Article Text

Download PDFPDF

1301 Conditional bispecific T cell engager design and dosing optimization via mechanistic modeling
  1. Jess Wu,
  2. Michael Vilkhovoy,
  3. Sara Shum,
  4. Johara Chouitar,
  5. Patrick LeRoy,
  6. Leticia Fridman,
  7. Eilene Kwok,
  8. Glendon Wu,
  9. Matyas Ecsedi,
  10. Antara Banerjee and
  11. Dean Bottino
  1. Takeda Pharmaceuticals, Inc., San Diego, CA, USA


Background Conditionally activated bispecific T cell engagers (TCEs) have the potential to provide a larger therapeutic window by reducing off-site on-target toxicity. Conditionally Bispecific Redirected Activation (COBRAs) are novel TCEs designed to be activated preferentially in the tumor microenvironment (TME).1,2 The conditionality of the COBRA prodrug is mediated upon binding of the high affinity target binding domain to the tumor antigen and by cleavage of the prodrug by matrix metalloproteases such as MMP9/2. The cleavage of the prodrug results in the release of the inactive CD3e VH/VL domains leading to the formation of the active dimer, responsible for tumor killing. In order to understand the dependence of treatment effect on drug and patient properties for this novel modality, we developed a mathematical model of COBRA’s mechanism of action.

Methods A mechanistic model was developed to describe the preclinical pharmacokinetics, target engagement and immune synapse formation in the TME, and tumor volume change due to resulting T-cell dependent tumor cell-mediated cytotoxicity (TDCC). Unknown parameters were optimized using in vitro TDCC data and in vivo efficacy data. Sensitivity analyses were conducted to investigate the influence of molecule- and TME-specific characteristics on synapse formation and treatment effect.

Results The model recapitulates the experimental data in vitro and in vivo. We used the model to identify parameters that govern the synapse formation and treatment outcome, such as COBRA affinity to TAA and CD3 (figure 1A), and the active protease concentration in the TME (figure 1B). The model also suggests that the drug-induced antigen internalization along with the abundance of target cells and effector T cells can impact the predicted treatment effect of COBRA and that COBRA concentration-synapse formation relationship does not exhibit the bell-shaped TCE concentration-synapse profile due to its conditionality on homodimerization once cleaved.

Conclusions The model framework was used to integrate preclinical datasets and project the short-term treatment response in human. This work demonstrates the impact of conditionality of the COBRA design on its dose-response relationship. Importantly, factors such as target abundances and protease activity in the TME can strongly influence synapse formation and the resulting anti-tumor effect, and therefore may serve as baseline biomarkers for patient stratification.


  1. Dettling DE, Kwok E, Quach L, Datt A, Degenhardt JD, Panchal A, et al. Regression of EGFR positive established solid tumors in mice with the conditionally active T cell engager TAK-186. J Immunother Cancer. 2022;10(6). Epub 2022/06/22. doi: 10.1136/jitc-2021-004336. PubMed PMID: 35728872.

  2. Panchal A, Seto P, Wall R, Hillier BJ, Zhu Y, Krakow J, et al. COBRA: a highly potent conditionally active T cell engager engineered for the treatment of solid tumors. MAbs. 2020;12(1):1792130. Epub 2020/07/21. doi: 10.1080/19420862.2020.1792130. PubMed PMID: 32684124; PubMed Central PMCID: PMCPMC7531513.

Ethics Approval The animal study protocol received ethics approval by In-Vivo Technologies, Inc Institutional Animal Care and Use Committee IVT-17-002-Y6.

Abstract 1301 Figure 1

Local sensitivity analysis showing the effect on the maximal number of synapses formed by varying KD for CD3 and A) KD for the TAA or B) the active protease concentration in the TME

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.