As FDA phases out mandatory animal testing, researchers map what can stand in
A Perspective in The Innovation charts where organoids, organ-on-a-chip systems, and AI models can replace animal studies—and where animals remain essential
image:
The graphical abstract contrasts traditional animal experimentation in the drug R&D pipeline—spanning disease modeling, efficacy assessment, and drug toxicity—with the transition to new approach methodologies (NAMs), including organoids, microphysiological systems, and computational modeling, bridged by the FDA's new paradigm. A comparison table outlines six dimensions where the two approaches differ, from physiological relevance and system simulation to screening throughput and regulatory validation.
view moreCredit: Zhu H., Lin X., Weng Q., et al., (2026). From animal models to new approach methodologies: Opportunities and challenges. The Innovation 7(11), 101260.
Every new medicine reaches the pharmacy counter only after passing through years of animal testing. That gatekeeper role is now under review: in April 2025, the US Food and Drug Administration announced it will phase out mandatory animal testing for certain therapeutics, including monoclonal antibodies, and agencies from the NIH to the European Medicines Agency are promoting alternatives known as new approach methodologies (NAMs). "We hope to systematically map the real capability boundaries of NAMs and explore a feasible path for them to evolve alongside animal experiments," the authors of a new Perspective in The Innovation write. The team is from Zhejiang University and Hangzhou City University. The pressure comes from biology. Only 35.3 percent of genes expressed in the human liver are shared with mice, so a compound that looks safe in rodents may not behave the same way in people. NAMs try to close that gap with three families of tools. Organoids—three-dimensional microtissues grown from human stem cells—now reproduce multi-zonal liver development and disease. Organ-on-a-chip platforms link organs together: Harvard's human body-on-chip connects eight organ modules, though each module holds about one millionth of an organ's cells, which can blunt detection of scarce biomolecules. A human liver chip nonetheless identified drugs that cause liver injury with 87 percent accuracy while clearing non-toxic compounds 100 percent of the time—toxicity that often goes undetected in animals. Computational models complete the toolbox: the first AI-guided small-molecule drug, rentosertib, has entered phase 2a trials, and one AI tool predicts the outcomes of nine classic animal tests with over 80 percent agreement—exceeding the reproducibility of the animal tests themselves. The authors are blunt about the limits. Organoids and chips cannot model whole-body immune responses such as cytokine storms; standardized organoid models take three to six months to develop—too slow for a novel pandemic—and AI models are only as good as their existing data. High-dose animal toxicology remains indispensable for setting first-in-human doses. And regulators diverge: the FDA, the European Medicines Agency, Japan's PMDA, and China's NMPA prioritize different validation frameworks, so conflicting interpretations of NAM data could slow drug approvals. The shift is grounded in the long-standing "3R" principles of replacement, reduction, and refinement, but the Perspective advocates a staged transition rather than outright replacement. Near-term measures include getting more out of animal studies—a University of Zurich technique now evaluates about 25 antibodies in a single mouse, cutting animals used in pharmacokinetic and biodistribution studies by up to a hundredfold—and validating human-relevant models through frameworks such as the FDA's ISTAND program. One organ-on-a-chip dataset has already carried a candidate drug, TNT005, into clinical trials—a sign of what verified NAMs may eventually deliver: medicines whose safety is assessed in systems built from human cells, rather than approximated in animals. ### Citation: Zhu H., Lin X., Weng Q., et al., (2026). From animal models to new approach methodologies: Opportunities and challenges. The Innovation 7(11), 101260. https://doi.org/10.1016/j.xinn.2026.101260
Journal
The Innovation
Method of Research
Literature review
Subject of Research
Not applicable
Article Title
From animal models to new approach methodologies: Opportunities and challenges
Article Publication Date
2-Nov-2026
No comments:
Post a Comment