Nathalie Perdaems (1), Valérie Duvivier (2), Sylvain Fouliard (3), Marylore Chenel (4)
(1) Early PK and PKPD modelling, Technologie Servier, Orléans, France, (2) Metabolism, Institut de recherche Servier, Suresnes, France, (3) Early PK and PKPD modelling, Institut de recherches internationales Servier, Suresnes, France (4) Clinical PK and Pharmacometrics, Institut de recherches internationales Servier, Suresnes, France
Objectives: In preclinical research, pharmacokinetic/pharmacodynamic (PK/PD) studies are very often not enough discussed between pharmacologists and modellers before the pivotal in vivo study, and sometimes are poorly designed. This work is to make the pharmacologists, in vivo techniciens and project leader in research and development aware to PK/PD modelling.
Methods: Using separate studies (a PK study and a PD study), a PK/PD model was developed for a S compound in type 2 diabetes disease. A PK model was built to describe the PK of the S compound in ob/ob mice after a single oral administration. Then a PK/PD model was built, using simulated average plasma concentrations of the S compound in ob/ob mice after repeated oral administrations for 3 days. The limitations of this approach and the interest to have PK in the same studies as PD to well characterize the PK/PD relationship was discussed with people in charge of the project. Simulations were performed to show what could be done with a proper model [1]. Optimal sampling was used to choose the PK sampling times for the 28 days PKPD study in ob/ob mice: feasible designs were evaluated after an optimization step [2].
Results: Using the separate studies, a PK/PD model was built but some parameters were estimated with a relatively low precision of estimation (RSE% up to 140%). The PK of the S compound was assumed to be linear and the characterization of the respective PK and the PD variabilities was not possible, as PK was fixed, all the variability was estimated on PD. The optimized sampling times were 0.2, 1.8, 3.2, 5.2, 11.8 and 23.6 h. The interest to have sampling times after 6 h was confirmed by evaluating different experimental designs to improve the precision of estimated parameters. After discussion with the pharmacologists, the design of the 28 days PK/PD study was modified: sampling times for the PK analysis of the S compound in the 4 groups (with 4 different doses of S compound) and the two last sampling time were 8 h and 24 h.
Conclusions: Discussions between pharmacologists and modellers as early as possible in the preclinical development helps to better design PK/PD studies in order to well characterize the PK/PD relationship. To be part of the drug development at the beginning allows to propose a modelling strategy for the translational drug development from animal to human.
References:
[1] Phoenix, WinNonlin 6.3 (Pharsight)
[2] PopED lite (University of Uppsala, Sweden)
Reference: PAGE 25 () Abstr 5753 [www.page-meeting.org/?abstract=5753]
Poster: Methodology - Study Design