NONMEM Users Network Archive

Hosted by Cognigen

RE: PD model

From: Parviz Ghahramani <parvizghahramani>
Date: Wed, 20 Jan 2010 16:13:26 +0000

Here are my thoughts on this topic and the models:


I agree with the comments made about the baseline and the method to deal wi=
th estimating the baseline pattern of QTc.


But, adding up concentrations of two different drugs should be the last r=
esort, because it assumes that the drugs are equipotent (with respect to =
causing QT prolongation) and that they follow the same underlying shape of =
PK-QT relationship. This is an assumption that could be rarely the case in =
real life. I would first try a different model. What needs to be added up i=
s the PD (i.e., QT prolongation) not the concentrations. If the individua=
l PK-QT relationships for Drugs A, B and the metabolite C are sufficientl=
y separated (i.e., different half-lives for the three compounds and have =
different PK-QT slopes or different EC50 and Emax values in the case of an =
Emax model), then you could try to fit a model like:


EFFA = SLA*CPA ;Effect of drug A=SlopeA x Plasma concent=
rations of drug A
EFFB= SLB*CPB ;Similar to above for drug B
EFFC= SLC*CPC ;Similar to above for metabolite C
EFF=EFFA+EFFB+EFFC ;Total drugs effect on QT
QTC = QTC0+EFF ;QTC0 is the baseline QTc effect (could be =
a diurnal variation model, etc)
Of course, the model requires enough data points and works if there is en=
ough separation between the slopes of the PK-QT effects. If the above mode=
l does not fit, I would next use the pre-clinical data (e.g., hERG test=
) to eliminate at least one of the moieties for potential QT prolongation a=
nd that should simplify the model.




Parviz Ghahramani, PhD, PharmD, MSc, MBA
Executive Director, Clinical Pharmacology and Drug Dynamics
Forest Research Institute
A Subsidiary of Forest Research Laboratories, Inc.
Harborside Financial Center, Plaza V
Jersey City, NJ 07311
201-427-8469 (office)
917- 828-3836 (cell)
201-427-8498 (fax)



Date: Thu, 14 Jan 2010 09:24:15 -0600
From: prhutson
Subject: Re: [NMusers] PD model
CC: nmusers

Otilia makes useful recommendations, and I would just reinforce the cauti=
on of having a good baseline and an understanding of the individual's diurn=
al changes. There is increasingly recognized to be a very large natural va=
riability in QTc that will need to be considered. Daily swings of 90msec a=
re apparently not uncommon when 24 hr recordings are made in normal adults.


Dear Otilia,

Thank you very much for your suggestions. They are very helpful.



---------- Original Message ----------
From: Lillin - de Vries, O. - <otilia.lillin
To: <chenyuhong
Subject: RE: [NMusers] PD model
Date: Thu, 14 Jan 2010 11:45:16 +0100

Dear Yuhong,
Here you have my 5 cents (in top-down fashion):
1. You need a PK-PD model quantifying the QTc prolongation; since you don=
't know what causes the QTc prolongation, this breaks down to testing:
   1a. DrugA - QTc model
   1b. DrugB - QTc model
   1c. DrugA+DrugB - QTc model
   1d. DrugC - QTc model
In order to be able to compare models 1a - 1d you need a good Baseline QTc =
model (i.e you need to describe well all non-drug related influences on QTc=
 like gender, circadian rhythm, age, placebo effect, etc - see e.g.=
 the paper of V. Piotrovsky, Pharmacokinetic-Pharmacodynamic Modeling in =
the Data Analysis and Interpretation of Drug-induced QT/QTc Prolongation,=
 AAPS Journal 2005).
If a combination of more than one of your three moieties can be responsible=
 for the effect on QTc (I do not have experience with this case) I would si=
mply add upp the concentrations (see the Nonmem code below for your conveni=
ence, without circadian rhythm for keeping it simple).

On the other hand be aware that a PK-PD model can not tell you for sure whi=
ch one of the moieties is responsible for the QTc prolongation; a model c=
an only quantify the magnitude of the effect and give you a hint on which m=
oiety is most probably causing it.

You need to make assumptions on physiological bases as well, and:
- check whether drugB alone causes QTc prolongation (model 1b? litterature?=
 previous studies with limited ECG? ...) if yes, you need model 1c.
- check the time point at which you have the largest QTc prolongation: does=
 it occur at Tmax_drugA? then a direct effect model (1a or 1c) are most pro=
bable; does it occur at Tmax_drugC? since C is a metabolite, it takes s=
ome time to be formed and probably Tmax_C > Tmax_A, this hints you in the=
 direction of the metabolite and you need a delayed effect model to describ=
e the parent's effect on QTc (1a) and the concentrations in the hypothetica=
l compartment should agree with the metabolite profile. In other words, y=
ou should be able to tell the effect from the parent(s) and metabolite from=
 each other.
Hope this helps.
OCC2=0 ; stead=
y state
IF (DAY.EQ.11) OCC2=1
QTC0 = THETA(1)+ETA(1) ;baseline QTc
SHFT = THETA(2) ;shift fact=
or placebo effect
QTCB = QTC0+SHFT*OCC2+THETA(3)*GEN ;baseline with placebo and gend=
er effect
SL = THETA(4) ; slope =
of drug effect
CP = CA + CB ; add upp=
 concentrations causing the effect
EFF = SL*CP ; linear=
 direct effect
Y = QTC+EPS(1)

Otilia Lillin-de Vries, MSc
Modeling and Simulation Expert

Pharmacokinetics, Pharmacodynamics & Pharmacometrics (P3)
Department of Drug Metabolism and Pharmacokinetics
T: +31 412 669321
M: + 31 6 22004827
F: +31 412 662506

Gasstraat Oost 10, 5349 AV, Oss
P.O. Box 20, 5340 BH, Oss
The Netherlands

From: owner-nmusers
 Behalf Of chenyuhong
Sent: Thursday, 14 January, 2010 2:45
To: nmusers
Subject: [NMusers] PD model

Dear All,

I am looking for a help. Currently I am working on a population PD model to=
 evaluate the effects of drugs on QT prolongation. Drug A and drug B are g=
iven to the study subjects (healthy volunteer) at the same time. Drug B is =
the inhibitor for the metabolism of drug A, also compound C is the metabo=
lite of drug A. I am wondering how to evaluate the effects for drug A or B=
 or the metabolite of drug A (compound C). These three moieties will be pre=
sent together in the blood for most of the time. Is anyone has experience a=
nd would like to share with us. Any comment will be greatly appreciated.

Best regards,



Diet Help
Reach your goals of being healthier and happier. Click here for diet tips a=
nd solutions.

This message and any attachments are solely for the intended recipient. If =
you are not the intended recipient, disclosure, copying, use or distr=
ibution of the information included in this message is prohibited --- Pleas=
e immediately and permanently delete.

Weight Loss Program
Best Weight Loss Program - Click Here!


Paul R. Hutson, Pharm.D.
Associate Professor
UW School of Pharmacy
777 Highland Avenue
Madison WI 53705-2222
Tel 608.263.2496
Fax 608.265.5421
Pager 608.265.7000, p7856
Hotmail: Trusted email with powerful SPAM protection.

Received on Wed Jan 20 2010 - 11:13:26 EST

The NONMEM Users Network is maintained by ICON plc. Requests to subscribe to the network should be sent to:

Once subscribed, you may contribute to the discussion by emailing: