Translating physiologically based Parameterization and Inter-Individual Variability into the Analysis of Population Pharmacokinetics
A. Solms (1,2), A. Schäftlein (1,3), M. Zeitlinger (4), C. Kloft (3), W. Huisinga (2)
(1) Graduate Research Training Program PharMetrX: Pharmacometrics & Computational Disease Modeling, Freie Universität Berlin and Universität Potsdam; (2) Computational Physiology Group, Institute of Mathematics, Universität Potsdam, Potsdam, Germany; (3) Department of Clinical Pharmacy and Biochemistry, Freie Universität Berlin, Berlin, Germany; (4) Department of Clinical Pharmacology, Medical University Vienna, Vienna, Austria;
Objectives: Physiologically based pharmacokinetics (PBPK) is a useful tool for predicting the PK of a drug. The physiological and anatomical parameterization allows to easily integrate patient characteristics and information about population specific variability. So far, there is no systematic approach how PBPK models can be used to analyze population PK data. Hence, the objective was to develop an approach integrating prior knowledge of the mechanism of (i) drug distribution and (ii) inter-individual variability (IIV), and estimate relevant drug- and physiology related parameters and the unexplained variability. The approach was exemplified on levofloxacin (Lev).
Methods: Lev plasma and interstitial fluid (ISF) microdialysis (µD) data was collected in [1,2,3]. PK was modeled based on a 13 compartment PBPK model with tissue distribution according to [4]. Individualization of the anatomy and physiology (e.g. tissue volumes) was done via the lean body mass scaling approach [5]. The data analysis was performed in R using an EM algorithm similar to [6] for nonlinear mixed effects modeling. Predicted PK, the target site of Lev, were compared to ISF µD measurements in adipose and muscle tissue.
Results: Drug transfer was realised by regional tissue blood flows; represented by the surrogate parameter cardiac output (Co). A random effect on Co was considered with a fixed distribution based on literature [7]. We found that tissue distribution strongly depended on Lev lipophilicity and its affinity to acidic phospholipids (AP). Lipophilicity was represented by the surrogate parameters logP, which was considered as a fixed effects parameter. Affinity to AP was represented by the surrogate parameter BP where additionally a random effect was assumed. The estimated parameter values well corresponded with published values. Model diagnostics, based on various tools, indicated excellent agreement between model predictions and plasma data. So far no established reference method is available to determine ISF kinetics; the comparison of model prediction and µD data revealed further insight in both techniques, e.g. their assumptions and limitations.
Conclusions: The new parameterization will lead to a different allocation of variabilities, offering a mechanism based interpretation. This gives a new perspective on parameter correlations, explained and unexplained variability in population PBPK modeling, as well as in the purely data based approach.
References:
[1] R. Bellmann, G. Kuchling, P. Dehghanyar, M. Zeitlinger, E. Minar, B.X. Mayer, M .Müller, C. Joukhadar, Tissue Pharmacokinetics of Levofloxacin in Human Soft Tissue Infections, Br J Clin Pharmacol 57 (5), 2003.
[2] M. Zeitlinger, P. Dehghanyar, B.X. Mayer, B.S. Schenk, U. Neckel, G. Heinz, A Georgopoulos, M. Müller, C. Joukhadar, Relevance of Soft-Tissue Penetration by Levofloxacin for Target Site Bacterial Killing in Patients with Sepsis, Antimicrobial Agents & Chemotherapy 47 (11), 2003.
[3] M. Zeitlinger, F. Traunmüller, A Abrahim, M.R. Müller, Z. Erdogan, M. Müller, C. Joukhadar, A pilot study testing whether concentrations of levofloxacin in interstitial space fluid of soft tissues may serve as a surrogate for predicting its pharmacokinetics in lung, Int J Antimicrobial Agents 29 (1), 2007.
[4] T. Rodgers, M. Rowland, Physiologically based pharmacokinetic modelling 2: predicting the tissue distribution of acids, very weak bases, neutrals and zwitterions, J Pharm Sci 95, 2006.
[5] W. Huisinga , A. Solms, S. Pilari, L. Fronton, Modeling Interindividual Variability in Physiologically Based Pharmacokinetics and Its Link to Mechanistic Covariate Modeling, CPT: Pharmacometrics & Systems Pharmacology 1, 2012.
[6] S. Walker, An EM Algorithm for Nonlinear Random Effects Models, Biometrics 52, 1996.
[7] A.A. Luisada, P.K. Bhat, V. Knighten, Changes of cardiac output caused by aging: an impedance cardiographic study, Angiology 31, 1980.