Camille Massaux1, Flora Musuamba Thsinanu1,3, Alexander Kulesza2, Jean-Michel Dogné1
1University of Namur, 2ESQlabs GmbH , 3Federal Agency for Medicines and Health Products (FAMHP)
Introduction: Physiologically based pharmacokinetic (PBPK) models are becoming increasingly important in drug development, particularly for the assessment of drug-drug interactions (DDI). A growing body of scientific literature demonstrates promising results regarding their predictive performance 1 2 3 and details their implementation in various software platforms 4 5. The European Medicines Agency (EMA) and the US Food and Drug Administration (FDA) have issued dedicated guidelines, with ICH M12 specifically recognizing PBPK as a valuable approach for DDI evaluation 6. In practice, PBPK analyses are typically conducted using commercial or open-source software platforms. This study focuses on two user-friendly PBPK software tools for DDI assessment: SIMCYP and PKSIM. The model structure and associated parameters are extensively documented for both platforms, making their commonalities and differences well understood in principle. However, the extent to which the choice of platform influences the final DDI results from a user or assessor perspective has not yet been thoroughly explored. Addressing this gap is crucial for drug developers and regulatory agencies, as it would provide insights into the potential impact on platform qualification, tool interchangeability, and the extrapolation of results from one platform to another. Objectives: The aim of this study is to assess the results and impact of implementing PBPK models in the Simcyp and PKSim software platforms for evaluating DDIs, using well-documented CYP3A4-related DDIs for oral contraceptives as case studies. Methods: Simulations were performed using Simcyp version 23, release 2, and PKSim version 11. The PBPK model implementation in both platforms consists of three key components: the drug model (based on the drug’s physicochemical properties), the systems model (reflecting the anatomical and physiological characteristics of the human body), and the population model (capturing variability in the systems model across different subgroups and target populations). First, the implementation of these three components is described in detail, highlighting commonalities and differences between the two platforms. Simulations of the same DDI assessments are then conducted in both software platforms, with numerical and graphical comparisons of the results. Literature was consulted to extract physicochemical parameters and pharmacokinetic properties for the oral contraceptive victim drugs, as well as for well-documented DDI perpetrators. The target population consisted of healthy Caucasian females aged 18 to 45 years, with 100 individuals (10 groups of 10 subjects) simulated in each scenario. Results: A significant number of differences were observed between the two software platforms. While the systems model structures are conceptually similar, there are notable differences in their implementation. For the whole body PBPK model available in both softwares, several differences were identified: For instance, Simcyp offers three absorption models whereas PkSim offers only one. On the other hand, PkSim offers five different methods for calculating the partition coefficients between plasma and the various organs and tissues included in the model, while Simcyp offers only three. Even for parameters common to both platforms, such as the concentration of reference enzymes, discrepancies were noted. For example, the CYP3A4 concentration is 137 pmol/mL in Simcyp compared to 108 pmol/mL in PkSim. Differences of varying significance were also observed with other oral contraceptives, such as gestodene and ethinylestradiol, drosperinone and ethinylestradiol or drosperinone alone. The more extreme case occured when 0.150 mg levonorgestrel and 0.03 mg ethinylestradiol were administered. In this case, Simcyp predicted a maximum plasma concentration of 2.02 ng/mL, whereas PKSim predicted a maximum concentration of only 0.80 ng/mL in the peripheral venous blood compartment. Conclusion: Several differences in implementation and respective user interface impair direct reproduction of a given PBPK study across platforms. When comparing the results obtained when assessing DDI for the same drugs and as close as possible matching parametrization in Simcyp and PK-Sim software platforms still the software choice may lead to contradictory conclusions related to magnitude of DDI. In consequence, unverified reproduction cross-platform can impact regulatory decisions and may have dramatic consequence for the patients, in particular for cases when there is no clinical data to be compared to software predictions. Consequently, the qualification of each platform should be carefully and separately considered even for the same application (e.g. CYP-related DDI). Model developers should be alerted that results obtained with one software are not always to be extrapolated to the others. Finally, when a study relies on a specific software, it is essential to clearly identify the software used, its version, as well as all the parameters and options that need to be entered or selected to perform the simulations presented in the study.
1. Min JS, Bae SK. Prediction of drug-drug interaction potential using physiologically based pharmacokinetic modeling. Arch Pharm Res 40 1356-1379. (2017) 2. Lewis GJ, Ahire D, Taskar KS. Physiologically-based pharmacokinetic modeling of prominent oral contraceptive agents and applications in drug-drug interactions. CPT Pharmacometrics Syst Pharmacol 13 563-575. (2024) 3. Cicali B, Lingineni K, Cristofoletti R, et al. Quantitative Assessment of Levonorgestrel Binding Partner Interplay and Drug-Drug Interactions Using Physiologically Based Pharmacokinetic Modeling. CPT Pharmacometrics Syst Pharmacol 10 48-58. (2021) 4. Pharmacology OS. PK-Sim® Documentation. [cited 2025 March 7]Available from: https://docs.open-systems-pharmacology.org/working-with-pk-sim/pk-sim-documentation 5. Ezuruike U, Zhang M, Pansari A, et al. Guide to development of compound files for PBPK modeling in the Simcyp population-based simulator. CPT Pharmacometrics Syst Pharmacol 11 805-821. (2022) 6. EMA. Drug interaction studies M12. [cited March 7 2025]Available from: https://database.ich.org/sites/default/files/ICH_M12_Step4_Guideline_2024_0521_0.pdf
Reference: PAGE 33 (2025) Abstr 11566 [www.page-meeting.org/?abstract=11566]
Poster: Methodology - New Modelling Approaches