B. Mc Hugh, N. Frey
Clinical Pharmacology, F. Hoffmann-La Roche, Basel, Switzerland
Objectives: To develop a structural PK model and a structural PK/PD model in healthy volunteers.
Methods: Single doses of 10, 25, 50, 100, 200, and 400 mg were administered to 48 healthy volunteers. Data consisted of 16 plasma enzyme inhibitor concentration and enzyme activity measurements between dosing and 30 hours. A NONMEM population analysis to describe the PK commenced with a simple model and progressed with adding complexity. A NONMEM population analysis to describe the PK/PD relationship was implemented by considering effect-compartment, indirect effect, and slow-binding models (Äbelö, et al.). Since the baseline enzyme activity was observed to vary greatly, pre-dose enzyme activity was treated as a covariate for each volunteer.
Results: A two-compartment model with first order absorption and a combination of first order and Michaelis-Menten elimination adequately described the PK. The First Order (FO) method of NONMEM was used. The multiplicative error component had a standard deviation near 17% and the additive component was three times the limit of quantification. With the exception of Michaelis-Menten constant and the metabolism velocity, the PK parameters were well-estimated (standard error < 30%). No bias was observed when the individual weighted residual error was depicted versus either time or predicted concentration. After a time lag estimated to be 0.24 [h], the absorption was rapid (ka= 9.3 [h-1]). Complete metabolic saturation resulted in a total clearance one third lower than its maximum value (the first order clearance was estimated to be 31 [L/h]; the Michaelis-Menten constant and maximum metabolism velocity were estimated to be 272 [mg/L] and 4.5 [mg/h], respectively). Regarding the population PD parameters, the slow-binding model, i.e., a slow dissociation of the drug from the enzyme, best described the PK/PD of the drug. The FO method of NONMEM was used, no bias was observed, and the parameters were well estimated (standard error < 10%). The estimated rate constants for enzyme deactivation and subsequent activation were 0.075*C [h-1] (C [mg/L] is the inhibitor concentration) and 0.85 [h-1], respectively. In all cases other than very low drug concentrations, the binding rate to the enzyme was predicted to be much faster than the dissociation rate.
Discussion/Conclusion: This present PK model with a Michaelis-Menten component of elimination is consistent with the existing knowledge of the drug, i.e., there is a high likelihood to saturate a metabolic pathway. The poorly estimated Michaelis-Menten parameter and the metabolism velocity can be attributed to the small study population. The preliminary structural PK and PK/PD models in healthy volunteers will be enhanced as additional study information becomes available and will facilitate the analysis of data from patients.
References:
Angela Äbelö, Johan Gabrielsson, Björn Holstein, Ulf G. Eriksson, Johan Holmberg, Mats O. Karlsson; Pharmacodynamic modelling of reversible gastric acid pump inhibition in dog and man; European Journal of Pharmaceutical Sciences 14 (2001) 339-346
Reference: PAGE 13 (2004) Abstr 499 [www.page-meeting.org/?abstract=499]
Poster: poster