The FDA’s Plan to Phase Out Animal Testing

Abstract

Replacing animal testing through “New Approach Methodologies” (NAMs) holds promise for developing cheaper and safer drugs without animal suffering. However, such an approach should be implemented carefully, and it cannot be rushed. We discuss the FDA Modernization Act 2.0 and 3.0 and the FDA’s roadmap to phase out animal testing.

Introduction

The U.S. FDA recently announced its commitment to replacing animal testing in developing monoclonal antibodies and other drugs with “New Approach Methodologies” (NAMs), such as Artificial Intelligence (AI)-based computational models [1]. The FDA hopes that phasing out animal testing will help reduce animal experimentation and enhance public health by providing safer and cheaper drugs that become available more quickly through a faster assessment process and lower R&D costs [1]. The National Institutes of Health (NIH) also announced in July 2025 that it will no longer create funding opportunities for projects focused only on animal testing [2].

While noble goals, phasing out animal testing with NAMs too quickly without a robust plan raises safety and effectiveness concerns. This article explores the FDA Modernization Act 2.0, proposed “FDA Modernization Act 3.0” initiatives, and the FDA’s roadmap to phase out animal testing. Many NAMs are not yet ready for prime time. Starting clinical trials immediately with less or no animal testing can have safety risks. A reasonable approach could be implementing an FDA premarket review or an independent third-party certification process for NAMs. Alternatively, since larger reform is time-consuming and difficult to achieve, sponsors could continue conducting animal testing alongside NAMs until sufficient data has been collected to demonstrate that NAM or a combination of NAMs is at least as reliable as animal testing for predicting drug safety. In other words, while we acknowledge the potential for NAMs to become an important tool for de-escalating risks for human clinical trials and ultimately reducing or possibly eliminating animal testing, rigorous validation is essential.

The FDA Modernization Act 2.0 and 3.0

The 21st century has seen the dawn of a variety of emerging biomedical and digital technologies, such as AI, self-assembled three-dimensional (3D) structures derived from human stem cells (“organoids”), 3D bioprinting, and microphysiological systems (“organ-on-a-chip”). In response to the rapid pace of technological progress today and to join a larger worldwide trend to reduce, refine, and/or replace animal testing (also known as the “3Rs”) [3], Congress passed the FDA Modernization Act 2.0, incorporated into Consolidated Appropriations Act, 2023 (CAA) Section 3209, on December 29, 2022 (Box 1) [4].

Box 1: FDA Modernization Act 2.0 and 3.0.

Over 100 human deaths caused by Elixir Sulfanilamide could have been prevented in 1937 by testing the drug on animals [5]. This crisis ultimately led to the final enactment of the 1938 Federal Food, Drug, and Cosmetic Act (FDCA), which continues to serve as the foundation for FDA regulation of medical products, including drugs [5]. In 1962, prompted by the thalidomide disaster in Europe, Congress amended the FDCA to explicitly authorize and direct the FDA to require “preclinical tests (including tests on animals)” in an IND submission as part of the necessary evidence to start human clinical trials [6]. This statutory language was removed by the FDA Modernization Act 2.0, enacted as part of the CAA, to incentivize NAMs as alternatives to animal testing [4].

In particular, CAA Section 3209 amended FDCA Section 505(i) by replacing the phrase “preclinical tests (including tests on animals)” with “nonclinical tests,” thereby stressing the use of NAMs as alternatives to animal testing as part of the evidence for an IND submission to justify testing the drug in a clinical trial [4]. It also revised the Public Health Service Act by replacing the term “animal studies (…)” with “an assessment of toxicity (…)” as part of the information needed for obtaining a license for certain biological products (i.e., those that are similar or interchangeable with other biological products; biosimilars) [4].

A “nonclinical test” is defined in the new FDCA Section 505(z) as “a test conducted in vitro, in silico, or in chemico, or a nonhuman in vivo test, that occurs before or during the clinical trial phase of the investigation of the safety and effectiveness of a drug [4].” The definition also contains a nonexclusive list (“may include”) of such tests, including “(1) [c]ell-based assays,” “(2) [o]rgan chips and microphysiological systems,” “(3) [c]omputer modeling,” “(4) [o]ther nonhuman or human biology-based test methods, such as bioprinting,” and “(5) [a]nimal testing [4].”

The FDA Modernization Act 3.0, introduced in the Senate by lead sponsor Senator Cory Booker on February 3, 2025 [7], and another bill with the same short title introduced in the House by lead sponsor Representative Buddy Carter on April 10, 2025, aim to require the FDA to update its IND regulations (21 C.F.R. Part 312) to match the language of the CAA, replacing any references to “animal” tests, studies, research, data, and models with “nonclinical” tests, studies, research, data, and models [8].

Since the enactment of CAA Section 3209, the FDA has not yet updated its Investigational New Drug (IND) regulations (21 C.F.R. Part 312) to match the CAA’s language. New legislative initiatives, known as the “FDA Modernization Act 3.0,” aim to change that (Box 1).

The FDA’s Roadmap to Phase Out Animal Testing

In April 2025, the FDA announced its plan to phase out animal testing in developing monoclonal antibodies and other drugs [1]. The announcement stated that implementation of this plan would start “immediately” and that the “inclusion of NAMs data is encouraged” for IND applications, such as those from AI-based computational modeling, organ-on-a-chip systems, or organoids that mimic human organs to test the safety of a drug [1]. To incentivize the use of NAMs, the FDA also promised to update its guidelines to reflect this new approach and to implement a streamlined review for “submit[ting] strong safety data from non-animal tests [1].”

Concurrently, the FDA published a roadmap outlining a strategic, step-by-step approach for reducing animal testing in preclinical safety studies using scientifically validated NAMs (Box 2) [9]. In October 2025, the FDA’s Center for Drug Evaluation and Research (CDER) also published a table listing drug development contexts in which CDER is open to streamlined nonclinical programs, including NAMs use [10].

Box 2: FDA’s Roadmap for Minimizing Animal Testing in Preclinical Safety Studies.

To minimize animal testing in preclinical safety studies, the FDA’s approach is to initially use NAMs for monoclonal antibody testing only and then expand their use to other drug types [9]. Key NAM categories mentioned in the roadmap include (i) in vitro human-derived systems like organoids and organs-on-chips, (ii) in silico approaches like AI predictive models, and (iii) other innovative platforms such as ex vivo human tissues [9]. The FDA further highlighted the need for an integrative strategy where multiple NAMs (e.g., three) could be combined to replace animal studies, promising a higher degree of accuracy and ethical acceptability [9].

Over the next three years, the FDA aims to reduce toxicity testing in animals by accepting pre-existing international drug toxicity data from countries with similar regulatory standards [9]. The agency also plans to encourage sponsors to submit NAM data along with less required animal data or, in select pilot cases, no animal data for IND submissions [9]. The FDA also emphasizes that outcomes of such studies should be closely monitored during the clinical trial to ensure “safety was not compromised [9].” Other FDA strategies include creating a comprehensive, open-access database with international human and animal toxicity data, reducing the primate toxicology testing period for specific monoclonal antibody studies from six to three months, possibly shortening timeframes for animal toxicity testing for additional drug categories, and conducting biannual tracking and quantification of changes in toxicity testing [9].

The FDA’s ambitious goal is “to make animal studies the exception rather than the norm for pre-clinical safety/toxicity testing” in just three to five years [9]. The FDA also plans to collaborate with other agencies like the NIH through the Interagency Coordinating Committee on the Validation of Alternative Methods (ICCVAM), for example, to establish pathways for validating NAMs [9].

Discussion and Recommendations

The FDA’s recent moves to phase out animal testing in preclinical studies show a strong commitment to implementing the FDA Modernization Act 2.0. Updating the IND regulations, as proposed (Box 1) [7–8], would be a logical next step to harmonize terminology, such as replacing terms like “animal” tests with nonclinical tests. Generally speaking, NAMs have great potential to lower R&D costs and address ethical concerns related to animal testing, such as testing on non-human primates [9,11]. They also might improve predictive accuracy in the long run, for example, through the use of human cells and tissues, since animal models are not necessarily perfect predictors for human safety [9,11].

However, despite this enthusiasm for NAMs, many scientists feel uneasy about the FDA’s newest initiative [2]. It will be crucial to adopt a careful, step-by-step approach. Reducing animal testing in preclinical studies is a complex task, let alone replacing it entirely with NAMs. The FDA’s goal “to make animal studies the exception rather than the norm for pre-clinical safety/toxicity testing” within the next three to five years [9] seems overly optimistic, given the absence of non-inferiority or superiority studies demonstrating NAMs are ‘ready for prime time.’ For example, AI is prone to biases, and computer modeling using AI would need to be ethically designed with robust screening for these biases [12]. But how could the FDA even manage to identify potential flaws in an AI model used by the sponsor, given the 30-day window to review an IND submission unless it orders a clinical hold (FDCA § 355(i)(2)-(3), 21 C.F.R. § 312.40)? Developing standards to determine which AI models can be used in the development of specific drugs for IND submissions will be crucial moving forward.

Using bioprinting as a “nonhuman or human biology-based test method[]” is another example that should currently be approached with caution. First, FDCA Section 505(z) does not define “bioprinting,” and the FDA has yet to issue any regulations or clear guidance that establishes the agency’s interpretation of what bioprinting is; no universally recognized scientific or legal definition currently exists [13–14]. Second, the regulatory pathway for bioprinted models used for nonclinical testing remains unclear. For example, should such a bioprinted model follow the same regulatory process as a bioprinted end product intended for transplantation into a patient?

Overall, the FDA should continue insisting on animal testing data to support the investigational use of new drugs until a specific NAM or a combination of them has proven to be as reliable as animal testing for a particular context of use. Concurrently, the NIH should support research that builds the necessary evidence base for such an approach, such as NAM non-inferiority or superiority studies.

NAM development currently occurs mostly in silos, and standards for their development and validation are needed. This requires stakeholder engagement, data collection and sharing, and more publications in peer-reviewed journals [15]. Moreover, the introduction of a premarket review or certification process for NAMs could be helpful. For example, NAMs could be required to undergo FDA review and obtain marketing authorization before sponsors can use them as nonclinical testing evidence. Alternatively, an independent third-party organization similar to the concept of EU Notified Bodies could assess their reliability and certify them. Once a NAM receives FDA marketing authorization or a third-party certification, such information should be publicly available so sponsors know what NAMs can be used for what contexts of use. The authorized or certified NAMs would also likely need to be rereviewed at regular intervals. Relying on sponsors to develop and self-evaluate their NAMs as a company is likely suboptimal in many cases and may lead to intransparency and safety risks.

Of course, introducing a premarket review or certification process requires broader reform that takes time and might be difficult to secure. An alternative approach could be for the FDA to encourage sponsors to use NAMs and submit NAMs data alongside data from animal tests over the next few years, rather than using them to conduct less or no animal testing [9]. Sponsors could optionally submit NAM(s) data without facing any repercussions, effectively as a ‘test run,’ by also providing information to the FDA on why they believe such NAM(s) data support their IND application [15]. Analyzing animal data and NAM(s) data in parallel could be beneficial over time to determine whether specific NAMs are reliable for particular contexts of use. The FDA’s planned open-access repository for human and animal toxicity data could certainly help track this information [9]. To further incentivize sponsors to voluntarily submit NAM(s) data alongside animal data, offering a prioritized review or other incentives like fee reductions could be useful [1].

Training FDA staff will also be essential to ensure clinical holds are issued when NAM(s) data raises safety concerns. It is unacceptable to allow a clinical trial to begin and only verify throughout the trial whether “safety was not compromised [9].” Otherwise, the reduction of “animal experiments” [9] would risk replacement with ‘human experiments.’ The FDA’s plan to phase out animal testing within five years could backfire and put participants at unnecessary risk. In some cases, replacing animal testing entirely with NAMs might not be feasible in the near future and may pose considerable safety risks, such as for whole-organ replacement therapies [11].

Concluding Remarks

In summary, adopting a premarket review or certification process for NAMs, or alternatively, gathering comparative data (i.e., NAMs and animal data), could be a reasonable step before reducing or replacing animal testing. A replacement should only occur if the data from one NAM or a combination of NAMs has been proven to be as reliable as animal data for specific contexts of use. There is still a long way to go. Validating NAMs, conducting non-inferiority or superiority studies, collecting and analyzing data, collaborating with stakeholders, and providing targeted FDA staff training are key milestones needed to successfully move toward fewer animal studies.