J.A.A. Reijers (1), T. van Donge (1), F.M.L. Schepers (1), J. Burggraaf (1), J. Stevens (1)
(1) Centre for Human Drug Research (CHDR), Leiden, The Netherlands
Objectives: To study pharmacokinetic biosimilarity using full concentration-time profiles from a population pharmacokinetic analysis of 2 biotherapeutics (trastuzumab) and compare the results to a standard non-compartmental analysis (NCA) that uses the observed concentration-time points.
Methods: The pharmacokinetic data were obtained from a biosimilarity trial (n=110) of trastuzumab in which intravenous doses of 0.49, 1.48, 2.96, 5.96 mg/kg test product and 6.44 mg/kg of reference product were administered to healthy volunteers[1]. Using NONMEM 7.2[2], a combined PK model was developed for the test and reference product, as were two PK models on test and reference product separately. To allow comparison between all models, structural similarity was pursued. Using the resulting individual concentration-time profiles, AUCs were calculated using linear trapezoidal rule. Then, using the observed concentration-time data, standard NCA methods were used to calculate the individual AUCs, which were compared to the model results.
Results: The PK of trastuzumab was best described using a three compartment model with combined linear- and Michaelis-Menten elimination. Interindividual variability could be identified on three PK parameters. Residual variability was best described by a combined proportional and additive error model. Lean body weight was identified as covariate on central volume of distribution and body mass index on the elimination rate constant. Trastuzumab product could not be identified as a significant covariate on any parameter. The geometric mean ratio (GMR, 95% confidence interval) of the AUCinf of the test/reference product for combined model was 81.66% (77.93-85.56%). For the separate models, GMR was 82.54% (78.70-86.57%). Following NCA, the GMR of the AUCinf was 82.32% (78.17-86.69%). After standard linear dose correction for the labelled dose of 6 mg/kg using separate model results, GMR for the AUCinf was 91.74% (87.46-96.24%) for population pharmacokinetic analysis and 89.55% (85.03-94.30%) for the NCA.
Conclusions: The obtained AUCs of the developed PPK models were comparable to the NCA results, regardless whether test and reference product were modelled in one combined or two separate models. Previously, the interchangeability of PPK models in biosimilarity research has been proven for therapeutic proteins[3]. We show that this also applies for the relatively large monoclonal antibody trastuzumab which displays non-linear pharmacokinetics.
References:
[1] Wisman LA, De Cock EP, Reijers JA, Kamerling IM, Van Os SH, de Kam ML, Burggraaf J, Voortman G. A phase I dose-escalation and bioequivalence study of a trastuzumab biosimilar in healthy volunteers. Clinical Drug Investigation (2014) 34:887-894.
[2] Beal SL, Sheiner LB, Boeckmann AJ & Bauer RJ (Eds.) NONMEM Users Guides. 1989-2011. Icon Development Solutions, Ellicott City, Maryland, USA.
[3] Dubois A, Gsteiger S, Balser S, Pigeolet E, Steimer JL, Pillai G, Menté F. Pharmacokinetic similarity of biologics: Analysis using nonlinear mixed-effects modeling. Clinical Pharmacology &Therapeutics (2012) 91(2): 234-242.
Reference: PAGE 25 () Abstr 5871 [www.page-meeting.org/?abstract=5871]
Poster: Methodology - Other topics