Lisa Hanquet 1, Happy Phanio Djokoto 1, Jean-Michel Dogné 1,2, Grace Shalom Govere 1,2, Hélène Haguet 1, Camille Massaux 1, Adrien Olama 1, Lisa Wellin 1, Flora Musuamba Tshinanu 1,2
1 Clinical Pharmacology and Toxicology Research Unit (URPC), Namur Research Institute for Life Sciences (NARILIS), University of Namur (Namur, Belgium), 2 Federal Agency for Medicines and Health Products (FAMHP) (Bruxelles, Belgium)
Objectives
CAR T cell kinetic models are increasingly enriched with biologically structural components that aim to represent processes such as differentiation, functional changes, and tissue distribution [1, 2]. Yet, the clinical datasets available for parameter estimation are often sparse and limited to a few peripheral readouts. In particular, long term persistence patterns and phenotypic evolution are rarely captured with sufficient resolution, limiting the ability to identify which biological determinants provide meaningful explanatory power [3]. Our objective was to systematically review biological determinants proposed to drive CAR T persistence, evaluate how they have been encoded in published models, and assess whether typical adult clinical datasets are sufficient to estimate the corresponding parameters in a robust way.
Methods
We first identified potential biological determinants of CAR T persistence using CAR T–specific and broader T cell biology literature. We then reviewed published model based analyses of CAR T therapies to determine, for each determinant, whether it had been represented structurally (e.g. as compartments, kinetic parameters, covariates, or initial conditions) and which clinical data supported its estimation. Data types considered included peripheral blood transgene or qPCR measurements, flow cytometry, immunophenotyping, cytokine concentrations, and, when available, tissue or bone marrow assessments. Finally, we compared how the structural complexity of the models aligned with the actual richness and resolution of the available data, identifying potential sources of non identifiability.
Results
We identified 11 candidate biological determinants of CAR T persistence, of which 6 had been explicitly incorporated into structural model components or covariate relationships. Across all studies, peripheral blood transgene or qPCR measurements formed the core dataset and consistently supported description of overall expansion–contraction patterns and proliferation related parameters [1]. When reported, tumour burden and clinical covariates such as prior treatment lines or CD4/CD8 composition were included as covariates, but did not by themselves resolve internal cellular heterogeneity [2, 4].
Models that attempted to represent differentiation cascades (e.g. TSCM, TCM, TEM, TEFF), exhaustion states, cytokine mediated survival, or manufacturing related effects introduced additional latent compartments and subset specific kinetic parameters [3, 4]. However, longitudinal immunophenotyping and cytokine profiles were often missing or measured sparsely, so that many of these parameters could not be uniquely identified from peripheral blood data alone[2, 5]. In practice, several biologically motivated extensions therefore relied on indirect inference rather than direct measurement support. Anatomical extensions, such as bone marrow as a sustained antigen source, further increased structural complexity but relied heavily on paediatric datasets or summary level parameters, limiting generalisability to adult practice and reinforcing dependence on non verifiable structural assumptions [2, 6].
Conclusions
Current adult clinical datasets, largely restricted to peripheral blood transgene measurements with limited longitudinal phenotyping, support only relatively simple CAR T persistence models for which most parameters are directly informed by observable data. More complex structures that encode internal stratification, latent functional states, or additional anatomical compartments are biologically appealing but are prone to non identifiability and reduced interpretability when not matched by equally granular longitudinal measurements. Strengthening empirical data collection—particularly systematic adult immunophenotyping and bone marrow or tissue level assessments—will be essential to justify and reliably estimate richer structural models of CAR T persistence.
References:
[1] Stein AM, Grupp SA, Levine JE, et al. Tisagenlecleucel model-based cellular kinetic analysis of chimeric antigen receptor–T cells. CPT: Pharmacometrics Syst Pharmacol 2019; 8: 285–295.
[2] Salem AM, Mugundu GM, Singh AP. Development of a multiscale mechanistic modeling framework integrating differential cellular kinetics of CAR T-cell subsets and immunophenotypes in cancer patients. CPT: Pharmacometrics & Systems Pharmacology 2023; 12: 1285–1304.
[3] Wolyncewicz G, Wayte R, Abadir E. Current insights of post-infusion CAR T expansion and persistence for large B-cell lymphoma. Cancers 2025; 17: 3167.
[4] Tao Z, Chyra Z, Kotulová J, et al. Impact of T cell characteristics on CAR-T cell therapy in hematological malignancies. Blood Cancer J 2024; 14: 213.
[5] Mueller-Schoell A, Puebla-Osorio N, Michelet R, et al. Early Survival Prediction Framework in CD19-Specific CAR-T Cell Immunotherapy Using a Quantitative Systems Pharmacology Model. Cancers 2021; 13: 2782.
[6] Martínez-Rubio Á, Chulián S, Blázquez Goñi C, et al. A Mathematical Description of the Bone Marrow Dynamics during CAR T-Cell Therapy in B-Cell Childhood Acute Lymphoblastic Leukemia. Int J Mol Sci; 22. Epub ahead of print 14 June 2021. DOI: 10.3390/ijms22126371.
Reference: PAGE 34 (2026) Abstr 12132 [www.page-meeting.org/?abstract=12132]
Poster: Drug/Disease Modelling - Oncology