Development of minimal physiologically-based pharmacokinetic-pharmacodynamic models for characterizing cellular kinetics of CAR T cells following local deliveries in mice

Abstract

Chimeric antigen receptor (CAR) T cell therapies have revolutionized the treatment of hematologic malignancies and have potentials for solid tumor treatment. To overcome limited CAR T cell infiltration to solid tumors, local delivery of CAR T cells is a practical strategy that has shown promising therapeutic outcome and safety profile in the clinic. It is of great interest to understand the impact of dosing routes on CAR T cell distribution, subsequent proliferation and tumor killing in a quantitative manner to identify key factors that contribute to CAR T efficacy and safety. In this study, we established mouse minimal physiologically-based pharmacokinetic (mPBPK) models combined with pharmacodynamic (PD) components to delineate CAR T cell distribution, proliferation, tumor growth, and tumor cell killing in the cases of pleural and liver tumors. The pleural tumor model reasonably captured published CAR T cellular kinetic and tumor growth profiles in mice. The mPBPK-PD simulation of a liver tumor mouse model showed a substantial increase in initial tumor infiltration and earlier CAR T cell proliferation with local hepatic artery delivery compared to portal vein and intravenous (i.v.) injections whereas portal vein injection showed little difference from i.v. administration, suggesting the importance of having the injection site close to tumor for maximal effect of non-systemic administration. Blood flow rate in the liver tumor was found to be a sensitive parameter for cellular kinetics and efficacy, indicating a potential role of tumor vascularization in the efficacy of CAR T cell therapies.

Utility of Minimal Physiologically Based Pharmacokinetic Models for Assessing Fractional Distribution, Oral Absorption, and Series-Compartment Models of Hepatic Clearance

“…The quantitative correlation of the SCM and DM was not elucidated until our recent work (Li and Jusko, 2023), where we demonstrated how the number of liver sub-compartments (n) in the SCM correlated with the dispersion number (D N ) of the DM that describes the relative spread of solute on passage through the liver. We also showed that the SCM closely resembles the DM 9 (CAR) T cells (Tsai et al, 2022) and the PK of antenatal betamethasone and dexamethasone in parturient women and their fetuses (Krzyzanski et al, 2023).…”

Best Practices and Considerations for Clinical Pharmacology and Pharmacometric Aspects for Optimal Development of CAR‐T and TCR‐T Cell Therapies: An Industry Perspective

“…68 Interestingly, Tsai et al recently further adopted some features of the full PBPK/PD models to develop a minimal PBPK/PD model and characterized CAR-T distribution, proliferation, and tumor dynamics after local delivery of CAR-Ts for pleural and liver tumors in mouse models. 101 Although limitation exists due to key model assumptions and data requirement, such a modeling framework can further support the development of local delivery of CAR-Ts for solid tumors. Informed by the minimal PBPK/PD model, the authors suggested local delivery to be more effective for CAR-Ts in solid tumors as compared with systemic delivery.…”

Model‐informed drug development of autologous CAR‐T cell therapy: Strategies to optimize CAR‐T cell exposure leveraging cell kinetic/dynamic modeling

“…In addition to providing information on the in vivo expansion, persistence, efficacy, and safety, both immune‐compromised and immune‐competent mice may allow one to study the distribution and accumulation of the CAR‐T cells in different organs. This information may be used to advance previously developed in vitro ‐ based cell‐level models into semi‐mechanistic 65 or PBPK/PD models 30,59 …”

“…46,57,58 Considering the manifold possibilities to influence the CAR-T cell phenotype (and with this the post-infusion expansion capacity) during CAR-T cell manufacturing, an abundance of options to combine design decisions regarding T cell activation, stimulation, expansion duration, and expansion conditions exist. Within an MIDD program, PBPK/quantitative systems pharmacology (QSP), 30,59 population PK/PD, 1,43,44,60 and emerging machine learning 22 approaches can be applied to incorporate the quantitative findings obtained in carefully chosen in vitro experiments investigating the impact of such design choices. The generated models can then be leveraged to perform hypothesis-generating simulations regarding the interactions of the actionable variables in the manufacturing phase, followed by model-informed design of the hypothesis-testing experiments.…”

“…This information may be used to advance previously developed in vitro-based cell-level models into semi-mechanistic 65 or PBPK/PD models. 30,59 Clinical development…”