Matuzumab – Evaluation of the Population Pharmacokinetic Model and Analyses of the Covariate Impact on the Pharmacokinetic Profile
K. Kuester(1,3), A. Kovar(2), C. Lüpfert(2), B. Brockhaus(2), C. Kloft(1,3)
(1)Freie Universitaet Berlin, Dept Clinical Pharmacy, Berlin, Germany; (2)Merck KGaA, Dept Human Pharmacology, Darmstadt, Germany; (3)Martin-Luther-Universitaet Halle-Wittenberg, Dept Clinical Pharmacy, Halle, Germany
Objectives: Matuzumab is a humanised monoclonal antibody of the immunoglobulin subclass IgG1 which targets the epidermal growth factor receptor (EGFR). A previously developed population pharmacokinetic (PK) model was to be evaluated and the impact of the covariate relation was to be analysed.
Methods: The developed model from 1256 serum concentrations (90 patients) had been analysed using the software program NONMEM (ADVAN6, TRANS1, TOL5 and the FOCE INTERACTION estimation method). For internal evaluation the bootstrap method, visual predictive checks (VPC: for single dosing and for multiple dosing, separately for each of the eleven dosing regimens) and case deletion diagnostics (CDD, deletion of either 10% of the study subjects [ID] or of 1 ID and reestimation of the parameters with the reduced datasets) were performed.
Results: The developed two compartment model, including a linear and a nonlinear elimination pathway, interindividual and interoccasion (IOV) variabilities and a covariate relation, showed its ability to accurately estimate all model parameters by the bootstrap method. The calculated bootstrap means, bias and relative bias (-3.9% and +4.7%) were obtained from 200 successful bootstrap runs. In the VPCs, the 90% prediction interval included most of the actual observed concentrations and the calculated medians were in accordance to the original data. CDD showed an influence of 1 ID on the IOV (without this ID the IOV value decreased by 17%). Closer examination revealed that the influence was due to 1 observation from this ID (neglecting this observation led to a reduction of 19% of the IOV).
The influence of the covariate relation fat-free mass (FFM) on the linear clearance was analysed by simulation of all IDs receiving 1200 mg weekly. Observed and simulated interindividual variability in steady-state concentrations was reduced by a proposed adapted dosing regimen on a basis of mg per fat-free mass kg. The regimen comprised the original dosing (e.g. 1200 mg weekly) for the first 4 weeks followed by an adapted weekly dosing (based on the fractional covariate influence on total clearance) with 50% of the original amount (e.g. 600 mg) plus a proportionally adapted amount (e.g. 11.4 mg/kg FFM).
Conclusion: The population pharmacokinetic model for matuzumab has successfully been evaluated by different methods. The variability in simulated steady-state concentrations could be reduced by an adapted dosing regimen.