In recent months, the Food and Drug Administration and National Institutes of Health have announced new initiatives to reduce and replace animal testing in biomedical research. Central to these efforts are “new approach methodologies,” such as lab-grown human-based models and computational technologies, promoted as more modern and human-relevant.
But amid this rush toward alternatives, we risk abandoning modern animal models that have become increasingly relevant to human biology. As a biomedical researcher combining advanced mouse models with human-based and computational approaches, I argue that biomedical discovery and drug development require greater investment in refining and complementing — not replacing — animal models.
Advertisement
Mice are the most commonly used animal model in biomedical research, and together with rats they make up around 95% of research animals. It’s no surprise, then, that many foundational discoveries have depended on them.
Although mouse research has long faced criticism for failing to predict human outcomes, today’s models are far more capable of emulating human responses than ever before.
Mice can be modified to carry human components — including genes, cells, and even tissues — to directly study human biology in the context of a whole, living organism rather than simply in a dish or test tube.
Advertisement
For example, in 1993, a clinical trial of the drug fialuridine, which had cleared preclinical animal testing, unexpectedly resulted in liver failure in nearly half of the participants. But researchers later discovered that by adding human cells to mouse livers, these “humanized” mice could predict the same drug toxicity seen in the clinical trial.
While lab-grown human liver models, one of the alternatives promoted by the FDA, may complement animal testing, they are insufficient on their own to capture treatment effects across interconnected organs or to study complex human conditions.
For instance, CAR T-cell immunotherapy, a groundbreaking treatment that uses a patient’s own immune cells to fight cancer, frequently causes toxicities affecting multiple organs, including the brain. While these severe effects have been difficult to study in traditional animal models, researchers using mice carrying human immune cells were able to uncover their causes. Clinical trials are now underway to address these effects and make this therapy safer for patients who urgently need it.
Mice can also be engineered with patient-specific gene variants, helping scientists better understand complex human conditions like autism and develop personalized therapies for genetic diseases. They have even been used in some cases to help guide patient care. For example, a patient’s cancer removed during surgery can be implanted into mice, allowing doctors to test different therapies and select the most effective follow-up treatment for that patient’s unique cancer.
Environmental factors, including diverse exposures and germs, also greatly affect human physiology, metabolism, and immune system. In contrast to the historically ultra-clean and highly controlled laboratory conditions, my colleagues and I are now using “naturalized” mice exposed to more diverse environmental factors to better capture these effects.
With more natural immune systems, these mice enabled researchers to reproduce the negative effects of drugs for autoimmune and inflammatory conditions that had previously failed in human clinical trials. Preclinical testing of new treatments for immune diseases — like rheumatoid arthritis and inflammatory bowel disease — in these mice is especially promising to identify therapies more likely to succeed in patients early on.
Advertisement
Scientists using such mice were also able to identify causes of severe toxicities commonly seen in patients treated with another groundbreaking therapy that releases the brakes on the immune system to fight cancer, which depended on the natural microbes found in these mice.
And many other such studies are ongoing. The promise of naturalized mice was highlighted at a scientific workshop hosted in 2024 by the National Institute of Allergy and Infectious Diseases of the NIH, where researchers from around the world presented evidence supporting their potential to improve our understanding of human disease and advance health.
And it’s not just mice. Many other animal models have evolved in ways that directly benefit human health. Notably, recent advances in genetically modified pig organs for human transplantation — in which harmful animal genes are removed and human ones are added in organs like the heart, liver, and kidney — are a promising step to address donor shortage and help patients with end-stage diseases.
Broadly reducing investment in animal models now, including the expertise, infrastructure, and ethical safeguards currently in place, would not only undermine critical progress but also leave us unprepared when complex medical and safety issues appear in the future.
As the FDA notes, “all possible side effects of a drug can’t be anticipated based on preapproval studies,” and, indeed, many adverse events emerge after approval. In such cases, researchers still use tailored animal models alongside human observations to isolate the causes and refine treatments over time.
In fact, investing in and complementing modern animal models now can reduce future animal use. When preclinical studies better predict human outcomes, fewer animals are needed for repetitive experiments that ultimately fail to translate.
Moving forward, we shouldn’t be choosing between animal models and human-based approaches — but rather integrating both as needed to best study human conditions.
Advertisement
As my colleagues and I have previously proposed and currently practice in our research, the most promising translational results come from combining animal models with human-based and computational approaches. These tools provide complementary insights that enhance how we understand disease and treat patients.
For instance, powerful artificial intelligence tools can benefit from animal data generated under experimental conditions not possible in humans to learn and uncover biologically relevant patterns. Tailored animal models can also be used to test promising AI predictions before they are first applied to people.
Integration of new methods into complex research systems, however, requires inclusive dialogue across diverse expertise and sufficient time for validation before implementing new policies.
As regulators and funders push toward alternatives, they must recognize that today’s animal models aren’t the same ones that failed to predict human responses in the past. Abandoning these advances now — just as humanized and naturalized models are finally delivering on their promise — would be a step backward for both scientific progress and patients who stand to benefit from future breakthroughs.
Agencies should instead invest in systems that unite animal models with other promising methods to most effectively accelerate biomedical research into treatments for patients.
Anis Barmada is a biomedical researcher at Yale School of Medicine, a P.D. Soros fellow, and public voices fellow of the OpEd Project.
