(2)/ALAG1
>
>DADT(3)=A(2)/ALAG1-K23*A(3)-K*A(3)+K32*A(4)
>
>DADT(4)=K23*A(3)-K32*A(4)
>
> $ERROR
>
>  DEL=0
>
>  IF (F.LE.0.0001) DEL=1
>
>  IPRE=F
>
>  W1= 1
>
>  W2= F
>
>  IRES= DV-IPRE
>
>  IWRE=IRES/(W1+W2)
>
>Y = F + W1*ERR(1) + W2*ERR(2)
>
>DV2=ABS(V2-TVV2)
>
> $EST METHOD=1 INTERACTION PRINT=5 MAX= SIG=3 MSFO=JLM.MSF
>
> $THETA
>
>   (0, 6);[CLP]
>
>   (0, 90);[V2]
>
>   (0, 19);[Q]
>
>   (0, 200);[V3]
>
>   (0, 0.16);[KA]
>
>   (0, 8);[CLI]
>
>   (0, 0.001);[KOUT]
>
>   (0, 250);[TLAG]
>
>   (0, 0.3);[ALAG1]
>
> $OMEGA
>
>   0.23 ;[CL] omega(1,1)
>
>   0.18;[V2] omega(2,2)
>
>   0 FIXED ;[Q] omega(3,3)
>
>   0.42;[V3] omega(4,4)
>
>   0.19;[KA] omega(5,5)
>
>   0.09;[TLAG for Ka]
>
>   0.1;[ALAG1 for CLI]
>
> $SIGMA
>
>   0.06 ;[P] sigma(1,1)
>
>   0.09 ;[A] sigma(2,2)
>
> $COV MATRIX=S
>
>
>
> Regards,
> Shankar Lanke Ph.D.
>
> University at Buffalo
> Office # 716-645-4853
> Fax # 716-645-2886
>
> Cell # 678-232-3567
>
>
>



-- 
Regards,
Shankar Lanke Ph.D.
University at Buffalo
Office # 716-645-4853
Fax # 716-645-2886
Cell # 678-232-3567

Re: [NMusers] Autoinduction model - An increased clearance(day 1- 14)

[Toufigh Gordi]

Hi Shankar,

We published an autoinduction model a few years back (Gordi et al., Br J Clin 
Pharmacol. 2005;59(2):189-98). The original paper was based on saliva samples 
but the model worked well using plasma data (Asimus and Gordi, Br J Clin 
Pharmacol. 2007;63(6):758-62). We also used it to describe the PK of the 
compound in a PK/PD model successfully (Gordi et al., Br J Clin Pharmacol. 
2005;60(6):594-604). The basic principal is similar to most other models, i.e., 
an indirect response model describes the enzyme compartment and the effect of 
the drug on production rate of the enzymes. One major difference is that we 
introduced a liver compartment into the system. In all other models I have 
seen, plasma concentrations drive the induction effect. This means that as time 
passes by, and drug concentrations in plasma decrease due to induction, you 
have less and less induction. Having the liver compartment, we allow drug 
concentrations (practically amounts, see the control stream below) in the liver 
upon absorption to induce the enzyme. This means that subsequent doses of the 
drug will have the same inducing effect, which should be closer to the reality.

I am showing a part of the control stream below. In our paper we estimated 
interoccasional variability and I'll be more than happy to help you with 
implementing that, if you want to. A couple of notes to make the coding easier 
to follow: this is based on salivary data, where concentrations reflected the 
free fraction of those in plasma (fu=14%), hence the S3 adjustment. The model 
also includes a precursor compartment to capture the induction lag time. 
Depending on your data, you may want to test a model without it. Finally, the 
model includes a nonlinearity in the intrinsic clearance and extraction ratio 
(and thereby bioavailability) of the drug, which was known for this particular 
compound. You should obviously test a simpler model.  

Let me know if you have any questions.

Toufigh


$SUBROUTINE ADVAN6 TRANS1 TOL=5
$MODEL NCOMP=5

COMP=(GUT DEFDOSE)   
COMP=LIVER
COMP=(SALIVA DEFOBS)   
COMP=ENZPOOL   
COMP=PREC

$PK

   
TEN   = THETA(1)   
KEN   = 0.693/TEN   
SIND  = THETA(2);slope for enzyme induction   
CLINT = THETA(3)*EXP(ETA(1))   
BFL   = 0.63*WT ;l/h for 55kg   
VS= THETA(4)*EXP(ETA(2)) 

. 
   
ALAG1 = THETA(5) 
KA= THETA(6)  
FU= 0.14   
KM= THETA(7)
   
VL= 1   
TPEN  = THETA(8)   
KPEN  = 1/TPEN   
S3= VS/FU   
BASENZ= 1   
BASEPR= KEN/KPEN   
F4= BASENZ   
F5= BASEPR   

$DES
   
BOX = CLINT*A(4)*KM/(A(2)+KM)   
E   = FU*BOX/(BFL+FU*BOX)   ;extraction ratio   
BAV = 1-E   ; bioavailability  
CLH = BFL*E   
DADT(1) = -KA*A(1)   
KS0 = BFL/VS   
DADT(2) = KA*A(1)-BFL*BAV*A(2)/VL+KS0*A(3)-BFL*E*A(2)/VL   
DADT(3) = BFL*BAV*A(2)/VL-KS0*A(3)   
DADT(4) = KPEN*A(5)-KEN*A(4)
DADT(5) = KEN*(1+SIND*A(2))-KPEN*A(5) 

Toufigh

Toufigh Gordi, PhD
President, PK/PD and Clinical Pharmacology Services
Rosa & Co. LLC: www.rosaandco.com
E-mail: tgo...@rosaandco.com
Tel.: 408-480-7314
Fax: 408-370-9810

The information contained in this e-mail message, e-mail message sequence, 
and/or any enclosures is confidential, and it may be privileged and protected 
from unauthorized use and/or disclosure. If you are not the intended recipient, 
any use, dissemination, distribution, or copying is strictly prohibited. If you 
received this e-mail in error, PLEASE NOTIFY THE SENDER immediately and destroy 
this message, any enclosures, and any copies of the message and/or the 
enclosures. Thank you.




On Mar 28, 2011, at 6:52 AM, Shankar Lanke wrote:

> Dear All,
> 
> I am working on a Pop PK data where the patients are treated with HIV drug. 
> An autoinduction is involved with prolonged administration of the drug. An 
> increased CL is expected from day 1 to day 14.
> We have intense data on day 1 and day 14 with sparse data between. Since a 
> lag period is involved for the induction I used the equation CL = CLinduced 
> -(CLinduced - CLpre)*exp(-kout*(t-Tlag)) described by Johan Gabrielsson as 
> more appropriate. 
> 
> Also when I included a lag period for absorption in my earlier model my fits 
> are better and OBF decreased by 200. 
> 
> However the final model with or without lag time for absorption + auto 
> induction model is either terminated or covariance step is being aborted.
> I changed the initial estimates several times but still no luck. Though the 
> Auto induction model aborts the fits are better than the lag time model 
> however the estimates for Vd are 4 fold less than the expected.
> 
> I appreciate your input and suggestions. Here is my code.
> 
> $SUBROUTINES ADVAN13 TRANS1 TOL=5   ;(I used ADVAN6 too)
> $MODEL
>NPAR=9 NCOMP=4
>COMP=(DEPOT,DEFDOSE)
>COMP=(LAG)
> 

Re: [NMusers] Autoinduction model - An increased clearance(day 1- 14)

[Ekaterina Gibiansky]

Dear Shankar,

In addition to what's been already said, you have a couple of problems
in your code. ALAG1 is a reserved name in NONMEM. You use it as
1/Ktransit for your transit absorption compartment. But NONMEM also
uses it to delay a dose entering  compartment 1 by ALAG1 time.
The second problem is how you formulate dependence of CL on TIME. Your
CL changes at the observation and dosing times only, not continuously.
You essentially substitute you function with piece-wise constant
function that change value at each observation and dosing time. So, the
results will depend on how often and where the observation times are.
Non-continuous parameters are also a sourse of numerical problems. To
make CL a continuous function of time, you need to have the function
inside $DES and use T rather than TIME.  

Regards,
Katya
Ekaterina Gibiansky, Ph.D.
CEO&CSO, QuantPharm LLC
Web: www.quantpharm.com
Email:   egibian...@quantpharm.com
 

On 3/28/2011 11:15 AM, Ulrika Simonsson wrote:

  
  
  
  
  Dear
Shankar,
   
  I
would guess that it is the term (t-tlag) which is negative in your code
as long as t
   
  Although
it physiologically makes sense to have a lagtime for the induction, the
estimation of this parameter is dependent on the information in the
data you are using. Try to fit a model without the lag time to see if
it is significant. In your code you are also adding the random effect
only on CL which says that the between patient variability is the same
at pre-induction state as at induced state. It could be worth exploring
if the model supports separate IIV in pre-induced CL compared to
induced CL. The time to steady state of induction which is determined
by kout in your model could also be different between individuals.
   
  There
are also example of more semi-physiological models for autoinduction
that could be worth exploring 
  *Hassan
et al. Br J Clin Pharmacol. 1999 Nov;48(5):669-77. A mechanism-based
pharmacokinetic-enzyme model for cyclophosphamide autoinduction in
breast cancer patients.
   
  In
this enzyme turnover model you set the enzyme amount to 1 at baseline
and estimate the change from baseline. There is no need to have
information about the enzyme levels.
   
  You
are also using only one transit absorption compartment, but it could be
worth exploring a more complex transit model where you estimate the
number of transit compartments unless you already have explored this
(Savic et al J Pharmacokinet Pharmacodyn. 2007 Oct;34(5):711-26)
   
   
  Best
regards,
   
  Ulrika
   
   
   
  Ulrika
Simonsson, PhD
  Assoc
Prof of Pharmacometrics
   
  Uppsala
Pharmacometrics
  Department
of Pharmaceutical Biosciences
  Uppsala University
  BMC, Box 591, 751 24 Uppsala
  Sweden
   
  
  From: owner-nmus...@globomaxnm.com
  [mailto:owner-nmus...@globomaxnm.com]
  On Behalf Of Shankar Lanke
  Sent: den 28 mars 2011 15:53
  To: nmusers@globomaxnm.com
  Subject: [NMusers] Autoinduction model - An increased
clearance(day 1- 14)
  
   
  Dear All,
  
   
  
  
  I am working on a Pop PK data where the patients
are treated with HIV drug. An autoinduction is involved with prolonged
administration of the drug. An increased CL is expected from day 1 to
day 14.
  
  
  We have intense data on day 1 and day 14 with
sparse data between. Since a lag period is involved for the induction I
used the equation CL = CLinduced -(CLinduced
- CLpre)*exp(-kout*(t-Tlag)) described by Johan
Gabrielsson as more appropriate. 
  
  
   
  
  
  Also when I included a lag
period for absorption in my earlier model my fits are better and OBF
decreased by 200. 
  
  
   
  
  
  However the final model
with or without lag time for absorption + auto induction model is
either terminated or covariance step is being aborted.
  
  
  I changed the initial
estimates several times but still no luck. Though the Auto induction
model aborts the fits are better than the lag time model however the
estimates for Vd are 4 fold less than the expected.
  
  
   
  
  
  I appreciate your input and
suggestions. Here is my code.
  
  
   
  
  
  
  $SUBROUTINES ADVAN13 TRANS1
TOL=5   ;(I used ADVAN6 too)
  
  
  $MODEL
  
  
     NPAR=9 NCOMP=4
  
  
     COMP=(DEPOT,DEFDOSE)
  
  
     COMP=(LAG)
  
  
     COMP=(OBSV,DEFOBS)
  
  
     COMP=(PERIP)
  
  
  $PK
  
  
     CLP=THETA(1)
  
  
     CLI=THETA(6)
  
  
     KOUT=THETA(7)
  
  
    
TLAG=THETA(8)*EXP(ETA(6))
  
  
    
  
  
     TVCL=CLI-(CLI-CLP)*EXP(-KOUT*(TIME-TLAG))
  
  
     CL=TVCL*EXP(ETA(1))
  
  
     TVV2=THETA(2)
  
  
    
V2=TVV2*EXP(ETA(2))
  
  
     TVQ=THETA(3)
  
  
    
Q=TVQ*EXP(ETA(3))
  
  
     TVV3=THETA(4)
  
  
    
V3=TVV3*EXP(ETA(4))
  
  
     TVKA=THETA(5)
  
  
    
KA=TVKA*EXP(ETA(5))
  
  
    
TVALAG1=THETA(9)
  
  
    
ALAG1=TVALAG1*EXP(ETA(7))
  
  
     S3=V2
  
  
  $DES
  
  
     K=CL/V2
  
  
     K23=Q/V2
  
  
     K32=Q/V3
  
  
     DADT(1)=-KA*A(1) 
  
  
    
DADT(2)=KA*A(1)-A(2)/ALAG1  
  
  
    
DADT(3)=A(2)/ALAG1-K23*A(3)-K*A

Re: [NMusers] Autoinduction model - An increased clearance(day 1- 14)

[Shankar Lanke]

Dear Toufigh,

Thank you very much for your email, there is wealth of information you have
provided me, I will look carefully into your suggestions and I will my model
soon.


Once again I appreciate your input.

On Mon, Mar 28, 2011 at 12:51 PM, Toufigh Gordi wrote:

> Hi Shankar,
>
> We published an autoinduction model a few years back (Gordi et al., Br J
> Clin Pharmacol. 2005;59(2):189-98). The original paper was based on saliva
> samples but the model worked well using plasma data (Asimus and Gordi, Br J
> Clin Pharmacol. 2007;63(6):758-62). We also used it to describe the PK of
> the compound in a PK/PD model successfully (Gordi et al., Br J Clin
> Pharmacol. 2005;60(6):594-604). The basic principal is similar to most other
> models, i.e., an indirect response model describes the enzyme compartment
> and the effect of the drug on production rate of the enzymes. One major
> difference is that we introduced a liver compartment into the system. In all
> other models I have seen, plasma concentrations drive the induction effect.
> This means that as time passes by, and drug concentrations in plasma
> decrease due to induction, you have less and less induction. Having the
> liver compartment, we allow drug concentrations (practically amounts, see
> the control stream below) in the liver upon absorption to induce the enzyme.
> This means that subsequent doses of the drug will have the same inducing
> effect, which should be closer to the reality.
>
> I am showing a part of the control stream below. In our paper we estimated
> interoccasional variability and I'll be more than happy to help you with
> implementing that, if you want to. A couple of notes to make the coding
> easier to follow: this is based on salivary data, where concentrations
> reflected the free fraction of those in plasma (fu=14%), hence the S3
> adjustment. The model also includes a precursor compartment to capture the
> induction lag time. Depending on your data, you may want to test a model
> without it. Finally, the model includes a nonlinearity in the intrinsic
> clearance and extraction ratio (and thereby bioavailability) of the drug,
> which was known for this particular compound. You should obviously test a
> simpler model.
>
> Let me know if you have any questions.
>
> Toufigh
>
>
> $SUBROUTINE ADVAN6 TRANS1 TOL=5
> $MODEL NCOMP=5
>
> COMP=(GUT DEFDOSE)
> COMP=LIVER
> COMP=(SALIVA DEFOBS)
> COMP=ENZPOOL
> COMP=PREC
>
> $PK
> 
>
> TEN   = THETA(1)
> KEN   = 0.693/TEN
> SIND  = THETA(2);slope for enzyme induction
> CLINT = THETA(3)*EXP(ETA(1))
> BFL   = 0.63*WT ;l/h for 55kg
> VS= THETA(4)*EXP(ETA(2))
>
> .
>
> ALAG1 = THETA(5)
> KA= THETA(6)
> FU= 0.14
> KM= THETA(7)
>
> VL= 1
> TPEN  = THETA(8)
> KPEN  = 1/TPEN
> S3= VS/FU
> BASENZ= 1
> BASEPR= KEN/KPEN
> F4= BASENZ
> F5= BASEPR
>
> $DES
>
> BOX = CLINT*A(4)*KM/(A(2)+KM)
> E   = FU*BOX/(BFL+FU*BOX)   ;extraction ratio
> BAV = 1-E   ; bioavailability
> CLH = BFL*E
> DADT(1) = -KA*A(1)
> KS0 = BFL/VS
> DADT(2) = KA*A(1)-BFL*BAV*A(2)/VL+KS0*A(3)-BFL*E*A(2)/VL
> DADT(3) = BFL*BAV*A(2)/VL-KS0*A(3)
> DADT(4) = KPEN*A(5)-KEN*A(4)
> DADT(5) = KEN*(1+SIND*A(2))-KPEN*A(5)
>
> Toufigh
>
> Toufigh Gordi, PhD
> President, PK/PD and Clinical Pharmacology Services
> Rosa & Co. LLC: www.rosaandco.com 
> E-mail: tgo...@rosaandco.com 
> Tel.: 408-480-7314
> Fax: 408-370-9810
>
> *The information contained in this e-mail message, e-mail message
> sequence, and/or any enclosures is confidential, and it may be privileged
> and protected from unauthorized use and/or disclosure. If you are not the
> intended recipient, any use, dissemination, distribution, or copying is
> strictly prohibited. If you received this e-mail in error, PLEASE NOTIFY THE
> SENDER immediately and destroy this message, any enclosures, and any copies
> of the message and/or the enclosures. Thank you.*
>
>
>
>
> On Mar 28, 2011, at 6:52 AM, Shankar Lanke wrote:
>
> Dear All,
>
> I am working on a Pop PK data where the patients are treated with HIV drug.
> An autoinduction is involved with prolonged administration of the drug. An
> increased CL is expected from day 1 to day 14.
> We have intense data on day 1 and day 14 with sparse data between. Since a
> lag period is involved for the induction I used the equation *CL* = *
> CLinduced* -(*CLinduced - CLpre)*exp*(-*kout*(t*-*Tlag*)) described by
> Johan Gabrielsson as more appropriate.
>
> Also when I included a lag period for absorption in my earlier model my
> fits are better and OBF decreased by 200.
>
> However the final model with or without lag time for absorption + auto
> induction model is either terminated or covariance step is being aborted.
> I changed the initial estimates several times but still no luck. Though the
> Auto induction model aborts the fits are better than the lag time model
> however the estimates for Vd are 4 fold less than the expected.
>
> I

RE: [NMusers] Time-to-event analysis with $DES

[Matt Hutmacher]

Hi Nick, Andy,

 

The Weibull hazard is not necessarily my first choice either.  However, I
should point out that it is not a much of an issue that the Weibull predicts
a zero hazard at time 0 (only defined for ALP>1).  One typically assumes the
survival fraction is = 1 at time = 0, so it doesn't matter what the hazard
is for any model at time =0.  Also,  If Beta = 0 (in yours or ALP=1 in
Andy's) it is well known that the Weibull = Exponential model, which has a
constant hazard throughout.  Note that if 00])?

 

Best regards,
Matt

 

 

From: owner-nmus...@globomaxnm.com [mailto:owner-nmus...@globomaxnm.com] On
Behalf Of Nick Holford
Sent: Monday, March 28, 2011 9:35 AM
To: Andrew Hooker
Cc: 'nmusers'
Subject: Re: [NMusers] Time-to-event analysis with $DES

 

Andy,

Thanks for this response. In fact I did not propose that particular Weibull
parameterization but simply pointed out that it seemed to be what the
original questioner seemed to be trying to use.

My own preference for parameterizing the Weibull is

  HAZARD=LAMDA0*EXP(BETA*LOG(TIME))







This parameterization makes is easier to understand that the Weibull
(ln(time)) is just a transformation of the Gompertz (time):

 

  HAZARD=LAMDA0*EXP(BETA*TIME)







Its then easy to extend hazards to more interesting and possibly more useful
hazards such as the Gombull (aka the Weipertz):





  HAZARD=LAMDA0*EXP(BETAG*TIME + BETAW*LOG(TIME))



 

One of the features of the Weibull that I dont like is that the hazard is
predicted to be zero when time is zero. This seems a particularly unlikely
biological occurrence and also requires extra coding to deal with this
special case when time is zero to avoid NaNs.

I'd be interested in knowing if you find any difference in the standard
errors with the ln(time) parameterization. But whatever you find it won't
change my behaviour because as you know standard errors are not much use for
making predictions :-)




Best wishes,




Nick



 


On 28/03/2011 3:05 p.m., Andrew Hooker wrote: 

Hi All

 

I would just like to point out that in the Weibull hazard that Nick
proposes:

 

HAZNOW=LAMD*ALPH*(TIME+DEL2)**(ALPH-1)

 

The units of LAMD would have to be TIME**(-ALPH).  That is, as soon as ALPH
changes then the units of LAMD change.  This seems strange and unnecessary
to include this type of correlation in the parameters.  If we instead
reparameterize the model to:

 

  HAZNOW=LAMD*ALPH*(LAMD*(TIME+DEL2))**(ALPH-1)

 

Then LAMD has units of 1/TIME no matter the value of ALPH.  I have tested
the two parameterizations in one simple simulation example and found that
the relative standard errors were significantly lower with this second
parameterization.

 

Best regards

Andy

 

 

Andrew Hooker, Ph.D.

Associate Professor of Pharmacometrics

Dept. of Pharmaceutical Biosciences

Uppsala University

Box 591, 751 24, Uppsala, Sweden

Phone: +46 18 471 4355

www.farmbio.uu.se/research/Research+groups/Pharmacometrics/ 

 

From: owner-nmus...@globomaxnm.com [mailto:owner-nmus...@globomaxnm.com] On
Behalf Of Nick Holford
Sent: den 24 mars 2011 07:54
To: nmusers
Subject: Re: [NMusers] Time-to-event analysis with $DES

 

Hyewon,

Sorry -- I just realized that EFF is not the drug effect but 1-drug effect
so that it has a value of 1 when AUC is zero. Please ignore my remarks about
EFF being zero when AUC is zero in my comments below and consider this code
to describe the effect of AUC on hazard:

 HAZNOW=LAMD*ALPH*(TIME+DEL2)**(ALPH-1)*(1-BETAE*AUC/(EC+AUC)) 

I think it is a very strong assumption that the drug will reduce the hazard
to zero at infinite AUC which you can test by estimating a parameter (such
as BETAE). If BETAE is significantly less than 1 then this means the
assumption is not supported.

Nick

On 24/03/2011 8:17 a.m., Nick Holford wrote: 

Hyewon,

The most obvious problem with your code is in $DES. You must use the
variable T not the variable TIME when referring to a time varying hazard.
TIME is the time on each data record. It does not change in $DES. T is the
time from the last data record upto the current data record and changes
within $DES.

You are predicting the likelihood of an event at the exact time of the event
observation record with multiple events per subject. As you seem to realize
you need to compute the cumulative hazard (CUMHAZ) either from TIME=0 or
from the TIME of the last non-censored event for each subject.

In my opinion the following code is clearer and less dependent on your data
structure than the method you are using which works only if your data has
just event records after the TIME=0 record. 

IF (MDV.EQ.0.AND.CS.EQ.0) THEN
   OLDCHZ=A(1) ; cum haz upto time of this event
ELSE
  OLDCHZ=OLDCHZ ; need to do this if OLDCHZ is a random variable
ENDIF

Your hazard model looks rather complicated. It seems to be based on the
product of a Weibull  baseline hazard 
LAMD*ALPH*(TIME+DEL2)**(ALPH-1)
then something odd involving the Weibull parameters
*EXP(-LAMD*(TIME+DEL2)**ALP

RE: [NMusers] Autoinduction model - An increased clearance(day 1- 14)

[mats karlsson]

Dear Shankar,

 

You don't need to have information about enzyme levels or precursor to use
the model mentioned by Rob, the idea of it is only to have a more
physiological mechanism for the delay (not a lag but something that develops
with a first-order delay) and magnitude (which is dependent on the
concentration of drug rather than an on/off ). The number of parameters is
no more than the one you're using now, but avoids the change-point which
often cause numerical problems.

 

Best regards,

Mats

 

Mats Karlsson, PhD

Professor of Pharmacometrics

Dept of Pharmaceutical Biosciences

Uppsala University

Sweden

 

Postal address: Box 591, 751 24 Uppsala, Sweden

Phone +46 18 4714105

Fax + 46 18 4714003

 

From: owner-nmus...@globomaxnm.com [mailto:owner-nmus...@globomaxnm.com] On
Behalf Of Shankar Lanke
Sent: Monday, March 28, 2011 4:51 PM
To: r.ter.he...@meandermc.nl
Cc: nmusers@globomaxnm.com
Subject: Re: [NMusers] Autoinduction model - An increased clearance(day 1-
14)

 

Dear Rob ter Heine,

 

I am working with Efavirenz, I working with 66 patients, 924 data points,
intense on day 1 and 14  and a trough con in between the two weeks.

 

I looked into the Physiological model presented by Dr. Karlsson earlier but
I did not used it since I dont have any information about ENZYME comp or
precursor.

 

I used the reasonable estimates based on earlier literature and aslo I tried
NPD approach. 

 

Thank you very much Rob ter Heine, I appreciate your input.

 

 

On Mon, Mar 28, 2011 at 10:36 AM,  wrote:

Dear Shankar,

 

How rich is your dataset? In other words: do you have enough data troughout
the induction period to estimate the lagtime? You could, for example try to
fix the lagtime to a reasonable time and estimate the inter-individual
variability. Another way of estimating the autoniduction is more
physiologically based with a theoretical enzyme compartment. For example,
see: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2014348/figure/fig01/

 

Which drug PK are you modelling? Most likely it is a non-nucleoside reverse
transcriptase inhibitor. The cyp3a4 autoinduction with efavirenz is
debatable and less profound than autoinduction with, for example,
nevirapine. 

 

Sincerely,

 

Rob ter Heine

 

  _  

Van: owner-nmus...@globomaxnm.com [mailto:owner-nmus...@globomaxnm.com]
Namens Shankar Lanke
Verzonden: maandag 28 maart 2011 15:53
Aan: nmusers@globomaxnm.com
Onderwerp: [NMusers] Autoinduction model - An increased clearance(day 1- 14)

Dear All, 

 

I am working on a Pop PK data where the patients are treated with HIV drug.
An autoinduction is involved with prolonged administration of the drug. An
increased CL is expected from day 1 to day 14.

We have intense data on day 1 and day 14 with sparse data between. Since a
lag period is involved for the induction I used the equation CL = CLinduced
-(CLinduced - CLpre)*exp(-kout*(t-Tlag)) described by Johan Gabrielsson as
more appropriate. 

 

Also when I included a lag period for absorption in my earlier model my fits
are better and OBF decreased by 200. 

 

However the final model with or without lag time for absorption + auto
induction model is either terminated or covariance step is being aborted.

I changed the initial estimates several times but still no luck. Though the
Auto induction model aborts the fits are better than the lag time model
however the estimates for Vd are 4 fold less than the expected.

 

I appreciate your input and suggestions. Here is my code.

 

$SUBROUTINES ADVAN13 TRANS1 TOL=5   ;(I used ADVAN6 too)

$MODEL

   NPAR=9 NCOMP=4

   COMP=(DEPOT,DEFDOSE)

   COMP=(LAG)

   COMP=(OBSV,DEFOBS)

   COMP=(PERIP)

$PK

   CLP=THETA(1)

   CLI=THETA(6)

   KOUT=THETA(7)

   TLAG=THETA(8)*EXP(ETA(6))

  

   TVCL=CLI-(CLI-CLP)*EXP(-KOUT*(TIME-TLAG))

   CL=TVCL*EXP(ETA(1))

   TVV2=THETA(2)

   V2=TVV2*EXP(ETA(2))

   TVQ=THETA(3)

   Q=TVQ*EXP(ETA(3))

   TVV3=THETA(4)

   V3=TVV3*EXP(ETA(4))

   TVKA=THETA(5)

   KA=TVKA*EXP(ETA(5))

   TVALAG1=THETA(9)

   ALAG1=TVALAG1*EXP(ETA(7))

   S3=V2

$DES

   K=CL/V2

   K23=Q/V2

   K32=Q/V3

   DADT(1)=-KA*A(1) 

   DADT(2)=KA*A(1)-A(2)/ALAG1  

   DADT(3)=A(2)/ALAG1-K23*A(3)-K*A(3)+K32*A(4) 

   DADT(4)=K23*A(3)-K32*A(4)  

$ERROR

 DEL=0

 IF (F.LE.0.0001) DEL=1

 IPRE=F

 W1= 1

 W2= F

 IRES= DV-IPRE

 IWRE=IRES/(W1+W2)

   Y = F + W1*ERR(1) + W2*ERR(2)

   DV2=ABS(V2-TVV2)

$EST METHOD=1 INTERACTION PRINT=5 MAX= SIG=3 MSFO=JLM.MSF 

$THETA 

  (0, 6);[CLP]

  (0, 90);[V2]

  (0, 19);[Q]

  (0, 200);[V3]

  (0, 0.16);[KA]

  (0, 8);[CLI]

  (0, 0.001);[KOUT]

  (0, 250);[TLAG]

  (0, 0.3);[ALAG1]

$OMEGA

  0.23 ;[CL] omega(1,1)

  0.18;[V2] omega(2,2)

  0 FIXED ;[Q] omega(3,3)

  0.42;[V3] omega(4,4)

  0.19;[KA] omega(5,5)

  0.09;[TLAG for Ka]

  0.1;[ALAG1 for CLI]

$SIGMA

  0.06 ;[P] sigma(1,1)

  0.09 ;[A] sigma(2,2)

$COV MATRIX=S



Regards,
Shankar Lanke Ph.D.  

University at Buffalo
Office # 716-645-4853
Fax # 716-645-2886 

Cell # 678-232-3567

Re: [NMusers] Autoinduction model - An increased clearance(day 1- 14)

[Shankar Lanke]

Dear Dr. Karlsson,

Thank you very much, I will apply the enzyme model for Auto Induction as
described Hassan et.al earlier.

I appreciate your feed back.

Once again Thank you very much Rob Ter Heine.


On Mon, Mar 28, 2011 at 6:03 PM, mats karlsson
wrote:

> Dear Shankar,
>
>
>
> You don’t need to have information about enzyme levels or precursor to use
> the model mentioned by Rob, the idea of it is only to have a more
> physiological mechanism for the delay (not a lag but something that develops
> with a first-order delay) and magnitude (which is dependent on the
> concentration of drug rather than an on/off ). The number of parameters is
> no more than the one you’re using now, but avoids the change-point which
> often cause numerical problems.
>
>
>
> Best regards,
>
> Mats
>
>
>
> Mats Karlsson, PhD
>
> Professor of Pharmacometrics
>
> Dept of Pharmaceutical Biosciences
>
> Uppsala University
>
> Sweden
>
>
>
> Postal address: Box 591, 751 24 Uppsala, Sweden
>
> Phone +46 18 4714105
>
> Fax + 46 18 4714003
>
>
>
> *From:* owner-nmus...@globomaxnm.com [mailto:owner-nmus...@globomaxnm.com]
> *On Behalf Of *Shankar Lanke
> *Sent:* Monday, March 28, 2011 4:51 PM
> *To:* r.ter.he...@meandermc.nl
> *Cc:* nmusers@globomaxnm.com
> *Subject:* Re: [NMusers] Autoinduction model - An increased clearance(day
> 1- 14)
>
>
>
> Dear Rob ter Heine,
>
>
>
> I am working with Efavirenz, I working with 66 patients, 924 data points,
> intense on day 1 and 14  and a trough con in between the two weeks.
>
>
>
> I looked into the Physiological model presented by Dr. Karlsson earlier but
> I did not used it since I dont have any information about ENZYME comp or
> precursor.
>
>
>
> I used the reasonable estimates based on earlier literature and aslo I
> tried NPD approach.
>
>
>
> Thank you very much Rob ter Heine, I appreciate your input.
>
>
>
>
>
> On Mon, Mar 28, 2011 at 10:36 AM,  wrote:
>
> Dear Shankar,
>
>
>
> How rich is your dataset? In other words: do you have enough data troughout
> the induction period to estimate the lagtime? You could, for example try to
> fix the lagtime to a reasonable time and estimate the inter-individual
> variability. Another way of estimating the autoniduction is more
> physiologically based with a theoretical enzyme compartment. For example,
> see: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2014348/figure/fig01/
>
>
>
> Which drug PK are you modelling? Most likely it is a non-nucleoside reverse
> transcriptase inhibitor. The cyp3a4 autoinduction with efavirenz is
> debatable and less profound than autoinduction with, for example,
> nevirapine.
>
>
>
> Sincerely,
>
>
>
> Rob ter Heine
>
>
> --
>
> *Van:* owner-nmus...@globomaxnm.com [mailto:owner-nmus...@globomaxnm.com]
> *Namens *Shankar Lanke
> *Verzonden:* maandag 28 maart 2011 15:53
> *Aan:* nmusers@globomaxnm.com
> *Onderwerp:* [NMusers] Autoinduction model - An increased clearance(day 1-
> 14)
>
> Dear All,
>
>
>
> I am working on a Pop PK data where the patients are treated with HIV drug.
> An autoinduction is involved with prolonged administration of the drug. An
> increased CL is expected from day 1 to day 14.
>
> We have intense data on day 1 and day 14 with sparse data between. Since a
> lag period is involved for the induction I used the equation *CL* = *
> CLinduced* -(*CLinduced - CLpre)*exp*(-*kout*(t*-*Tlag*)) described by
> Johan Gabrielsson as more appropriate.
>
>
>
> Also when I included a lag period for absorption in my earlier model my
> fits are better and OBF decreased by 200.
>
>
>
> However the final model with or without lag time for absorption + auto
> induction model is either terminated or covariance step is being aborted.
>
> I changed the initial estimates several times but still no luck. Though the
> Auto induction model aborts the fits are better than the lag time model
> however the estimates for Vd are 4 fold less than the expected.
>
>
>
> I appreciate your input and suggestions. Here is my code.
>
>
>
> $SUBROUTINES ADVAN13 TRANS1 TOL=5   ;(I used ADVAN6 too)
>
> $MODEL
>
>NPAR=9 NCOMP=4
>
>COMP=(DEPOT,DEFDOSE)
>
>COMP=(LAG)
>
>COMP=(OBSV,DEFOBS)
>
>COMP=(PERIP)
>
> $PK
>
>CLP=THETA(1)
>
>CLI=THETA(6)
>
>KOUT=THETA(7)
>
>TLAG=THETA(8)*EXP(ETA(6))
>
>
>
>TVCL=CLI-(CLI-CLP)*EXP(-KOUT*(TIME-TLAG))
>
>CL=TVCL*EXP(ETA(1))
>
>TVV2=THETA(2)
>
>V2=TVV2*EXP(ETA(2))
>
>TVQ=THETA(3)
>
>Q=TVQ*EXP(ETA(3))
>
>TVV3=THETA(4)
>
>V3=TVV3*EXP(ETA(4))
>
>TVKA=THETA(5)
>
>KA=TVKA*EXP(ETA(5))
>
>TVALAG1=THETA(9)
>
>ALAG1=TVALAG1*EXP(ETA(7))
>
>S3=V2
>
> $DES
>
>K=CL/V2
>
>K23=Q/V2
>
>K32=Q/V3
>
>DADT(1)=-KA*A(1)
>
>DADT(2)=KA*A(1)-A(2)/ALAG1
>
>DADT(3)=A(2)/ALAG1-K23*A(3)-K*A(3)+K32*A(4)
>
>DADT(4)=K23*A(3)-K32*A(4)
>
> $ERROR
>
>  DEL=0
>
>  IF (F.LE.0.0001) DEL=1
>
>  IPRE=F
>
>  W1= 1
>
>  W2= F
>
>  IRES= DV-IPRE
>
>  IWRE=IRES/(W1+W2)
>
>Y = F + W1*ERR(1) + W2*ERR(2)
>