Artefactual inflation of pharmacokinetic difference between two Granulocyte Colony Stimulating Factor (G-CSF) drug products by non compartmental analysis.
E. Pigeolet (1), P. Wiczling (3), P. Lowe (1), S. Balser (2), A. Berghout (2) and W. Krzyzanski (4)
(1) Novartis Pharma AG; (2) Sandoz Biopharmaceutical Development; (3) Biopharmaceutics & Pharmacodynamics, Med. University Gdansk; (4) Pharmaceutical Sciences, State University New York
Objectives: Conventional bioequivalence analysis between Zarzio(r) and Neupogen(r) (two formulations of G-CSF) showed that their Cmax and AUC ratios were decreasing between single and repeated administration at 2.5 and 5 but not 10 ug/kg daily dose. The aim of the modeling analysis was to assess whether the drift in these ratios when doses are decreased or repeated could be explained by the mechanisms underlying the well known pharmacokinetic non-linearity for this drug.
Methods: Rich sampling pharmacokinetic and pharmacodynamic (blood neutrophil count) profiles of 112 healthy male and female volunteers were evaluated. G-CSF was administered as repeated s.c. daily administration for one week of 2.5, 5 and 10 ug/kg doses and single i.v. dose (5 ug/kg) in a cross-over design. A semi-mechanistic population PK/PD model was built from these data. From the parameter estimates and assuming a rapid equilibrium, the ratio of unbound drugs (what is measured by the assay) was computed for 10, 2.5 and a low dose of 1 ug/kg doses.
Results: The pharmacokinetic part was a one compartment binding model. The pharmacodynamic part of the model was described by bone marrow, blood and tissue compartments for neutrophils. The serum G-CSF concentration was stimulating the proliferation of neutrophils. The total number of neutrophils was driving the amount of receptors capturing the unbound drug in the binding model. The pharmacokinetic part only was fitted on the single administration data with very good diagnostic plots. The PK/PD model was fitted on the full dataset with reasonably good diagnostics although some bias could still be detected. The unbound amount of G-CSF was computed from the parameters of the PK only model except the receptor amount whose relative increase after repeated dose was simulated from the full PK/PD model. If there was between the 2 drug products a true 4% difference in the drug amount systemically available at all doses, the model based computed ratio of unbound drug amount was 0.961, 0,958 and 0,952 after a single dose and 0.955, 0.937 and 0.892 after 7 daily administrations of 10, 2.5 and 1 ug/kg respectively, indeed indicating a drift away from unity, when dose is decreased, especially after repeated administration.
Conclusions: The inflation of a small pharmacokinetic difference when doses are decreased or repeated is a plausible hypothesis for a drift of Cmax and AUC ratio away from unity when measured through the unbound G-CSF concentrations. As suggested by the model, this inflation is due to the non-linear receptor mediated drug disposition.
