A DDT for ICANS Risk Assessment in T Cell–Engaging Bispecific Antibody Immunotherapies: Strategy and Rationale

Work in progress …

****************

T cell–Engaging Immunotherapies Pose a Predictable but Complex Risk
T cell–engaging bispecific antibodies (BsAbs), particularly those targeting CD3 and a tumor-associated antigen, have emerged as a powerful class of immunotherapies. These agents leverage endogenous T cells to induce targeted tumor cell death, often achieving deep and durable responses in hematologic malignancies. Similarly, chimeric antigen receptor (CAR) T cell therapies—which involve engineering a patient’s own T cells to recognize tumor antigens—have shown curative potential in certain blood cancers. As of 2024, more than 600 bsAbs are in clinical trials, with over 100 specifically designed as T cell engagers (TCEs) for blood cancers, and all 7 FDA-approved bsAbs to date fall into this category.

Both bsAbs and CAR T therapies share a common mechanism of immune activation that introduces well-documented risks—chief among them cytokine release syndrome (CRS), and in some cases, a downstream neurotoxic condition known as immune effector cell–associated neurotoxicity syndrome (ICANS). CRS occurs frequently in bsAb-treated patients (20-60% of patients) and is generally managed with IL-6 inhibitors, corticosteroids, and supportive care. ICANS occurs occurs downstream of CRS in 5-20% of patients. Symptoms ranging from delirium, to seizures and cerebral edema. ICANs carries significant risk of long term cognitive and neurological deficits and can be lethal. Unlike CRS, ICANS lacks both reliable predictors and targeted interventions. These risks are the “black box” warnings on both CAR T and BsAbs therapies

Pre-clinical Assesments of CRS are Limited and Pre-clinical ICANS models do not exist
Current preclinical approaches— animal models and in vitro cytokine release assays (CRAs) using PBMCs—fail to recapitulate the complex human immune environment and lack predictive power for neurotoxicity. In particular, no validated in vitro or in vivo models currently exist for ICANS. Several in vitro CRA formats have been developed to assess cytokine release potential. These include whole blood assays, solid-phase PBMC assays, and high-density PBMC cultures. While informative, these models are limited in scope: they often exclude important immune components like neutrophils, lack native plasma proteins such as fibrinogen, and do not provide functional readouts of vascular or neurological effects. There are no pre-clinical in vitro models for ICANs.

Humanized mouse models have been used to evaluate CRS, but they do not consistently reproduce the inflammatory network observed in humans. Moreover, species-specific differences in T cell subsets and receptor expression can yield false negatives—as seen in the infamous TGN1412 trial (a T cell engaging monoclonal). Critically, these models have no capacity to assess the risk of ICANS. Recognizing this, regulatory guidance from FDA increasingly emphasizes the need for human-relevant, mechanism-informed in vitro systems for immunotoxicity assessment.

µSiM-ICANS and Integrated Assay Workflow for CRS

To address current preclinical limitations, we propose the µSiM-ICANS drug development tool: a microphysiological system (MPS) that models the human blood-brain barrier (BBB) and its response to systemic inflammatory stimuli. This tool builds on extensive work in the McGrath laboratory to generate iPSC-derived BBB models representing both “healthy” and “vulnerable” phenotypes under inflammatory stress. The platform recapitulates key hallmarks of ICANS pathogenesis, including changes in BBB permeability, fibrinogen extravasation, leukocyte infiltration, and astrocyte activation under physiological flow. The model is further enhanced by MPS-integrated sensors developed by the Miller laboratory, which enable real-time monitoring of cytokine release syndrome (CRS) in an upstream chip and detection of fibrinogen arrival on the brain-facing side of the µSiM-ICANS.

The µSiM-ICANS tool integrates with existing whole blood CRAs. In our proposed workflow, bsAb candidates are first tested in a human whole blood assay to establish CRS cytokine profiles. These plasma samples are then applied to the µSiM-BBB chip to assess downstream effects on barrier integrity and neural injury markers. Functional readouts—such as increased BBB permeability, fibrinogen extravasation, and astrocyte reactivity—serve as translational indicators of neurotoxicity risk. This two-stage system can be applicable both to preclinical dose-setting and to stratification of patients during early-phase trials.

Validation Plan with Glofitamab and Benchmarking Against Patient Samples
Initial studies will use Glofitamab—a clinically approved bsAb—as a benchmark agent. Cytokine levels in stimulated healthy donor whole blood will be compared to those found in sepsis, delirium, and ICANS patient plasma. We will calibrate plasma dilutions to maintain physiological cytokine levels and test these samples on the µSiM-ICANS to map dose-dependent injury phenotypes. Outputs will include BBB integrity, immune cell infiltration, and astrocyte activation. The goal is to derive a no-observed-adverse-effect level (NOAEL) and establish thresholds for safe dosing in first-in-human trials.

Context of Use Statements

Pre-clinical: The µSiM-ICANS is a drug development tool (DDT) intended to evaluate the potential of novel T cell-engaging bispecific antibodies (BsAbs) to induce systemic inflammatory responses that compromise blood-brain barrier (BBB) integrity and contribute to central nervous system (CNS) injury. The assay employs a human BBB chip model exposed to plasma collected from healthy donors following ex vivo stimulation with BsAb candidates. Comparative controls include reference BsAbs known to induce cytokine release syndrome (CRS) and immune effector cell-associated neurotoxicity syndrome (ICANS), as well as negative controls. The tool supports preclinical safety and dose assessment by generating donor-specific CRS profiles and assessing their capacity to induce key markers of ICANS-related neurotoxicity in vitro.