Microdialysate-corrected mid-interval model versus microdialysate-based integral model - Population pharmacokinetics of levofloxacin in peripheral tissues
A. Schaeftlein(1,2), A. Solms(2,3), M. Zeitlinger(4), W. Huisinga(2), C. Kloft(1)
(1) Department of Clinical Pharmacy and Biochemistry, Freie Universitaet Berlin, Berlin, Germany (2) Graduate Research Training Program PharMetrX: Pharmacometrics & Computational Disease Modeling, Martin-Luther-Universitaet Halle-Wittenberg, Freie Universitaet Berlin and Universitaet Potsdam, Germany (3) Computational Physiology Group, Institute of Mathematics, University of Potsdam, Potsdam, Germany (4) Department of Clinical Pharmacology, Medical University Vienna, Vienna, Austria
Objectives: Microdialysis (µD) has become the method of choice to determine unbound interstitial fluid (ISF) concentration of antiinfectives in peripheral tissues (PT) [1]. This interval sampling method requires the correction of the measured microdialysate concentrations (CµD) by the recovery rate (RR). The aim of this analysis was to compare two population PK modelling approaches with respect to descriptive and predictive performance for CµD of levofloxacin (LEV). A second objective was to investigate covariates (demographics, clinical chemistry, disease severity) on the PK of LEV into the ISF of PT.Methods: Plasma and µD concentrations in adipose and muscle ISF of 39 patients from 5 clinical trials [2-4] receiving 500 mg LEV once daily were analysed using NONMEM 7.2. by two approaches: (i) the microdialysate-corrected mid-interval (MCM) model, correcting CµD by RR prior to the data analysis and assigning the corrected CµD to the middle of the sampling interval; (ii) the microdialysate-based integral (MBI) model [5] which simultaneously analyses RR and CµD data and assigns CµD to the end of the sampling interval. Comparison was guided by plausibility and precision of parameter estimates, GOF plots and VPCs. Covariate selection was based on OFV, relevant influence on the PK and the ability to explain interindividual variability (IIV) on PK parameters.
Results: PK parameter estimates of the MBI model better agreed with published ones [6], also revealing higher precision than of the MCM model. In contrast to the MBI model the MCM model did not adequately describe the concentration-time profiles in both ISF of PT and its predictive performance was worse. Albumin being a marker for the colloid osmotic pressure significantly influenced intercompartmental CL (plasma to adipose ISF) explaining ~20% of IIV. Additionally renal function and disease severity showed an impact on CL of LEV.
Conclusions: CµD in the ISF of PT was best described and predicted by the MBI model. This approach enabled the differentiation between µD-specific processes and physiologically-based distribution of LEV. Based on this, more mechanistically-motivated models will be developed to explain the distribution of antiinfectives in ISF of PT.
References:
[1] N. Plock et al. Eur J Pharm Sci 25:1 (2005).
[2] R. Bellmann et al. Br J Clin Pharmacol 57: 563 (2003).
[3] M. Zeitlinger et al. AAC 47: 3548 (2003).
[4] M. Zeitlinger et al. Int J Antimicrob Agents 29:44 (2007).
[5] K. Tunblad et al. Pharm Res, 21:1698 (2004).
[6] GL. Drusano et al. AAC 46:586 (2002).
