Back to the Top
I'm looking for information on PK/PD modeling options and approaches in an upcoming 2x2 cross-over design study.
Subjects (n = 40) will receive either placebo, compound 1, compound 2, or the combination, all as single dose (oral) occasions, on separate days with sufficient wash-out. The PK characteristics of two compounds of interest are available, and sufficient data-points will be taken to get decent fitted curves, However, the assessment of effect (the PD part) is limited for practical reasons. I.e. in our case the PD effect will be obtained at 3 time points, of which one as just prior to dose (t = 0), the second at around Tmax and the last one a while after Tmax. PD measurement are done on all 4 study days of the 2x2 design.
The Tmax values for cpd 1 and 2 are about the same, but Cmax values will differ and Kel's differ. For one compound we also have reasons to expect a large inter-individual variation in metabolism, due to CYP polymorphism, resulting in a rather wide range of Kel values. With such a design it is not obvious to model the PD effect given the limited data points as function of time.
The question is now whether colleagues have experience with building a PK-PD model in such a design, or that the effect measurements are too limited to get useful information.
Dr Guus Duchateau
science leader Bioavailability & ADME
Unilever R&D Vlaardingen
PO Box 114, 3130 AC Vlaardingen
The Netherlands
Back to the Top
Dear Guus,
The information you provide on the type of effect is rather limited to allow for a specific recommendation. However, let's discuss the design and whether it would be possible to model a PK/PD correlation.
I am not sure why a sample is taken at the presumed tmax. I understand that it is rather common to sample at tmax but I believe it has more to do with peoples' habits than modern science. The idea of effect measurement at tmax is to be correlate Cmax to the maximum observed effect. In reality, however, for majority of cases, the observed effect does not occur at tmax but follows drug concentrations with some lag time. If you are thinking of a PK/PD model, which I strongly believe is the right way to analyze the data, it is not a requirement to take the sample at tmax. What would help, especially if you plan a mixed effect modeling approach, would be to spread the sampling times in different individuals so you are covering a wide range of time points and hence effects. What I normally recommend to my clients when they can collect limited number of samples is to not ask for a specific time point but rather give a time interval for the sample to be collected. As an example, one sample pre-dose, ano
Back to the Top
The following message was posted to: PharmPK
Dear Guus
> The question is now whether colleagues have experience with building a PK-PD model
> in such a design, or that the effect measurements are too limited to get useful
> information.
Quick answer - no idea.
Longer answer.
The sampling times that you indicate to take your PD measurements may be, by chance, at very good times and may be, by chance, at very poor times. It is possible also that your PD model could be quite complex and require a greater number observation times to characterise.
So it seems to me that this is really a question about optimal resource allocation. You need to balance the risk of not learning about your PD from your current study given the resources you will put into this study against the resource requirements for you to design a robust study, i.e. where you use prior data, perhaps in animals or from similar compounds, then model this data and then propose a likely time course of PD effect and then design a robust study accordingly.
Regards
Steve
--
Professor Stephen Duffull
Chair of Clinical Pharmacy
School of Pharmacy
University of Otago
PO Box 56 Dunedin
New Zealand
E: stephen.duffull.-a-.otago.ac.nz
Back to the Top
Guus,
It may be helpful for you to read something about PKPD so that you can understand better what you are trying to achieve and how to achieve it. You use SHAM (shape height area moment) PK terms (Cmax, Tmax, Kel) but these are not really much help for PKPD so I suggest you come to grips with model based PK approaches. Then it will become easier to understand how to link drug concentrations to effect.
Once you understand the fundamentals of PKPD you will then find it easier to appreciate how to optimize the design for a clinical study such as the one you have outlined.
If you know nothing about how the mechanism of drug effect and how it's time course is related to plasma concentration then you could improve your design easily by not putting all your eggs in one basket. Spread out the sampling time for the response measurements so that they give you more information instead of limiting yourself to fixed "Tmax" and "a while after Tmax" conc times. You have a lot of subjects (40) and with 4 obs/subject after active treatment you could have 160 different times instead of just 2 times to help you learn about the time course of response in relation to plasma concentration.
It is hard to know why you are confident that all drug effects will have washed out on the different study days. The pre-dose response (and conc) measurement will let you check on this assumption as well as letting you learn about the time course of the biological response in the untreated state. However, if you find your assumption is wrong what will you do? This may require a model for the 'baseline' also known as the natural history or disease progression model as well as a PKPD model that can predict responses at times when you had assumed all drug effects would have washed out. Here is an example of a big trial that failed to come to any useful conclusion by making a poor assumption about washout.
The Parkinson Study Group. Levodopa and the progression of Parkinson's disease. N Engl J Med. 2004;351(24):2498-508.
Jumping into a combination study of compound 1 and compound 2 seems a bit premature when you seem to know nothing about the response to each compound alone. The design of combination studies is quite a bit harder and depends very much on just what kind of question you are trying to answer. Because the Tmax for both compounds is similar there will be a very high correlation between the ascending part of the concentration time courses of both compounds and you may learn very little about the combined effects by having an effect measurement at Tmax.
These ideas have been around for a long time and there are centres of excellence not far away from you e.g. http://www.tipharma.com/projects/mechanism-based-pk-pd-modeling.html
Nick
Back to the Top
The following message was posted to: PharmPK
Dear All
When a drug is administered, after absorption it reaches the systemic
circulation and from there it moves to site of action or get distributed to
some of the tissues (kidney, liver, Cancer cells if it is a targeted for
that) and in mean time it get eliminated from systemic circulation by
metabolism through liver/kidney etc.
But, even the tissue where it has been distributed keeps on proliferating
or growing thus increasing the number of cells available for metabolism or
in other term with increase in the number of cell, the metabolic capability
increases thus that may lead to increase metabolism of the drug in that
specific region or tissue.
My question is that
1. Is there will be any effect on the PD out come because of the increased
metabolism/ increased metabolic activity of the tissue where the drug is
distributed
or it is simply controlled by the systemic clearance?
2. Is there any PK PD model is available for the same?
3. If no PK PD model available, what approaches one should try for to
generate one.
Warm Regards
--
Bhupesh
Want to post a follow-up message on this topic?
If this link does not work with your browser send a follow-up message to PharmPK@boomer.org with "PK-PD modeling" as the subject | Support PharmPK by using the |
Copyright 1995-2011 David W. A. Bourne (david@boomer.org)