Physiologically based pharmacokinetic modeling – application for renal and hepatic impairment
Annika Schneider(1,2,3), Katrin Coböken(3), Sebastian Frechen(3), Jan-Frederik Schlender(3), Michael Block(3)
(1)Ruprecht-Karls Universität Heidelberg, Germany (2)Rheinisch-Westfälische Technische Hochschule Aachen, Germany (3)Systems Pharmacology & Medicine, Bayer AG, Leverkusen, Germany
Objectives: Hepatic and renal impairment can have an impact on pharmacokinetics in patients compared to individuals with normal hepatic or renal status. Currently, the regulatory acceptance of model-based assessments for such populations is still in discussion. The objective of the presented approach was to evaluate the potential and critical gaps of physiologically based pharmacokinetic (PBPK) modeling and its predictive power for an assessment of these disease states. Moreover, it was aimed to evaluate potential improvements. By use of adequately selected paradigm compounds, a path forward for this assessment was developed and is presented here.
Methods: In a stepwise approach, the current state of knowledge on physiological changes related to hepatic and renal impairment were compiled and integrated into a PBPK framework. Physiological changes considered in the PBPK framework were informed for blood flows, plasma protein concentrations, hematocrit, liver enzyme activities and glomerular filtration rate (GFR). By this, in-silico renal (mild/ moderate/ severe) and hepatic impairment (Child-Pugh A/B/C) populations were established and population simulations with several paradigm compounds were performed to test the compiled information for both diseases. All simulations were performed using the Open Systems Pharmacology Suite [1] and simulation results in the form of concentration-time profiles, maximum plasma concentration (Cmax) and area under the plasma concentration-time curve (AUC) values were compared to respective clinical data obtained from literature [2-8]. As a qualification criterion for concentration-time profiles, a two-fold prediction range around the observed values was defined. Cmax and AUC values were evaluated using the geometric mean fold error (GMFE).
For hepatic impairment midazolam and alfentanil were chosen as paradigm compounds due to their nearly exclusive metabolism by CYP3A4. Amikacin was selected as a test case for renal impairment because of its nearly solely renal excretion. Midazolam pharmacokinetics was also examined for renal impairment to evaluate the influence of renal impairment on CYP3A4. In a next step the approach was applied to lidocaine and two subsequent metabolites to assess a more complex elimination scheme. For this purpose, a PBPK model for lidocaine and its metabolites was built.
Results: The PBPK approach was able to predict the mean pharmacokinetics of alfentanil, amikacin, and midazolam under renal and hepatic impairment well. 88% of all plasma concentration values were within a two-fold prediction range around the observed values. Cmax and AUC values were predicted with a GMFE of 1.698 and 1.351, respectively. The application to lidocaine and its metabolites also showed good agreement of available mean clinical data to the mean predictions with 74% of plasma concentration predictions lying within the two-fold prediction range. The GMFE of the Cmax and AUC values were 1.68 and 1.515, respectively. Nevertheless, the approach struggled with the prediction of the right variability within and between populations, especially for hepatic impairment populations. Further analysis of this limitation identified different underlying reasons, one of them being the limited translatability of the classification rules into physiological parameters, as well as the pathophysiological heterogeneity of populations with the same classification score.
Conclusions: The hereby presented approach was able to cover the rationale to predict the pharmacokinetics in renally and hepatically impaired populations based on healthy individuals. The current challenges towards a direct relation of categorization by Child-Pugh classes or apparent GFR values to physiological parameters were discussed, revealing key elements for further improvement. Additionally, a PBPK model for lidocaine and its metabolites was built which was well able to predict pharmacokinetics of these compounds in both patient populations.
References:
[1] http://www.open-systems-pharmacology.org/
[2] Orlando R et al. Br J Clin Pharmacol. (2003) 55, 86-93.
[3] Orlando R et al. Clin Pharmacol Ther. (2004) 75, 80-88.
[4] de Martin S et al. Clin Pharmacol Ther. (2006) 80, 597-606.
[5] Ferrier M et al. Anesthesiology (1985) 62, 480-484.
[6] Blair D et al. Antimicrob Agents Chemother. (1982) 22, 376-379.
[7] Thomson B et al. Am J Kidney Dis. (2015) 65, 574-582.
[8] Albarmawi A et al. Br J Clin Pharmacol. (2014) 77, 160-169.