Every FDA-approved T cell-engaging bispecific antibody carries a boxed warning for cytokine release syndrome (CRS). Seven of nine require mandatory hospitalisation during step-up dosing. In pivotal trials of BCMA-targeting bispecifics, any-grade CRS occurred in 59% of patients. Severe CRS (grade 3 or higher) occurred in 2%.
These numbers shape every first-in-human trial for this drug class. Because CRS cannot be reliably predicted preclinically, sponsors set conservative starting doses and escalate slowly. Patients wait longer to reach therapeutic exposure. Trials take longer to complete. The cost is measured in time, money, and delayed clinical benefit. And always, the potential for clinical disaster.
The FDA has just accepted a platform into the ISTAND qualification program that is designed to solve this problem. The µSiM-CRS-ICANS, developed by the University of Rochester’s TraCe-bMPS, is a microphysiological system that generates human-relevant CRS and immune effector cell-associated neurotoxicity syndrome (ICANS) risk assessments from a single donor blood sample. It was accepted into the ISTAND Program on 24 June 2026 under DDT-IST-000065.
The Current Preclinical Toolkit Cannot Effectively Predict CRS
PBMC-based cytokine release assays are the current standard. They produce supraphysiological responses and strip out neutrophils, platelets, and plasma factors that drive clinical CRS. Neutrophils are major contributors to the cytokine cascade in patients, but they are lost entirely during PBMC isolation. These assays overestimate some risks and miss others.
Non-human primate models are worse. In the TGN1412 first-in-human trial in 2006, preclinical testing in cynomolgus macaques showed no evidence of cytokine storm potential at 500 times the human dose. All six volunteers developed life-threatening systemic inflammatory responses within hours. The species difference was later identified: macaque CD4+ effector memory T cells do not express CD28 and simply cannot respond to the drug the way human cells do.
For ICANS, the neurotoxicity that can accompany CRS, the situation is starker. There is no preclinical model. IND applications proceed without ICANS-specific safety data. Neurotoxicity is evaluated only after patients are in the clinic.
The FDA’s own 2025 Roadmap cites the failure of animal models to predict CRS as a reason to prioritise monoclonal antibodies for reduced animal use. The 2023 FDA immunotoxicity guidance identifies microphysiological systems as an emerging approach for characterising cytokine release from T cell-engaging mAbs.
What the Platform Does
The µSiM-CRS-ICANS uses whole blood stimulated with a T cell-engaging mAb to generate a physiological CRS profile on-chip. Unlike PBMC assays, whole blood retains neutrophils, platelets, complement, and fibrinogen. Vascular endothelial cells on the chip contribute IL-6 and IL-1β, two cytokines central to CRS pathophysiology that are absent from cell-free assays. Integrated photonic biosensors provide real-time cytokine monitoring across a validated panel: IL-1β, IL-6, IL-8, IL-10, IFN-γ, TNF-α, and GM-CSF.
A downstream ICANS module then measures what happens when CRS-derived plasma reaches a model of the human neurovascular unit. iPSC-derived brain microvascular endothelial cells co-cultured with pericytes sit on an ultrathin nanoporous membrane. The module quantifies blood-brain barrier permeability, fibrinogen leakage, and neutrophil transmigration as surrogate markers of neuroinflammatory injury.
The Clinical Impact: Faster to Therapeutic Dose
The practical consequence of better CRS prediction is not just safer trials. It is faster trials.
When sponsors lack confidence in CRS risk estimates, they start low and escalate cautiously. Step-up dosing regimens for approved bispecifics can span weeks before patients reach a therapeutic dose. For drugs like teclistamab, patients are hospitalised for 48 hours after each step-up dose.
A platform that provides human-relevant CRS data with defined sensitivity and specificity changes the starting point. If a sponsor can demonstrate that a candidate has a quantified, low CRS risk profile across a donor panel, the justification for a higher initial dose is stronger. Higher starting doses mean fewer escalation steps. Fewer steps mean patients reach therapeutic exposure sooner. For patients with relapsed or refractory disease, weeks matter.
The proposed context of use makes this explicit: the platform would inform maximum safe starting dose justification for IND applications supporting first-in-human trials. The regulatory deliverable is an IND with integrated CRS and ICANS safety data that does not currently exist anywhere in the preclinical package.
What This Means
Fifteen bispecific antibodies approved. More than 650 clinical trials active. A market projected to grow from USD 18 billion to over USD 600 billion by 2035. And the preclinical safety toolkit for this drug class still cannot reliably answer the most basic question: will this molecule trigger a cytokine storm?
The µSiM-CRS-ICANS is the first ISTAND submission to tackle CRS prediction for monoclonal antibodies directly. If qualified, it would give sponsors something they do not have today: human-relevant, mechanism-informed CRS and ICANS data to support dose selection before a drug reaches its first patient. That changes the risk calculus for every bispecific antibody in the pipeline.
More broadly, this may indicate a shift in the FDA’s thinking around IND approvals. Concepts like “DILI” (Drug Induced Liver Injury) are important but broad categories that are often catchalls for an array of toxicity mechanisms. Focusing ISTAND on predicting the specific mechanisms that drive clinical failure ties closely into the agencies recent overhauls to the IND process. What technologies will we see next in ISTAND?
InnovApproach Consulting tracks every ISTAND submission and analyses what each acceptance means for the NAMs-informed drug development landscape. For advisory support on MPS qualification strategy, ISTAND submissions, or NAMs-informed preclinical packages, get in touch.
No responses yet