Carla Troisi 1, Ingrid Michon 1, Marco Siccardi 1, Sophie Fischer-Holzhausen 1, Pavel Balazki 1, Stephan Schaller 1
1 ESQlabs GmbH (Saterland, Germany)
Objectives:
Drug-drug interactions (DDIs) are a major concern in drug development and clinical practice. Mechanistic physiologically based pharmacokinetic (PBPK) modeling gains acceptance from regulatory agencies as an approach to estimating clinical DDI magnitude, designing clinical DDI trials, and potentially avoiding the need to conduct multiple DDI trials is used to predict DDIs [1,2]. However, manual execution of large perpetrator-victim matrices is time-consuming, error-prone, and difficult to reproduce. We aim to present an open-source, end-to-end workflow for conducting scalable, reproducible, and robust DDI analyses. This workflow leverages the Open Systems Pharmacology (OSP) Suite and ESQlabs R packages ecosystem to facilitate efficient simulations and analyses of complex DDI scenarios, including multiple perpetrators and victims across various dosing strategies and disease states.
Methods:
The workflow starts by developing a PBPK model for the index compound using PK-Sim [3] and MoBi [4], incorporating compound-specific properties and pre-implemented whole-body PBPK models across multiple species and disease states. Perpetrators and victims are added either by:
– loading OSP community, qualified templates [5] for strong/moderate/weak CYP/UGT/P-gp substrates and inhibitors or, when library coverage is insufficient,
– adopting published, validated (but not yet formally qualified) templates [6], or
– building bespoke compound models when necessary.
While all simulations can be executed directly within PK-Sim or MoBi, large perpetrator-victim scenario matrices rapidly become impractical to manage manually. To enable scalable and reproducible execution, the workflow uses OSP’s R interface [7] together with the {esqlabsR} package [8], supported by the graphical user interface ESQapp [9] to: (i) parameterize study arms from pre‑filled, version‑controlled configuration tables; (ii) programmatically generate and execute simulation batches; (iii) extract outputs as tidy data frames; and (iv) compute PK parameters and DDI ratios (AUC, Cmax, AUCR, CmaxR), run basic statistics, and auto‑generate tables and publication‑ready figures ({ggplot2} [10] / {esqlabsR}).
Results:
The workflow demonstrates substantial improvements in scalability, reproducibility, and automation for DDI analysis. It allows the efficient execution of large DDI matrices involving multiple perpetrators and victims, supports various populations and dosing strategies, and minimizes the need for manual setup. The platform generates accurate pharmacokinetic parameters, including (but not limited to) AUC (AUC0-inf, AUCtau, and other customized AUC) and Cmax values, and computes DDI ratios such as AUCR and CmaxR with minimal user input. A complete, customizable R script is provided to facilitate automated analysis and generate customized metrics and graphs for DDI evaluation. The entire system is 100% free, open-source, and easily maintainable, allowing researchers to seamlessly integrate and extend it for their own specific DDI needs.
Conclusions:
This workflow provides the pharmacometric community with a flexible, efficient, and reproducible solution for conducting DDI studies. By automating the execution and analysis of large simulation grids, the workflow accelerates the translation of mechanistic hypotheses into decision-ready results. It supports diverse DDI analyses across different populations, dosing regimens, and therapeutic areas, offering a significant contribution to the field. The ready-to-use R script and user-friendly interface make it accessible for researchers at all levels, ensuring a high-impact, open-source solution for DDI modeling and simulation.
References:
[1] ICH M12 guidelines, available at ICH_M12_Step4_Guideline_2024_0521.pdf
[2] Li, Yangkexin, Henry Sun, and Zuoli Zhang. 2025. “The Evolution and Future Directions of PBPK Modeling in FDA Regulatory Review” Pharmaceutics 17, no. 11: 1413. https://doi.org/10.3390/pharmaceutics17111413
[3] PK-Sim, https://github.com/Open-Systems-Pharmacology/PK-Sim
[4] MoBi, https://github.com/Open-Systems-Pharmacology/MoBi
[5] Model library of qualified models, https://github.com/Open-Systems-Pharmacology/OSP-PBPK-Model-Library
[6] Model library validated templates, Open Systems Pharmacology section “Find a repository”
[7] OSPSuite‑R; Sevestre M, Balazki P, Solodenko J, Patil I, MIL F (2026). ospsuite: R package to manipulate OSPSuite Models. R package version 12.4.1, https://github.com/open-systems-pharmacology/ospsuite-r.
[8] esqlabsR package, Balazki P, Eitel J, Patil I, Vavilov S (2023). esqlabsR: esqLABS utilities package. R package version 5.0.0, https://github.com/esqLABS/esqlabsR.
[9] ESQapp, Mil F, Kostiv A (2026). ESQapp: Shiny App to edit esqLABSR Edit Simulation Scenarios. R package version 2.2.0.9004, https://github.com/esqLABS/ESQapp.
[10] ggplot2, Wickham H (2016). ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag New York. ISBN 978-3-319-24277-4, https://ggplot2.tidyverse.org.
Reference: PAGE 34 (2026) Abstr 12092 [www.page-meeting.org/?abstract=12092]
Poster: Methodology - Other topics