NAMs Are Coming for Animal Models. Is Your 3D Culture Ready?

Team Dynamic Matrices | 2 February 2026

For most of the history of drug development, animal data was the gate to human trials by statute. That is no longer true, and the pace of change since 2022 has been faster than most labs have adjusted to. New Approach Methodologies, which explicitly include organoids, iPSC-derived assays and organs-on-chips, are moving from supplementary evidence toward primary evidence. That raises the bar for how those models are built, and the matrix underneath them is now a regulatory question rather than an academic one.

What has changed, with dates

• December 2022. The FDA Modernization Act 2.0 (Pub. L. 117-328) removed the statutory requirement for animal testing before investigational new drug applications, allowing "cell-based assays, organ chips and microphysiological systems, computer models" and other non-animal methods to be submitted instead.
• April 2025. The FDA published a roadmap to reduce animal testing in preclinical safety studies, starting with monoclonal antibodies, with a stated goal of making animal studies "the exception rather than the rule" within three to five years, and named organoids and organ-on-chip data as acceptable evidence. In the same month the NIH announced it would prioritise human-based research technologies and stop funding new animal-only proposals where alternatives exist.
• Ongoing. The FDA Modernization Act 3.0, which directs the FDA to establish a formal qualification process for NAMs, has passed both chambers of Congress (Senate, December 2025; House, July 2026) and is pending reconciliation and signature. The EMA and the OECD are on parallel tracks, and ICH guidance is being revised.

The direction is settled. What is open is how quickly each program has to be ready.

What "acceptable evidence" demands

Academic reproducibility problems are costly but contained: a failed replication is a bad week. Regulatory reproducibility problems are disqualifying: a submission that cannot show consistent, characterised results across batches and sites does not carry weight. A NAM-based submission will be asked:

• What exactly was the model made of, including the matrix, and can that be reproduced from a specification?
• How consistent are results across matrix lots, operators and sites?
• Were the mechanical and biochemical properties of the culture environment defined and reported?
• Does the materials chain of a model that claims to replace animals itself depend on animals?

Those are the questions organoid researchers have been asking themselves about variability for a decade. They now have a regulatory audience.

Why the usual workaround does not scale

Most labs manage matrix variability by qualifying a favoured lot and stockpiling it. That works for a project. It does not work for a submission, because the lot will be gone by the time the data are defended, and "we used lot 1234567 which we no longer have" is not a reproducible specification. Kozlowski, Crook and Ku (Communications Biology, 2021) and Aisenbrey and Murphy (Nature Reviews Materials, 2020) both make the point that a matrix whose composition can be stated is a precondition for a model whose behaviour can be defended.

The animal-derived matrix paradox

There is an irony regulators have started to notice. Basement membrane extract is harvested from mouse sarcomas. An organoid grown in it, submitted as evidence that animal testing is unnecessary, is only partial progress toward the goal it is meant to serve. As reviewers scrutinise the complete materials chain of NAM submissions, an animal-derived, undefined matrix inside the model is an awkward contradiction. It is not disqualifying today. It is the kind of thing that gets asked about, and the answer "we could not find an alternative" is becoming less true each year.

What a chemically defined matrix gives you

Composition that can be specified and manufactured rather than reconstructed from tissue. Mechanical properties that are set by design and reported on a certificate of analysis. Lot-to-lot identity, so a replication three years later uses the same material. No animal in the chain. Those are the properties a submission needs, and they are structural to a defined synthetic matrix rather than something added later.

DyNAtrix was built to those properties: fully synthetic, animal-free, defined polymer and DNA crosslinker composition, stiffness and stress relaxation specified per lot, endotoxin tested. We publish the numbers because a regulator will ask for them (Peng et al., Nature Nanotechnology, 2023).

A preparation checklist

The three-to-five-year window allows preparation and not procrastination. Labs that address matrix reproducibility only when a submission deadline arrives will be validating under pressure, on top of years of data generated in an undefined material. What we suggest now:

1. Audit current protocols for matrix variability. Which assays use an undefined matrix, and how much do results move between lots?
2. Default new programs with regulatory ambitions to a chemically defined, animal-free matrix. Migrate existing ones at a natural break after a side-by-side validation.
3. Document the matrix like a cell line. Product, lot, composition, mechanical properties, endotoxin, in every methods section and every internal record.
4. Report distributions, not representative images, so cross-batch consistency is on file.
5. Keep the animal question in view across the whole materials chain: matrix, medium supplements, coatings.

The labs that thrive under NAM regulation will be the ones whose 3D systems already meet its reproducibility standards, not the ones retrospectively validating years of undefined-matrix data. Today's matrix choice is tomorrow's regulatory readiness decision.