Carmine Schiavone, Stephan Schaller 1, Wilbert de Witte 1, Jorin Diemer 1, Alexander Kulesza 1,2
1 ESQlabs GmbH (Saterland, ), 2 Université Namur (Namur, Italia)
Bispecific T-cell engagers (TCEs) offer potent therapeutic benefits, yet their clinical translation is frequently hindered by non-linear pharmacokinetics and severe immune-related toxicities, such as cytokine release syndrome (CRS). Concurrently, there is a growing regulatory push toward new approach methodologies (NAMs) that minimize animal testing, creating a critical need for mechanistic, human-relevant modeling workflows to predict safety and efficacy. While physiologically based pharmacokinetic (PBPK) models for TCEs have been developed, highly reusable and scalable whole-body platforms remain scarce.
To bridge this gap, we constructed an open-source, modular PBPK–quantitative systems pharmacology (QSP) ecosystem within the PK-Sim/MoBi platform. Designed for maximum reusability, this framework separates a standardized PBPK chassis—accounting for blood-flow limited distribution, vascular permeability, lymphatic transport, and FcRn-mediated recycling—from customizable pharmacodynamic components.
The modularity of this framework allows it to be truly “plug-and-play”; researchers can seamlessly add a new piece to introduce a novel biological effect without having to rebuild the underlying model. Our current ecosystem utilizes three primary, interchangeable mechanistic blocks:
Trimer Formation: Calculates the kinetic formation of the Target-Antibody-CD3 complex to mechanistically capture the drug’s primary mode of action.
Cytokine Response: Translates the trimer signal into a biological response, driving cytokine production while incorporating first-order decay and inhibitory feedback loops.
Cell Depletion: Employs a transit-compartment structure to dynamically simulate the drug’s impact on the proliferation rate and subsequent depletion of circulating target cells.
To further enhance predictive accuracy, we have recently expanded the Cell Depletion module to directly integrate extrapolated laboratory findings. MoBi’s modular architecture and R scripting interface enables a consistent integration of in vitro cytotoxicity and cytokine release data (e.g., from hPBMC assays) within the entire translational WB-PBPK-based TCE modeling framework. By explicitly recapitulating experimental conditions (such as target expression levels, effector and target cell concentrations, E:T ratio, compartment volume, and incubation time), the platform anchors drug concentration while mapping observed dose–response relationships to mechanistic drivers of activity, notably synapse/trimer formation. Under these controlled conditions, effect parameters (e.g., EC50 expressed in units of trimers per target cell) are estimated by fitting to in vitro data while preserving their in vivo mechanistic interpretation and exact implementation. Importantly, this extension enables quantitative extrapolation of the characteristic bell-shaped in vitro dose–response curve observed for TCEs, arising from concentration-dependent shifts between productive trimer formation and non-productive binary complexes [11]. This strategy constitutes a structurally informed in vitro–in vivo extrapolation (IVIVE), enabling the identification of in vivo-relevant potency parameters directly from in vitro experiments and thereby strengthening translational predictivity.
We validated the scalability of this end-to-end framework across three diverse drugs and multiple species. For blinatumomab (CD19×CD3), the model captured plasma PK together with IL-6 dynamics and peripheral B-cell depletion in a chimpanzee dataset across 0.06, 0.10, and 0.12 μmu μg/kg dose levels. For epcoritamab (CD3×CD20), the framework reproduced cynomolgus monkey PK/PD behavior across regimens, including strong B-cell depletion at 0.1 and 1 mg/kg repeated IV dosing, while linking exposure differences to cytokine profiles. Crucially, for teclistamab (CD3×BCMA), simulations matched expected human PK patterns across IV and SC regimens (0.27 to 1.5 mg/kg) and accurately reproduced IL-6 trajectories under step-up dosing scenarios (MajesTEC-1). These simulations were consistent with reported rapid peripheral B-cell depletion during clinical therapy, highlighting direct relevance to human translation and regimen design.
Overall, these satisfactory cross-species fits indicate that the modular PBPK-QSP workflow can be “personalized” at the level of drug- and regimen-specific inputs while reusing the same mechanistic blocks. By decoupling system biology (e.g., IL-6 dynamics, cell depletion) from drug-specific parameters, this validated approach minimizes the need for re-implementation, reduces reliance on animal testing (NAM-aligned), and accelerates translational safety assessment. Ultimately, this plug-and-play versatility supports a streamlined workflow where novel bispecific constructs and complex dosing regimens can be rapidly evaluated within a unified, high-quality computational environment.
References:
FDA. Roadmap to Reducing Animal Testing in Preclinical Safety Studies. 10 Apr 2025. https://www.fda.gov/files/newsroom/published/roadmap_to_reducing_animal_testing_in_preclinical_safety_studies.pdf
Schropp J, Khot A, Shah DK, Koch G. Mechanistic PBPK modeling for biologics. CPT Pharmacometrics Syst Pharmacol. 2019;8:177–187. https://doi.org/10.1002/psp4.12369
Lippert J, Burghaus R, et al. Open Systems Pharmacology Community. CPT Pharmacometrics Syst Pharmacol. 2019;8:878–882. https://doi.org/10.1002/psp4.12473
Niederalt C, Kuepfer L, et al. Generic whole‑body PBPK model for therapeutic proteins in PK‑Sim. J Pharmacokinet Pharmacodyn. 2018;45:235–257. https://doi.org/10.1007/s10928-017-9559-4
Chen X, Kamperschroer C, Wong G, Xuan D. Modeling cytokine release for T cell‑engaging bispecific antibodies. Clin Transl Sci. 2019;12:600–608. https://doi.org/10.1111/cts.12662
Schlereth B, et al. CD19×CD3 in chimpanzees: T‑cell activation and B‑cell depletion. Cancer Immunol Immunother. 2006;55:503–514. https://doi.org/10.1007/s00262-005-0001-1
Engelberts PJ, et al. DuoBody‑CD3xCD20: preclinical PK/cytokines/depletion; supports SC dosing. EBioMedicine. 2020;52:102625. https://doi.org/10.1016/j.ebiom.2019.102625
Willemin M‑E, Wang Lin SX, De Zwart L, et al. Evaluating drug interaction potential from cytokine release syndrome using a PBPK model: teclistamab case study. CPT Pharmacometrics Syst Pharmacol. 2024. https://doi.org/10.1002/psp4.13144
Girgis S, et al. Translational modeling predicts efficacious dosing range of teclistamab. Target Oncol. 2022;17:433–439. https://doi.org/10.1007/s11523-022-00893-y
Frerichs KA, et al. Teclistamab impairs humoral immunity; peripheral B‑cell depletion; IVIG relevance. Blood Adv. 2024;8:194–206. https://doi.org/10.1182/bloodadvances.2023011658
Douglass EF, Miller CJ, Sparer G, Shapiro H, Spiegel DA. A Comprehensive Mathematical Model for Three-Body Binding Equilibria. J Am Chem Soc. 2013;135(16):6092–99. https://doi.org/10.1021/ja311795d
Reference: PAGE 34 (2026) Abstr 12159 [www.page-meeting.org/?abstract=12159]
Poster: Drug/Disease Modelling - Oncology