Back to the Top
In the analysis of pharmacodynamic data are there ever any circumstances
where transformation of the response is justified for model development?
For example if one is measuring blood pressure but models the area under
the effect curve. Is this okay?
Back to the Top
[A few replies - db]
From: "Rory Conolly"
Date: Wed, 20 Dec 2000 17:34:10 -0500
To: david.at.boomer.org
Subject: RE: PharmPK Pharmacodynamics
The following message was posted to: PharmPK
Think about how the ultimate endpoint - disease state - is related to
whatever PD endpoint you are measuring. For some disease endpoints, AUC of
the PD endpoint may make sense while maximum effect, or AUC above some
threshold level of effect may be more relevant for others.
================================
Rory B. Conolly, Sc.D., D.A.B.T.
Senior Scientist
CIIT Centers for Health Research
6 Davis Drive
Research Triangle Park
North Carolina 27709 USA
voice: (919) 558-1330
fax: (919) 558-1300
e-mail: rconolly.-a-.ciit.org
================================
---
From: ml11439.at.goodnet.com (Michael J. Leibold)
Date: Thu, 21 Dec 2000 01:15:23 -0700 (MST)
To: david.-a-.boomer.org
Subject: Re: PharmPK Pharmacodynamics
The following message was posted to: PharmPK
Pharmacodynamics,
In a basic sense, pharmacodymamic models are frequently based
on a percent normal response, or a percent response versus control.
For example, warfarin pharmacodynamics are based on the percent
normal prothrombin complex activity, which is a transformation of
the prothrombin time. So, it would seem that your data could be
modeled as a percent change versus control, if there is a correlation
with plasma levels of drug.
Mike Leibold, PharmD,RPh
ML11439.at.goodnet.com
---
From: bangarurk.aaa.drreddys.com
Date: Thu, 21 Dec 2000 16:16:34 +0530
To: david.-at-.boomer.org
Subject: Re: PharmPK Pharmacodynamics
The following message was posted to: PharmPK
Yes, the transformation of data is justified as an example the blood
pressure drop as a parameter gives only the percent drop and does not tell
about the efficacy but if you just Area between 30% drop and 10% drop which
is critical it gives a fairer estimation of the efficacy parameter and you
can develop a model using the AUC. I hope it is convincing.
RK bangaru
bangarurk.-at-.drreddys,com
---
From: "Jogarao Gobburu 301-594-5354 FAX 301-480-3212"
Subject: Re: PharmPK Pharmacodynamics
To: "postmaster"
MIME-Version: 1.0
Posting-date: Thu, 21 Dec 2000 08:52:24 -0500 (EST)
Importance: normal
Priority: normal
Sensitivity: Company-Confidential
A1-type: MAIL
X-WSS-ID: 165CD53875212-01-01
Hello,
1. Usually, it is better to model using the raw data. On the contrary,
it is probably more mechanisitically sound to model [H+] (proton
concentrations) than the measured pH. But again, these decisions are
dependent upon the aim (see below).
2. Good modeling practices call for clear identification of the
modeling exercise's objective(s). I provide 2 examples below:
2a. Aim: To design an optimal drug delivery rate
If your aim is to design an optimal dosing regimen then AUEC will not
help you. Why not develop a concentration - BP relationship? I am
assuming you have measured the drug concentrations. You can derive any
secondary PD parameter that you might want, from such a model.
2b. Aim: To evaluate BP as a surrogate for predicting the probability of
the occurrence of some long-term cardiovascular event (time to occurrence,
yes/no, etc..). Let us say, you neither have drug concentrations nor can
simulate them.
If you are tying to model the effect of the drug on the BP, and further,
the cumulative effect of lowering BP (say) on some long term
cardiovascular event (LTCVE). Though not by choice, you might have to
live with dose - AUEC - LTCVE relationship.
If you have the concentrations, you can develop a dose-conc- BP/AUEC -
LTCVE relationship.
Hope this helps.
Regards,
Joga Gobburu
Pharmacometrics,
CDER, FDA.
PharmPK Discussion List Archive Index page
Copyright 1995-2010 David W. A. Bourne (david@boomer.org)