C. Falcoz (2), S. Bozec (1), D. Cravo (1), H. Merdjan (2), S. Bolze (1)
(1 )Poxel SA, Lyon, France, (2) Pharsight Consulting Services, a division of Certara, St. Louis, MO, US
Objectives: The aim was to predict a likely clinical efficacious dose for PXL, a compound with a new mechanism of action developed in type 2 diabetes, based on in vitro (IVT) efficacy data, PKPD modeling in ob/ob mice, one of the disease models, and a single endpoint, glucose.
Methods: The exploration of early and sparse PK and glucose data justified a PKPD approach based on glucose only levels. Specific PK and PKPD studies were designed under experimental constraints, and population PK and PKPD (PopPK, PopPKPD) models developed in normal and ob/ob mice. Phoenix® was used for exploratory analysis and Phoenix® NLMETM for nonlinear mixed-effect modeling. Simple allometry was used to predict human clearance (CL). A likely clinical efficacious dose was predicted using the IC50 estimated in mice, scaled to man based on IVT differences between species and plasma binding.
Results: Data from the first pharmacological experiments showed that glucose levels 1 h post dose after 8 days of treatment were correlated to PXL concentrations. Glucose and PXL concentrations were available from the new studies over 8 h following the morning dose (0, 25, 50 or 100 mg/kg BID). PopPK. A two-compartment model was developed, with between-subject variability on CL, V and V2; the kinetics was proportional with dose. PopPKPD. There was an apparent diurnal pattern for glucose; however data were not available over 24 h and a constant baseline was used. The best model was an indirect-response model with inhibition of glucose production. Some model-misspecification was partly due to the circadian baseline which could not be accounted for. Using individual PK parameters allowed a better estimation of IC50. Allometric scaling. Scaling with unbound concentrations led to lower and tighter predictions of human CL (1.1 L/h). Human dose. The IVT difference in sensitivity between human and animal was 8 fold. The human dose providing on average 50 and 80% of maximal effect was predicted at 25 and 100 mg/d, respectively.
Conclusions: Integrating early sparse and literature information allowed designing studies which led to successful modeling. A two-compartment PK model was adequate. Using a population PKPD approach and IVT data in animal and man, human doses likely to produce a defined lowering effect on glucose were predicted, which should guide the FTIM study.
References:
[1] Landersdorfer CB, Jusko WJ (2008). Pharmacokinetic/pharmacodynamic modelling in diabetes mellitus. Clin Pharmacokinet 47:417-48
[2] Sukumaran S, Jusko WJ, DuBois DC, Almon RR (2011). Mechanistic modeling of the effects of glucocorticoids and circadian rhythms on adipokine expression. J Pharmacol Exp Ther 337:734-46
[3] Ring et al. (2011). Predicting Drug Clearance in Humans using Non-clinical Data. Part 3: Comparative Assessment of Prediction Methods of Human Clearance. J Pharm Sci 100:4090-4110
Reference: PAGE 21 () Abstr 2420 [www.page-meeting.org/?abstract=2420]
Poster: Other Modelling Applications