Stuart Beal, Ph.D.
Univ. of California at San Francisco
How to design a PK/PD trial? An answer for a specific situation involves many considerations of many different types. Following current practice, it will probably evolve as the result of repeated group discussions. These discussions will entail attempts to imagine suitable designs. Software which outputs suitable designs under the multitude of constraints that must be imposed on the trial will probably not exist for along while, if ever. However, software already exists to immediately and conveniently simulate trial responses under a design that may be specified as an input, allowing this design to be tested, to be modified depending on the result of the testing, to be retested, etc.
The NONMEM system can be used for this purpose. An example of simulating population PK data under a couple of different designs will be given. A design must be chosen according to what type of information results from using it, and this in turn depends on the results of data analysis. While the idea of using NONMEM to simulate population PK data may be relatively unfamiliar to many, the idea of using NONMEM to analyze such data is quite familiar. It is the iteration of simulation and analysis together that allows a trial design to be tested.
NONMEM can handle continuous PK/PD outcomes, but also other types of out- comes as well. An example of simulation and analysis of an animal toxico- logical experiment involving dichotomuous responses will also be given.
Reference: PAGE 5 () Abstr 569 [www.page-meeting.org/?abstract=569]
Poster: oral presentation