A pharmacokinetic/pharmacodynamic analysis of central and peripheral effects of GSK3 inhibitors
K. Tunblad(1), E. van Schaick(2), A. Mörtberg(1), K. Agerman(1), S. Rosqvist(1), P. I. Arvidsson(1), S. Visser(1)
(1) DMPK, Neuroscience and Chemistry Departments, CNSP iMed Science Södertälje, AstraZeneca, Sweden; (2) Exprimo NV, Belgium
Objective: The aim was to perform a pharmacokinetic-pharmacodynamic analysis of the central and peripheral effects of Glycogen Synthase Kinase 3 (GSK3) inhibitors in rat pups, and to correlate these effects to in vitro pharmacology measures. The influence of plasma protein binding and distribution to the brain was also investigated.
Methods: PKPD data for several GSK3 inhibitors at several dose levels was collected. Tau-phosphorylation in the hippocampus (central biomarker) was measured at two epitopes (AT8 and AT180), and the phosphorylation of glycogen synthase (P-GS) in muscle was measured as a peripheral biomarker. Exposure in plasma and in hippocampus and central effects and the peripheral effect were measured at one time-point in each individual. For the model building a mixed approach was used where "individual data" were created by grouping data per study and per compound. All compounds were simultaneously fitted in one model, where drug specific and system specific parameters were estimated. The variability between different study/drug combinations was also evaluated.
Results: The plasma concentrations were used to drive the response. The central effects were best described by indirect response models, while a direct effect model described the peripheral effect. The average half-life for the delay in the central effect was 2.7 h for AT180 and 4.1 h for AT8. IC50's were estimated for all effect measures and the rank order of these values was similar for the three effect measures. In the indirect response models the system specific parameters (Kin, Kout and hill slope) were estimated. In the direct effect model the baseline and hill slope were estimated. In both models Imax was fixed to 100%. There was a good correlation between in vitro data and in vivo parameter estimates. Also, the IC50 for peripheral and central effects showed a good correlation. Correcting for plasma protein binding and distribution to the brain did not improve the correlations.
Conclusion: The results indicate that the response on P-GS seems to be a good predictor of compound potency for central effect, although not for the time-course of this effect. This provides a translational opportunity to human by characterizing P-GS as a peripheral biomarker also in humans. Additionally the good in vitro-in vivo correlation can be used to benchmark new candidate drugs aimed for inhibition of GSK3 in vitro, thereby reducing the need for in vivo screening experiments.