A PB-QSP platform model for T cell-dependent bispecific antibodies
Stephan Schaller
esqLABS GmbH
Objectives: T-cell-dependent bispecific (TDB) antibody is a promising therapeutic modality in drug development for both hematologic and solid malignancies. The TDB antibody bypasses the normal physiology of MHC-regulated T-cell activation by forming a tri-molecular immune synapse between the CD3 and target antigen, which trigger tumor cell killing by T-cell activation. Therefore, the degree of tumor cell killing by TDB is contingent on the distribution to- and binding with target at- the site of action, both of which may depend on the relative affinities of the TDB binding arms to the target vs. effector T-cells [1]. This complexity prompts the development of a quantitative platform model for TDB that captures the inter-relationship of binding affinities, biodistribution, and target engagement to assist in the discovery (e.g., design of drug-specific parameters, understanding proof of concept) and development (e.g., design of dosing regimen) of TDBs.
Methods: Here, we describe the development of a PB-QSP platform for TDBs using the open-source software OSP Suite (PK-Sim® and MoBi®, www.open-systems-pharmacology.org) [2], [3]. The platform was developed in three stages: First, the PK-Sim mouse physiology was adapted with in-house data for the corresponding test animal model. A dynamic T-cell distribution model was then developed in PK-Sim as a full PBPK model. T-cell kinetics were calibrated and validated using tissue time-concentration data from the literature on exogenously administered T-cells [4] and tissue distribution of resting T-cells [5] in mice. The T-cell kinetics model considered tissue-specific transmigration and tissue retention rates to capture the organ-specific distribution of T-cells, as well as the internalization and recycling kinetics of T-cell receptor CD3. In the second stage, the platform was expanded with full PBPK models to describe the binding kinetics of three different gD/CD3 monospecific antibodies binding only to T-cells with different affinities. The data on monospecific CD3 binding with three different binding affinities enabled the calibration of the CD3 turnover and internalization rates. In a third stage, the platform was expanded with a full PBPK model for HER2/CD3 TDBs binding to T-cells and target cells to calibrate HER2 turnover. All antibody PBPK models were calibrated and validated using in-house preclinical mouse studies of TDB distribution with varying affinities.
Results: The TDB/T-cell PB-QSP platform qualitatively and quantitatively captures the observed tissue distributions at the site of action, on-target/off-tumor, and off-target. The model adequately described the unique “push-and-pull” distribution patterns between T-cell-rich organs and tumor tissue as a function of relative binding affinities of the TDB to HER2 and CD3 [1]. The model showed that a high CD3 binding affinity reduces half-life and exposure in blood but increases gD/CD3 TDB distribution to T-cell-rich tissues. At the same time, it decreases both gD/CD3 and HER2/CD3 TDB distribution to HER2+ tumors. The addition of a HER2+ tumor also reduced plasma and lymph node distribution.
Conclusions: The TDB/T-cell PB-QSP platform can be applied to evaluate the tri-molecular immune synapse formation of TDBs with varying binding effector and target affinities in mouse xenografts. It allows convenient extension with effect models for assessing the impact of TDB binding properties on TDB tissue distribution and T-cell tumor-killing efficacy. Future extensions of this platform include the integration of the downstream pharmacodynamics and target cell killing by the TDB [6] and translating the model to other nonclinical species and humans.
References:
[1] D. Mandikian et al., “Relative Target Affinities of T-Cell-Dependent Bispecific Antibodies Determine Biodistribution in a Solid Tumor Mouse Model,” Mol. Cancer Ther., vol. 17, no. 4, pp. 776–785, 2018, doi: 10.1158/1535-7163.MCT-17-0657.
[2] S. Willmann, J. Lippert, M. Sevestre, J. Solodenko, F. Fois, and W. Schmitt, “PK-Sim®: a physiologically based pharmacokinetic ‘whole-body’ model,” Biosilico, vol. 1, pp. 121–124, 2003.
[3] J. Lippert et al., “Open Systems Pharmacology community - an open access, open source, open science approach to modeling and simulation in pharmaceutical sciences,” CPT Pharmacomet. Syst. Pharmacol., Oct. 2019, doi: 10.1002/psp4.12473.
[4] A. Khot, S. Matsueda, V. A. Thomas, R. C. Koya, and D. K. Shah, “Measurement and Quantitative Characterization of Whole-Body Pharmacokinetics of Exogenously Administered T Cells in Mice,” J. Pharmacol. Exp. Ther., vol. 368, no. 3, pp. 503–513, Mar. 2019, doi: 10.1124/jpet.118.252858.
[5] R. H. Rubin, D. Baltimore, B. K. Chen, R. A. Wilkinson, and A. J. Fischman, “In vivo tissue distribution of CD4 lymphocytes in mice determined by radioimmunoscintigraphy with an 111In-labeled anti-CD4 monoclonal antibody,” Proc. Natl. Acad. Sci. U. S. A., vol. 93, no. 15, pp. 7460–7463, Jul. 1996, doi: 10.1073/pnas.93.15.7460.
[6] I. Hosseini et al., “Mitigating the risk of cytokine release syndrome in a Phase I trial of CD20/CD3 bispecific antibody mosunetuzumab in NHL: impact of translational system modeling,” Npj Syst. Biol. Appl., vol. 6, no. 1, Art. no. 1, Aug. 2020, doi: 10.1038/s41540-020-00145-7.
