Bridging Innovation and Reliability: Why Animal Studies Still Matter in the Age of NAMs

The landscape of preclinical research is shifting fast—and 2025 may mark a defining moment. In April, the FDA announced its plan to remove the requirement for animal testing of monoclonal antibodies and other drugs, and in July, the National Institutes of Health announced it will no longer fund studies that rely exclusively on animal models, signaling a powerful endorsement of New Approach Methodologies (NAMs). These non-animal technologies promise faster timelines, ethical advantages, and, in many cases, clear alignment with human biology.

The science behind NAMs is evolving rapidly, offering innovative, ethically motivated approaches that can enrich mechanistic understanding and support the 3Rs (replacement, reduction, and refinement of animal use). Innovations like organ-on-chip systems, human-derived organoids, and in silico modeling are capturing headlines and industry attention. Startups are emerging with targeted platforms designed to reduce or replace animal use, while regulators are beginning to map out use cases where NAMs may eventually substitute for traditional models.

However, despite this momentum, NAMs are not yet a full substitute for the depth, complexity, and reliability of well-designed animal models. Within the broader scientific and regulatory ecosystem, there remains substantial uncertainty. Many of these tools are still early in their validation lifecycle, and their reproducibility, scalability, and ability to model complex systemic biology have not yet reached the level of established animal models—particularly in high-risk or late-stage programs.

It is essential to recognize that the enthusiasm for NAMs, while warranted, must be tempered with scientific pragmatism. For now, the smartest path forward lies not in replacement of animal models, but in integration.

What NAMs can (and can’t) do

NAMs, as a category, encompass a broad spectrum of methodologies: in vitro human cell/organoid models, high-throughput screening assays, computational toxicology, and microphysiological systems, among others. These technologies are attractive for several reasons: they offer a chance to reduce animal use, shorten timelines, and generate mechanistic insights based on human biology. In some cases—particularly early in drug discovery—they may also be more cost-effective.

The promise becomes especially compelling when one considers disease areas where animal models fall short. Pulmonary fibrosis, for instance, is notoriously difficult to model in animals. Traditional systems often yield false positives, showing efficacy that fails to translate in the clinic. In contrast, precision-cut lung slices from human tissue—a form of NAM—can offer more predictive efficacy data by preserving diseased architecture and enabling direct testing of human tissue responses. But these tools are not comprehensive. While a lung slice can inform efficacy within a specific microenvironment, it cannot answer questions about pharmacokinetics, off-target effects, or systemic toxicity. It provides a piece of the picture—not the whole frame.

This is a recurring theme across NAM categories: Organ-on-chip systems can recreate specific barriers or functions of a single organ but typically lack insight into multi-organ interplay, immune system interactions, and chronic exposure assessments that are vital for understanding real-world drug behavior.

Complex questions still require complex systems

Drug development does not advance based on single answers. It advances based on a body of evidence. Each preclinical question—whether it’s target engagement, biodistribution, chronic dosing, or off-target effects—requires a different type of model to answer with confidence.

For instance, when evaluating how a compound circulates through the body, accumulates in tissues, or impacts systems beyond its intended target, animal models still provide the most reliable framework. This is particularly true for questions involving immune responses, metabolism, or delayed toxicity—areas where NAMs are still in the early stages of technical maturation.

Even when NAMs show clear promise, their application must be narrowly focused and scientifically justified. They are highly fit-for-purpose tools, not generalized substitutes.

Regulatory clarity still lags

The policy climate is shifting, but regulatory certainty remains elusive. While agencies like the FDA have begun offering roadmaps for the use of NAMs in specific areas—such as monoclonal antibody development—guidance for broader categories remains fragmented or undefined.

This ambiguity puts drug sponsors in a difficult position: How can they confidently rely on a model system that may not yet meet regulatory expectations for safety and efficacy data?

For example, while the FDA has suggested that in vitro models may be sufficient for certain monoclonal antibody safety assessments (particularly when animal models are known to be non-predictive), similar allowances have not yet been extended to small molecules or other biologics. As a result, most IND-enabling programs still rely on validated animal models to support regulatory submissions.

This doesn’t mean NAMs should be excluded from safety packages—far from it. But it underscores the need for scientific judgment and regulatory fluency when integrating them into study designs. Until NAMs are backed by sufficient historical data and performance benchmarks, they should be viewed as complementary—not as replacements.

Integration: It’s and, not or

Rather than framing the discussion as a binary choice between NAMs and animal models, industry would be better served by shifting to a model of thoughtful, intentional integration. The best preclinical strategies today leverage NAMs where they add value—such as mechanistic studies, early screening, or target-specific modeling—while continuing to rely on animal studies for broader, systemic questions.

This integrated approach offers several advantages. It allows sponsors to reduce animal use where appropriate without sacrificing the depth or reliability of their safety data. It also helps de-risk clinical transitions by building a more complete preclinical picture.

The key to success lies in choosing the right tool for the right question. Over-reliance on any single modality—whether animal or non-animal—risks leaving critical questions unanswered.

Rethinking CRO engagement

This evolving landscape also changes what sponsors should expect from their contract research partners. In the past, a CRO might have been selected based on access to specific models or lab infrastructure. Today, what matters more is the CRO’s scientific judgment and ability to design adaptive, data-driven programs.

Sponsors should be asking CROs: How do you determine whether a NAM or animal model is the right fit for a given study? Do you have experience integrating novel technologies into standard study designs? Are your teams staying current with regulatory expectations across both modalities?

CROs that take a balanced, evidence-first approach—rather than pushing a particular platform—are best positioned to help sponsors navigate this transitional period. They should also be prepared to advise when an animal model is not predictive and when a NAM offers meaningful value, as well as when a hybrid approach is the most strategic path forward.

A call for evidence-driven evolution

The future of preclinical research is not either/or. It is adaptive, integrated, and driven by the question at hand. NAMs are a valuable addition to the scientific toolbox, and their role will undoubtedly grow. But their integration must be guided by data—not ideology or policy mandates alone.

If the industry is to deliver on its promise of safer, more effective therapies, it must stay focused on what actually works. That means designing programs that are not just innovative but also fit-for-purpose and scientifically robust.

We should embrace NAMs—but not at the expense of reliability. The cost of prematurely sidelining proven models is too high. The path forward requires both optimism and caution, both innovation and experience.

In the end, the goal remains unchanged: to generate the clearest, most actionable data possible to protect patients and advance meaningful therapies. That goal is best served not by choosing sides, but by choosing wisely.