Rebekka Fendt 1,2, Annika Schneider 1,2, Jan-Frederik Schlender 2, Ute Hofmann 3,4, Elke Schäffeler 3,4, Reinhold Kerb 3, Matthias Schwab 3,5, Lars Kuepfer 1,2
(1) Institute of Applied Microbiology, RWTH Aachen University, Aachen, Germany (2) Systems Pharmacology & Medicine, Bayer AG, Leverkusen, Germany (3) Dr. Margarete Fischer-Bosch-Institute of Clinical Pharmacology, Stuttgart, Germany (4) University of Tuebingen, Tuebingen, Germany (5) Depts. of Clinical Pharmacology and Biochemistry and Pharmacy, University of Tuebingen, Tuebingen, Germany
Objectives:
In this work, individualized physiologically-based pharmacokinetic (PBPK) models were built based on information from a clinical study in healthy volunteers. The mechanistic nature of PBPK models allows incorporating knowledge about the studied healthy volunteers such as biometrics (height, weight, age, sex) or physiological data (hematocrit, glomerular filtration rate, liver blood flow). The goal of the study was to evaluate whether incorporation of this individual information improves the agreement of the volunteer-specific PK simulations with corresponding PK measurements.
Methods:
PK data were obtained recently in a drug cocktail trial [1]. In this study, six drugs (caffeine, codeine, midazolam, pravastatin, talinolol and torsemide) were given simultaneously to healthy individuals (n=106). In addition to the plasma PK data, the data set included information on biometrics, physiology, pharmacogenetics and lifestyle of the study volunteers. A previously established midazolam PBPK model was employed to simulate individual PBPK profiles. All simulations were performed with PK-Sim® as part of the Open Systems Pharmacology Suite [2] and MATLAB.
Results:
PK data of midazolam, a probe drug for CYP3A4 activity, were analysed in this study. As a starting point, the midazolam PBPK model was fitted to the mean concentration of all volunteers. The model was simulated with the biometrics of the PK-Sim® reference individual (caucasian, male, 30 years, 73 kg, 1.76 m).
The PBPK model of midazolam was able to describe the time course of the mean data very well, including the maximum concentration (Cmax) and the time of the maximum concentration (tmax). All observed data were within the 2-fold range of predicted concentrations and the concordance correlation coefficient was 0.99. The maximum predicted versus observed ratio was 1.3.
To systematically assess the benefit mediated by the additional consideration of volunteer-specific information, the original PBPK model of the reference individual was individualized with respect to height, weight, age and sex.
In step 1, PK data of each volunteer were compared to the base PBPK simulation described above. The concordance correlation coefficient was 0.75 and 80.1% of the observed data were within the 2-fold range of predicted concentrations.
In step 2, the biometrics (height, weight, age, sex) of the virtual volunteers were set to the reported biometrics of the study volunteers and compared to the respective PK data. The results of the individualized PBPK models were similar to the simulations with the reference individual in step 1. The concordance correlation coefficient was 0.74 and 76.1% of the observed data were within the 2-fold range of predicted concentrations.
In step 3, CYP3A4 expression was fitted individually to PK data of each volunteer using the reference individual. The concordance correlation coefficient was 0.88 and 97.0% of the observed data were within the 2-fold range of predicted concentrations.
In step 4, CYP3A4 expression was fitted to the volunteer-specific PK profiles using the biometrics of the respective individual. These results were again similar to the results that were obtained for PBPK simulations with a reference individual. The concordance correlation coefficient was 0.87 and 97.2% of the observed data were within the 2-fold range of predicted concentrations.
Conclusions:
Inclusion of biometric information (height, weight, age, sex) of study individuals did not improve the predictive performance of midazolam PBPK model simulations. The individual PK data are already well described by a simulation with a reference individual (80% of the data being in the 2-fold range). Similar results were obtained by applying the same approach to a caffeine PBPK model. As a next step, physiological information about the glomerular filtration rate, albumin concentration, hematocrit and liver blood flow rates will be fed into the model. The presented approach contributes to a profound understanding of PK variability which is crucial for optimization of efficacy and safety in clinical science.
References:
[1] Kuepfer, L., R. Kerb, et al. (2014). “Clinical Translation in the Virtual Liver Network.” CPT Pharmacometrics Syst. Pharmacol. 3(7): e127.
[2] http://www.open-systems-pharmacology.org/
Reference: PAGE 28 (2019) Abstr 9102 [www.page-meeting.org/?abstract=9102]
Poster: Drug/Disease Modelling - Absorption & PBPK