Alina Volkova 1,2, Yuri Kosinsky 1, Ricard Masia 3, Yu Huang 3, Benjamin Lee 3, Kirill Peskov 1,2,4
1 M&S Decisions (Dubai, United Arab Emirates), 2 Marchuk Institute of Numerical Mathematics of the Russian Academy of Sciences (INM RAS) (Moscow, Russia), 3 Translational Medicine, Crossbow Therapeutics Inc. (Cambridge, USA), 4 Research Center of Model-Informed Drug Development, Sechenov First Moscow State Medical University (Moscow, Russia)
Objectives: T-cell receptor mimetic (TCRm) T-cell engagers (TCEs) that recognize peptide-Human Leukocyte Antigen (pHLA) complexes represent an emerging class of immunotherapies. Recently, a potent and specific TCRm TCE targeting a telomerase reverse transcriptase (TERT) pHLA complex on the surface of tumor cells in acute myeloid leukemia (AML) has been reported [1]. First in human (FIH) dose selection for this compound may be challenging, particularly due to potentially low and heterogeneous surface density of TERT pHLA complexes on AML blasts and healthy myeloid cells. The objective of this work was to develop an integrated quantitative systems pharmacology (QSP) model that mechanistically links data of in vitro and in vivo preclinical experiments with clinical outcomes to support safe FIH dose selection for a TERT-directed TCRm TCE in patients with AML.
Methods: The human AML QSP model integrated pharmacokinetics (PK) of TERT-directed TCRm TCE, immunological synapse (IS) formation in bone marrow (BM), followed by target cells lysis, and Interleukin-6 (IL-6) elevation in blood plasma. In parallel, the CD123×CD3 TCE vibecotamab was included as a clinical benchmark. Drug disposition was described using a minimal physiologically based PK model incorporating target-mediated drug disposition. IS formation was driven by two mechanistic processes: T-cell-target cell encounter and TCE-mediated cell-cell adhesion. The BM compartment explicitly represented both leukemic blasts and normal myeloid cells expressing the relevant targets, enabling competition for drug binding and IS formation. Model parameters governing IS formation, cytotoxicity and interferon-γ release by activated T-cells were calibrated using in vitro data for both TCEs. IL-6 elevation after TCE dosing was modeled as a function of IS formation with a negative feedback regulation term to reproduce attenuation of cytokine release upon repeated dosing. The model parameters linking IS dynamics and IL-6 release were calibrated using published cytokine measurements in AML patients treated with vibecotamab across multiple dosing regimens [2]. A virtual AML population was generated by varying disease burden and immune status (baseline target and T-cell numbers in BM, target expression levels) together with interindividual variability in PK parameters, enabling joint prediction of cytotoxic efficacy and IL-6 dynamics across patients.
Results: First, we established a generalized in vitro T-cell-dependent cellular cytotoxicity (TDCC) framework to quantitatively characterize IS formation and cytotoxicity for the TERT directed TCRm TCE and vibecotamab across concurrent experimental settings, ensuring a shared description of system specific determinants of cytotoxicity. This unified framework reduced confounding by system level variability and enabled more robust identification and quantitative comparison of drug specific cytotoxicity parameters for both agents. Building on this, the calibrated synapse level model was used to describe TCE-driven IS formation in the human BM, enabling prediction of exposure-efficacy and exposure-safety relationships for the TERT directed TCRm TCE across multiple dosing regimens in AML patients.
Virtual clinical trial simulations identified both pharmacologically active and safe starting doses by predicting the fraction of patients achieving a clinical response while maintaining ≥95% of patients below an IL-6 peak threshold of 600 pg/mL, a level associated with clinically meaningful risk of cytokine release syndrome (CRS) in TCE therapies [3]. Explicit representation of leukemic and normal target-expressing cells in BM enabled simultaneous prediction of anti-leukemic activity and on-target toxicity effect, thereby quantifying the efficacy-safety trade-off across dosing regimens.
Conclusions: This integrated QSP framework provides a quantitative basis for FIH dose and schedule selection of a TERT-directed TCRm TCE in AML by jointly characterizing exposure-efficacy and exposure-cytokine-release relationships in humans. By combining preclinical data with clinical benchmarking and modeling of CRS risk, the approach supports the selection of tolerable, pharmacologically active starting doses, thereby reducing early-phase development risk and supporting model-informed First-in-Human dose selection.
References:
[1] Huang Y, Ban B, Jimenez J, et al.; Discovery and preclinical characterization of novel TCR-mimetic T-cell engagers targeting TERT peptide-HLA complex for the treatment of solid and hematologic malignancies. Cancer Res 2025; 85(8_Supplement_1):3507. doi: 10.1158/1538-7445.AM2025-3507.
[2] Ravandi F, Bashey A, Foran J, et al.; Phase 1 study of vibecotamab identifies an optimized dose for treatment of relapsed/refractory acute myeloid leukemia. Blood Advances 2023; 7(21). doi: 10.1182/bloodadvances.2023010956.
[3] Frances N, Bacac M, Bray-French K, et al.; Novel in Vivo and in Vitro Pharmacokinetic/Pharmacodynamic-Based Human Starting Dose Selection for Glofitamab. J Pharm Sci. 2022;111(4):1208-1218. doi: 10.1016/j.xphs.2021.12.019.
Reference: PAGE 34 (2026) Abstr 12254 [www.page-meeting.org/?abstract=12254]
Poster: Drug/Disease Modelling - Oncology