The more complex immuno-oncology (IO) therapeutics become, the more damage a late-emerging preclinical toxicology signal can do. Antibody-drug conjugates (ADCs), monoclonal antibodies, T cell therapies and other biologics are designed for precision, but safety liabilities can still appear when lead selection, in vivo planning and clinical strategy are already well underway.
For ADCs, the challenge is rarely simple. Toxicity can emerge from target-dependent effects, target-independent uptake, payload release, linker behavior, tissue sensitivity or a combination of these factors. Reviews of ADC toxicity have shown that dose-limiting toxicities are often shared across ADCs that carry similar payloads, even when they target different antigens, making it especially important to understand functional tissue effects earlier in development.¹
Regulators are also encouraging more human-relevant approaches. In April 2025, the FDA announced a plan to reduce, refine or potentially replace animal testing in the development of monoclonal antibodies and other drugs, specifically encouraging New Approach Methodologies (NAM) data in investigational new drug applications.² The FDA roadmap describes a stepwise move toward scientifically validated NAMs, including advanced in vitro assays, organ-on-chip systems and computational models.³ The NIH has also announced a larger push to expand human-based research technologies while reducing animal use where scientifically appropriate.4
In Europe, the EMA has taken a similar step, issuing a draft qualification opinion in March 2026 for a new approach methodology that would replace standard animal control groups with virtual control groups in preclinical dose-range finding studies. The agency described this move as a blueprint for future NAM applications.5
For IO developers, the question is: how can teams generate more relevant safety data without making early development harder to manage?
Many IO programs still rely on a mixture of animal data and primary cell studies. Both have value, but their limitations are well-documented. Species differences can obscure safety signals. Donor variability confounds interpretation. Some tissues simply cannot be sourced consistently.
Compared to these systems, iPSC models provide controlled differentiation, large homogeneous batches, and dependable behaviour across experiments. With iPSCs, results are very steady and reproducible. This is vital when comparing multiple candidates or studying subtle functional effects.
Animal studies are still important in preclinical testing, but they are not equally informative for every question. IO biologics can interact with human tissues in ways that are difficult to reproduce in animals, particularly when receptor expression, immune biology or tissue distribution differs by species.
This can become a big problem with a big price tag. A candidate may look promising in efficacy screens and still carry tissue risks that are not detected until later. In ADC programs, those risks can stem from multiple sources that binding data alone won’t reveal, such as:
Expression profiling can help identify where a target appears. Binding assays can show whether a molecule attaches to a tissue. But what happens next? Does the candidate alter cell health, trigger stress, affect function or induce cytotoxicity?
That is where human-relevant, functional biologics testing can add value.
A more useful early safety screen should do several things well:
Real-time impedance monitoring is one way to capture that dynamic response. Impedance-based assays measure changes in cell behavior over time, including effects related to cell attachment, morphology, proliferation and viability. Studies evaluating xCELLigence and related real-time cell analysis platforms have shown their utility for pharmacology and toxicology investigations in early drug development.6
For IO developers, that view can help distinguish a mild, delayed response from rapid cytotoxicity, support dose-response interpretation and help compare how different tissues react to the same candidate.
The Ncyte® IO Safety Panel was developed to address the need for earlier, functional, human-relevant in vitro safety data. The panel evaluates cytotoxic risk across seven healthy human iPSC-derived tissues in a single standardized workflow:
This seven-tissue configuration is a starting point. The panel can be expanded or adapted based on the unique requirements of a drug development program. This design gives IO teams a broader view of tissue sensitivity before committing to subsequent in vivo work. It also reduces some of the variability when separate tissue assays are run in different labs, under different conditions or at different time points.
As part of Ncardia’s broader iPSC-based drug discovery platform, the panel is intended for candidates such as ADCs, monoclonal antibodies and cell therapies. By assessing functional toxicity before in vivo commitment, it helps teams make more informed decisions about which leads to advance, which risks require follow-up and where additional mechanistic work may be needed.
This is especially useful when candidates are difficult to separate by efficacy alone. In IO programs, several leads may show similar tumor cell killing. A differentiated safety profile can help determine which candidate has the best chance of moving forward successfully.
Ncardia tested the well-characterized ADC cetuximab-MMAE to determine whether the system could detect tissue-specific toxicity patterns that reflect known clinical observations.
The panel showed dose-dependent toxicity across specific tissues. Higher concentrations produced expected liver and lung signals. Melanocytes showed sensitivity even at low concentrations, which is consistent with the high frequency of skin-related adverse events observed clinically with EGFR-targeted therapies. The intestinal epithelial model did not show direct toxicity, a finding that is consistent with evidence that some gastrointestinal adverse events in patients are immune-related, rather than being driven by direct epithelial injury.
No model answers every safety question. The current panel is designed for functional cellular screening across healthy tissues, not to fully model immune-mediated adverse events. Future versions may incorporate additional cell types and an immune component to address that additional layer of IO biology.
A useful NAM needs a defined context of use, reproducible performance and data that help teams make better decisions. The strongest models will be the ones that are transparent about their value and their boundaries.
The shift toward NAMs will be gradual. Most developers will not stop using animal studies overnight, and regulators are not asking them to override proper scientific judgment. NIH has clarified that researchers may continue to use animal models when scientifically appropriate and justifiable, while also encouraging human-based approaches and model selection based on relevance to the research question.8
The T cell therapy field provides a precedent here. Several years ago, TCR-T cell developers faced a lack of accurate animal models that could reliably predict human safety outcomes. That need drove the development of iPSC-based functional safety panels like this one. The FDA has since accepted this type of in vitro dataset as part of IND submissions for TCR-based programs.
For many IO teams, the sensible path is to run human-relevant NAMs alongside established models to get:
The opportunity is already significant. Earlier preclinical toxicology can help teams identify avoidable risks, compare candidates more confidently and refine dosing or design strategies before the most expensive decisions are made.
The next stage of IO development will largely depend on asking better questions. Human iPSC-derived cells give researchers a way to ask those questions against the biology that ultimately matters: how healthy human tissue responds. For ADCs, antibodies and cell therapies, that makes safety pharmacology one of the most valuable risk mitigation strategies available in early development.
iPSC models can provide human-derived tissue systems for assessing functional responses earlier in development. In biologics testing, this can help teams evaluate tissue sensitivity, dose-response, and cytotoxicity patterns before committing to larger in vivo programs.
IO candidates can affect more than one healthy tissue. A multi-tissue panel allows researchers to evaluate cytotoxicity across relevant cell types in a single workflow, helping them compare tissue-specific responses under consistent experimental conditions. The xCELLigence readout offers a “simple” but highly valuable cytotoxicity readout, which can then be followed by more complex MOA studies on any risk tissues.
Binding-based screens can show whether a candidate interacts with a tissue. The Ncyte IO Safety Panel is designed to measure functional effects over time using real-time impedance monitoring. That helps researchers assess whether binding or exposure leads to changes in cell health, stress or cytotoxicity. With in vivo models often lacking translatable expression profiles, this panel is currently the only way to accurately quantify functional cytotoxic effects on physiologically relevant human cell types.