Recently, the U.S. Senate unanimously passed the FDA Modernization Act 3.0 , requiring the FDA to revise its regulatory rules within one year of the act's enactment— replacing the previous term "animal testing" with "non-clinical trials ." This new concept does not eliminate animal research, but rather integrates it, along with cutting-edge technologies such as AI models, organoids, and organ-on-a-chip, into a more scientific evaluation system.

01

The reform began

The direct background of the FDA Modernization Act 3.0 can be traced back to the FDA Modernization Act 2.0 passed in 2022. In Act 2.0, Congress removed the statutory requirement that "animal testing is mandatory" in the development of new drugs , shifting the FDA's legal authorization from "animal testing as a necessary prerequisite" to "allowing the use of multiple non-clinical methods to support safety evaluation." However, this legal change has not been simultaneously reflected in the FDA's specific regulatory framework . Until the end of 2025, 21 CFR (Title 21 of the Code of Federal Regulations) still largely retains clauses with "animal testing/data/research/models" as their core wording. This creates an inconsistency between the "metaphysical" and the "practical": the law has loosened, but the regulatory text remains within the old framework.

The core content of the FDA Modernization Act 3.0 is precisely to address this issue explicitly. The act requires the U.S. Department of Health and Human Services (HHS), through the FDA, to issue an "interim final rule" within one year of the act's enactment, systematically revising the relevant provisions in 21 CFR concerning new drug development and non-clinical safety evaluation. The key to the revision lies in standardizing regulatory language : replacing phrases such as "animal testing/data/research/model" in the regulations with the corresponding "non-clinical" items.

On the surface, it seems like a word game, like the debate between Hao Liang and his consort, but behind it lies a clear institutional direction.

"Non-clinical trials" is not a newly introduced concept, but rather a regulatory term long used by the FDA to refer to all research activities conducted before human clinical trials to support safety and risk assessments. The 3.0 bill does not aim to "replace animal studies with new methods," but rather to shift the regulatory default from a single approach of animal testing to a multi-strategy approach encompassing human in vitro systems (such as organoids, organ-on-a-chip/microphysiological systems), computational and modeling methods (such as PBPK, quantitative systems pharmacology models, AI/machine learning toxicity and immunogenicity prediction), and high-throughput human cell assays and human tissue ex vivo assessments, to support the assessment of human safety and risk at different levels. The legal pathway for non-clinical safety evaluations will be broadened in the regulatory text, providing a framework for accepting new methods in subsequent regulatory practice and specific review judgments.

The new bill is not just a word game; it has already begun to be implemented in the real world.

While the Senate passage of the FDA Modernization Act 3.0 is scheduled for December 2025, regulatory adjustments did not begin at that time. In fact, prior to the formal requirement to amend the regulatory text, the FDA had already initiated a series of internal policy and implementation adjustments in the area of non-clinical evaluation.

In April 2025, the FDA released a public roadmap for reforming non-clinical safety evaluations, explicitly stating that it would gradually reduce reliance on animal testing for certain drug types and explore incorporating more new methods into an acceptable safety evidence framework. This roadmap did not apply a one-size-fits-all approach to all therapeutic areas or molecular types, but rather chose to start with areas where regulatory experience was relatively mature and the mechanisms were more clearly understood, with monoclonal antibodies listed as the first priority.

In this roadmap, the FDA explicitly states that "New Approach Methodologies" (NAMs) will be an important supplement to non-clinical evaluation systems. These methods include AI-based computational models, organ-on-a-chip systems, organoid models, and other human-related testing systems. It is important to note that the FDA does not describe NAMs as "the default alternative to animal testing" in its policy statement, but rather positions them as a set of tools that can be used to construct safety arguments, and their acceptability depends on the quality of evidence provided and the ability to interpret risks.

This policy orientation is highly consistent with the "nonclinical" language required by Regulation 3.0. The FDA repeatedly emphasizes in its roadmap whether nonclinical evidence can support a reasonable assessment of human risk. From a regulatory enforcement perspective, the FDA states that it may provide procedural incentives to sponsors using high-quality non-animal methods in terms of review pace and communication mechanisms , such as reducing repetitive supplementary requirements if the nonclinical safety package is well-developed and logically clear. While these statements do not constitute an explicit commitment, they reflect that the FDA has begun internally exploring how to translate methodological diversity into review efficiency.

02

Long-standing problems with non-clinical predictive ability

The deeper reason driving the FDA to adjust its non-clinical evaluation framework stems from a long-standing and unavoidable problem within the drug development system: the predictive power of safety and efficacy assessments in non-clinical studies for clinical outcomes remains consistently low . The gap between the results of non-clinical studies guided solely by animal trials and subsequent human trials has already drawn the FDA's attention.

For many years, in the fields of small molecules, antibody drugs, and more complex biologics, a large number of candidate drugs that have shown acceptable safety and pharmacological activity in animal studies have failed in humans at a high rate . These failures include safety issues, insufficient efficacy, or failure of mechanism extrapolation. The common thread is that the risk perceptions established in animal models often cannot be validated in humans. This fact does not mean that animal models are "ineffective" or "useless," but rather that in an increasing number of disease areas, the information provided by animal models is insufficient to independently support a reliable assessment of risk in humans.

This issue becomes more prominent as drug development increasingly targets and targets more complex biological mechanisms of action. In fields such as immune regulation, neurological diseases, and metabolic regulation, diseases themselves are highly dependent on human-specific physiological and pathological characteristics. Differences in key aspects of animal models, such as immune structure, metabolic pathways, and disease progression, objectively limit their predictive value. Regulatory bodies are gradually recognizing that using animal testing as the default core evidence for non-clinical evaluation systematically amplifies uncertainty.

At the same time, the cost structure of non-clinical research failures is also changing. As clinical trials expand in scale, enrollment criteria become more stringent, and trial cycles lengthen, the cost of delaying risk exposure to the clinical stage is rising rapidly. Against this backdrop, regulatory agencies are showing significantly increased focus on whether unfeasible projects can be identified at earlier stages.

Given this, instead of clinging to a not-so-reliable standard, it's better to explore new methodologies.

It is under this pressure that the FDA began to re-examine the structure of evidence: whether non-clinical data can clearly explain the source, pathway, and uncertainty boundaries of potential risks. Empirical extrapolation based solely on animal models is increasingly failing to meet this requirement. The FDA's growing emphasis on WoE (Weight of Evidence) reflects a shift away from relying solely on a single model, and towards constructing a chain of evidence through cross-validation of multiple models (animal + digital + in vitro).

03

Potential impact on R&D and industrial structure

The changes will first be reflected in the R&D decision-making level.

Under the traditional approach, the core task of non-clinical safety evaluation is relatively clear: to assess toxicity, pharmacokinetics, and preliminary efficacy through a mature and predictable combination of animal studies, thereby supporting clinical trials. Regulatory communication often focuses on whether the trials are compliant. With the introduction of the new legislation, this logic is being disrupted. Sponsors now need to answer whether the submitted non-clinical evidence reasonably supports the assessment of human risk.

This shift will fundamentally impact the design of non-clinical studies. More and more companies will consider combining multiple pieces of evidence early in the project, rather than passively supplementing data after animal trials are completed. This doesn't mean animal research is marginalized, but rather that its role shifts from a "single core piece of evidence" to part of a holistic risk assessment. In this structure, the value of non-clinical research will depend on its ability to clearly explain the sources, mechanisms, and uncertainties of potential risks. This requires companies to have stronger interdisciplinary integration capabilities in the non-clinical phase, incorporating different types of data into the same risk assessment, rather than treating them as independent supplementary materials.

At the industry level, this change is reshaping preclinical research and CRO structures.

For a long time, CROs with animal testing as their core competency have constituted the infrastructure for non-clinical research, with a highly standardized and large-scale service model. However, with the increasing diversity of methodologies, this single-evidence industry model will be impacted. Providing only animal data may no longer be sufficient to meet the overall needs of FDA regulatory communication.

At the same time, demand for services related to human-related models, computational methods, and multimodal data integration is rising. Under regulatory frameworks, these methods also face stringent validation requirements, including reproducibility, outcome stability, and correlation with clinical outcomes. What truly matters is not a single technology, but the ability to integrate multiple non-clinical methods into an evidence framework that is regulatory-understandable and acceptable.

Leading CROs are beginning to incorporate "non-animal methods" into their long-term capability configurations, rather than as a hedge against policy risks . Charles River Laboratories launched the Alternative Methods Advancement Project, systematically expanding human-related testing and computational methods while maintaining traditional animal testing services, reflecting the industry's realistic assessment of a potential restructuring of non-clinical evaluation structures.

Market-level funding flows further reinforce this signal. Grand View Research projects that the organ-on-a-chip market will grow from approximately $157 million in 2024 to $952 million in 2030, representing a CAGR of over 35%; the organ-related market is expected to expand from approximately $1.4 billion in 2025 to $4 billion in 2035. This rapid growth is not based on a single regulatory event, but rather stems from the anticipated increasing demand for human-related models within the long-term research and development framework.

04

Post-transformation uncertainty

While the FDA Modernization Act 3.0 has clearly defined the direction for adjusting regulatory language at the legislative level, its impact on actual review practice still heavily depends on subsequent rule refinement and enforcement . What the industry truly needs to focus on is under what conditions and in what manner these methods will be incorporated into specific review decisions.

The primary uncertainty lies in the legislative process . The FDA Modernization Act 3.0 passed the Senate in December 2025 and was sent to the House of Representatives on December 17th. Currently, the bill is in the House stage but has not yet completed its review and vote. In the subsequent process, if the House passes the text of the bill, it will be submitted to the President for signature; after the President's signature, the bill will officially become federal law. After the bill becomes law, a statutory time limit will be triggered, requiring the FDA to complete the relevant regulatory revisions and issue an interim final rule within one year.

The second uncertainty stems from the specific wording of the yet-to-be-released "interim final rule." While the bill stipulates wording substitutions and time limits, it leaves considerable room for interpretation, applicability, and supporting guidelines when the rule is implemented.

Furthermore, the differences between different therapeutic areas and molecular types remain an unavoidable reality in non-clinical reforms. The FDA has made it clear that it will not simultaneously promote the same degree of adjustment in all areas. Immunotherapy, biologics, and targeted drugs with relatively clear mechanisms may enter the review practice with higher methodological diversity earlier, while in areas with highly complex toxicity risks and significant long-term safety uncertainties, animal research may still maintain its core position for a considerable period. This differentiated implementation will give "non-clinical research" different practical connotations in different scenarios.

Finally, the diversification of non-clinical methods places higher demands on reviewers in terms of data interpretation, model understanding, and interdisciplinary judgment. Even after the regulations are revised, the stable formation of review standards will still take time and will gradually emerge through the accumulation of individual cases. This process may lead to phased differences in implementation and also means that the industry needs to adapt to a more uncertain regulatory interaction environment for some time.

Overall, the FDA Modernization Act 3.0 addresses the issue of rule consistency, rather than implementation details. Its long-term impact will truly be determined by the specific application methods after the rules are issued, the pace of implementation in different areas, and how regulatory agencies balance methodological openness with risk control in practice. These factors will collectively shape the actual form of the non-clinical evaluation system in the coming years.

https://news.yaozh.com/archive/46833.html