- On 9 Sep 2002 at 14:02:08, Weijiang Zhang (ZHANGW.at.mail.rx.uga.edu) sent the message

Back to the Top

Hello everyone:

Could anyone give me some help for determining the minimum number of

samples (or subjects) needed for a population pharmacokinetics? Is

there an easy to estimate the number of samples needed using the

variability (for example, the %CV of important PK parameters is

around 100%)? Or I have to do some kind of simulation?

Thank you very much!

Best Regards,

Weijiang Zhang - On 9 Sep 2002 at 18:19:51, "Serge Guzy" (GUZY.aaa.xoma.com) sent the message

Back to the Top

I DON'T THINK THERE IS A GOLD STANDARD BECAUSE EACH PROBLEM YOU ARE

DEALING with IS ASSOCIATED WITH A DIFFERENT KIND OF SENSITIVITY OF

THE PARAMETERS WITH RESPECT TO THE OBJECTIVE FUNCTION YOU ARE TRYING

TO OPTIMIZE. The type of model you are using is of course an

important component to take into consideration (1 compt model vs 2

compt model)but all other components are also

important(intraindividual noise, interindividual variability etc...).

Combination of simulation with subsequent fitting and reliability

procedures will help you to assess this kind of information for the

specific problem you are dealing with.

Serge Guzy, Ph.D.

Head of Pharmacometrics - On 9 Sep 2002 at 23:40:47, "Steve Duffull" (sduffull.-at-.pharmacy.uq.edu.au) sent the message

Back to the Top

Weijiang,

> Could anyone give me some help for determining the minimum number of

> samples (or subjects) needed for a population pharmacokinetics?

This is not an uncommon question. There is no easy answer. It will

depend on the likely PK profile that you are expecting to see and on

why you are doing the PK study. In essence the more you look for the

more you will see - so intensive sampling may well reveal details

that sparse sampling cannot support. Therefore if you are in the

exploratory phase then you should try and aim for a more intensive

sampling strategy than if you are in a more confirming phase of

development (where 'sparser' sampling may be reasonable).

If you are thinking of simulating then you obviously have a good feel

for the data that you are likely to see and therefore could also

optimise the design using optimality-type procedures (DOE). The PFIM

functions in MATLAB and SPLUS can help here (see link from website

below).

As a general rule, for less intensive designs, you should always try

and set the *minimum* number of samples to be equal to the number of

fixed effect parameters to estimate; otherwise the efficiency of the

design may be considerably reduced. The number of patients and

number of occasions will depend on the variance models that you

believe are true.

Without more details it is hard to be more specific.

Regards

Steve

Stephen Duffull

School of Pharmacy

University of Queensland

Brisbane 4072

Australia

http://www.uq.edu.au/pharmacy/duffull.htm - On 10 Sep 2002 at 10:13:49, "Sandrine MICALLEF" (Sandrine.Micallef.at.ineris.fr) sent the message

Back to the Top

Weijiang,

You can find a way to answer such a problem in the following paper :

Bois, F. Y., T. J. Smith, et al. (1999). "Optimal

design for a study of butadiene toxicokinetics in humans."

Toxicological Sciences: an Official Journal of the Society of

Toxicology 49(2): 213-224.

This can help you to answer many questions.

Regards,

Sandrine.

Sandrine MICALLEF

Toxicology Lab

INERIS

Parc ALATA, PB2

60550 Verneuil en Halatte

FRANCE - On 10 Sep 2002 at 12:08:08, Weijiang Zhang (ZHANGW.-a-.mail.rx.uga.edu) sent the message

Back to the Top

Thank you all for your response about my question. It helped me to

clear my mind and know where to start to think about this issue.

Thanks.

In fact, this is a pediatric pharmacokinetic study (for <1 year old

infant). The primary objective of the study is to get the general PK

profile in this population. Therefore I think I don't need to use

the mixed effect model to identify the covariate and residual

variability, probably naive pooled approach will be enough. Right?

This compound fits one compartmental model well in adults, the

sampling schedule in adults is 0, 0.5, 1, 1.5, 2, 3.5, 6, 9, and 12

hrs.

Any suggestions or comments will be highly appreciated.

Thanks.

Best Regards,

Weijiang - On 10 Sep 2002 at 17:25:24, David Bourne (david.aaa.boomer.org) sent the message

Back to the Top

[Three replies - db]

From: "Mishina, Elena V"

Date: Tue, 10 Sep 2002 14:26:14 -0400

To: david.at.boomer.org

Subject: RE: PharmPK Re: Sampling number for population pharmacokinetics

Dear Weijiang,

Usually when the drug is to be studied in pediatrics, the FDA requirement is

to have 4 groups of pediatric patients to include all ages from infants to

adolescents. If your study is planned as a one-site study to assess the PK

in infants as a part of multiple sites study you should not use a naive pool

approach for your group of children. Instead, you should have a few samples

taken from each patient and include your data into the population data

analysis for the whole pediatric population. It is very important to

evaluate the influence of the covariates on the PK parameters estimated for

the pediatric patients. This will help to make a decision on the dose

adjustment for children (if necessary).

Regards,

Elena Mishina, Ph.D.

Senior Clinical Pharmacologist

US Food & Drug Administration

---

From: Nick Holford

Date: Wed, 11 Sep 2002 07:01:50 +1200

To: david.-a-.boomer.org

Subject: Re: PharmPK Re: Sampling number for population pharmacokinetics

"Weijiang Zhang (by way of David Bourne)" wrote:

> In fact, this is a pediatric pharmacokinetic study (for <1 year old

> infant). The primary objective of the study is to get the general PK

> profile in this population. Therefore I think I don't need to use

> the mixed effect model to identify the covariate and residual

> variability, probably naive pooled approach will be enough. Right?

Wrong! In the 0 to 1 year age range is exactly the period of life

when major changes take place in PK parameters. There are very real

fixed (weight, age) and random (between subject variability) effects

that need to account for individual differences. A non-linear mixed

effect model is essential to be able to describe the changes

conditional on weight and age. A naive pooled analysis is guaranteed

to be misleading.

See Anderson BJ, van Lingen R, Hansen TA, Lin Y-C, Holford NHG.

Acetaminophen developmental pharmacokinetics in premature neonates

and infants. Anesthesiology 2002;96(6):1336-1345 for a recent example

and references to earlier literature.

--

Nick Holford, Divn Pharmacology & Clinical Pharmacology

University of Auckland, 85 Park Rd, Private Bag 92019, Auckland, New Zealand

email:n.holford.at.auckland.ac.nz

http://www.health.auckland.ac.nz/pharmacology/staff/nholford/

---

From: "Sam Liao"

Date: Tue, 10 Sep 2002 16:23:11 -0400

To: david.aaa.boomer.org

Subject: RE: PharmPK Re: Sampling number for population pharmacokinetics

Hi Weijiang:

You can use less frequent sampling schedule in your ped PK study if you use

the mixed effect model.

Such as 0, 1, 2, 4, 8, and 12 hr post-dose, which is more likely be

acceptable to the physician and parents.

Best regards,

Sam Liao, Ph.D.

PharMax Research

270 Kerry Lane,

Blue Bell, PA 19422

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 "Sampling number for population pharmacokinetics" as the subject

PharmPK Discussion List Archive Index page

Copyright 1995-2010 David W. A. Bourne (david@boomer.org)