From:From: "Hu, Chuanpu" 
Subject:[NMusers] PD modeling question  
Date:Wed, February 27, 2002 10:24 am  


I am looking for suggestions for modeling the time course for the following
type of effect. The rate of onset of the drug effect is believed to be much
faster than the rate of offset, suggesting nonlinear mechanism. This may be
consistent with covalent binding and receptor turnover, or some homeostatic
mechanism. The exact mechanism of action is not well understood, and I am
looking for empirical, hopefully physiologically plausible PD models to link
predicted (imputed) plasma concentration levels to drug effect. 

Any help would be appreciated.


Chuanpu Hu, Ph.D.
Modeling and Simulation, U.S.
Tel: 919-483-8205  
Fax: 919-483-6380


From:"Gobburu, Jogarao V"   
Subject:RE: [NMusers] PD modeling question  
Date:Wed, February 27, 2002 11:33 am  

Dear Chuanpu,
I have to admit that I do not know enough about the specific case, so these
are just some thoughts for your consideration.

It appears from your message that your drug affects ENZyme which regulates X
and that you are measuring X. 
Although it might be challenging to delineate the exact mechanism, in my
honest opinion, we need to have a reasonable understanding of the underlying
mechanism. This is essential because model selection is going to be
difficult, otherwise. 

The two approaches you mention include: 1. having a binding component and/or
ENZ turnover and 2. homeostatic mechanism. I am not sure how you can
attribute the second mechanism to your observation. I would expect, the
second reason would speed-up the (ENZ turn-over) offset (let us say simply,
X returning to baseline). According to that, inspite of having increasing
amounts of the drug in the body, the effect should be diminishing (eg:

ACE inhibitors are known for their tight binding to ACE, experience from
them might be helpful. Also, Gisleskog et al present a nice model that takes
into account the turn-over of DHT (you migh or might not need the
irreversible part of the pd model). These will be good place to start (if
you have not considered them already).

Ref1. Toutain PL, Lefebvre HP, King JN. Benazeprilat disposition and effect
in dogs revisited with a pharmacokinetic/pharmacodynamic modeling approach.
J Pharmacol Exp Ther. 2000 Mar;292(3):1087-93.  

Ref2. Gisleskog PO, Hermann D, Hammarlund-Udenaes M, Karlsson MO. 
A model for the turnover of dihydrotestosterone in the presence of the
irreversible 5 alpha-reductase inhibitors GI198745 and finasteride. Clin
Pharmacol Ther. 1998 Dec;64(6):636-47.

Ref3. Bauer JA, Fung HL. Pharmacodynamic models of nitroglycerin-induced
hemodynamic tolerance in experimental heart failure. Pharm Res. 1994

Joga Gobburu,



Subject:Re: [NMusers] PD modeling question  
Date:Wed, February 27, 2002 11:58 am  

The simplest approach is to model the receptor's dynamics shifting between 
active (R*) and inactive (R) states:

(1) dR*/dt = kon R C - koff R*,

where C is observed drug conc. or Ce,
and subject to the constraint R + R* = Rtot.

Then, if 

(2) E/Emax = R*/Rtot, 

dividing (1) by Rtot and substituting E/Emax for
R*/Rtot and rearranging yields

dE/dt = Emax kon C - (kon C + koff)E,

which has onset as rapid as you wish (controlled by C),
and offset as slow as you wish (controlled by koff).
A more complex relationship than the linear one (2) between E and R*
is also possible.

To identify this model and distinguish it from "effect compartment model", 
especially when experiment is undertaken in
saturable effect range, requires EXQUISITE data.