When it came time for Itzy Morales Pantoja to start her Ph.D. in cellular and molecular medicine, she chose a laboratory that used stem cells—not only animals—for its research. Morales Pantoja had just spent two years studying multiple sclerosis in mouse models. As an undergraduate, she’d been responsible for giving the animals painful injections to induce the disease and then observing as they lost their ability to move. She did her best to treat the mice gently, but she knew they were suffering. “As soon as I got close to them, they’d start peeing—a sign of stress,” she says. “They knew what was coming.”
Even though the mouse work was emotionally “very, very difficult,” Morales Pantoja remained committed to her research out of a desire to help her sister, who has multiple sclerosis. Three years after the project wrapped up, however, Morales Pantoja was crushed to find that none of her results would be of any direct help to people like her sister. An antioxidant she’d tested seemed promising in mice, but in human samples it was ineffective.
This was a disappointment but not a surprise. Around 90 percent of novel drugs that work in animal models fail in human clinical trials—an attrition rate that contributes to a $2.3-billion average price tag for every new drug that comes to market.
On supporting science journalism
If you’re enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.
Today Morales Pantoja is a postdoctoral fellow at the Johns Hopkins Center for Alternatives to Animal Testing, where she is helping to develop lab-grown models of the human brain. The goal is to advance scientific understanding of neurodegeneration while moving the field beyond what some researchers see as an antiquated reliance on animal models.
Millions of rodents, dogs, monkeys, rabbits, birds, cats, fish, and other animals are used every year for research purposes worldwide. Exact numbers are hard to come by, but advocacy group Cruelty Free International estimated that 192 million animals were used in 2015. Most of this work occurs in four broad domains: cosmetics and personal products, chemical toxicity testing, drug development, and drug-discovery research.
Animal-based studies have contributed to important findings and lifesaving medical advancements. The COVID vaccines, for instance, were developed in animals, including mice and nonhuman primates. Animal models have also been critical in advancing AIDS drugs and in developing treatments for leukemia and other cancers, among many other uses.
But animal studies often fall short of producing useful results. They may weed out possibly effective drugs or miss toxicity in humans. They have failed to deliver breakthroughs in certain fields of medicine, including neurological conditions. A 2014 study estimated that candidate therapies for Alzheimer’s disease developed in animal models have failed in clinical trials about 99.6 percent of the time. “As questions about human biology and variability get more complex, we are bumping up against the limits of animal models,” says Paul Locke, an environmental health scientist and attorney at the Johns Hopkins Bloomberg School of Public Health. “The thing you run into with animals—and there’s no way to get around this—is that animal biology is just too different from human biology.” Other species are no longer providing the insights about human biology—including at the cellular and subcellular levels—that scientists today need to achieve innovation.
A growing, multidisciplinary community of researchers around the world is investigating alternatives to animal models. Some are motivated by concerns about animal welfare, but for many, sparing the lives of millions of creatures is just an added bonus. They are driven primarily to create technologies and methods that will approximate human biology and variability better than animals do.
For the past decade or so dozens of labs, start-ups and nonprofit groups have pursued alternative methods that range from machine-learning tools that predict chemical toxicity to living “organs-on-a-chip” that can be combined to replicate human organ systems. Their efforts have now matured to the point where some labs are phasing animals out entirely. Research is beginning to show that these new methods often provide significantly more accurate answers than animals do.
Legislation is beginning to reflect these developments. In 2021 the European Parliament passed a resolution to phase out animal testing in research. Australia’s national science agency has begun seriously exploring nonanimal models for medical product development. In 2022 President Joe Biden signed a bill that did away with a long-standing U.S. Food and Drug Administration requirement for animal testing as a part of every new drug application. In May 2023 Maryland passed a first-of-its-kind law mandating that animal-testing labs contribute to a fund that will be directed to other labs developing human-relevant alternatives. Another federal bill, introduced in 2024, would pave the way for the FDA to begin accepting data from new methodologies on a wide scale.
This convergence of developments in legislation, industry and science will bring about “a sea change in how we conduct biomedical research,” says Danilo Tagle, director of a group at the U.S. National Institutes of Health that is leading an institution-wide push to invest in alternatives to animal models. This year the NIH is launching a $300-million fund that specifically supports the development, validation and testing of nonanimal alternatives for drug screening, disease modeling, and more. This resource will be on top of the 8 percent of the NIH’s $40-billion research budget already awarded for alternative methods, a percentage that has been rising for the past 15 years. As Tagle says, “We’re seeing a convergence in legislation, industry and scientific developments.”
In 1937, when 12 patients came to Archibald “Archie” Calhoun complaining of infections, the physician from Covington County, Mississippi, did what he often did: he wrote them prescriptions for sulfanilamide, an antibiotic he’d used for years. Within days, six of the patients were dead. The pharmaceutical company that produced sulfanilamide had added a new ingredient to the raspberry-flavored formula: diethylene glycol, a type of antifreeze, which turned out to be deadly. “This realization has given me such days and nights of mental and spiritual agony as I did not believe a human being could undergo and survive,” Calhoun wrote afterward.
Paul Locke of Johns Hopkins and others say it’s not a question of whether animal testing will be phased out of most research but when.
The “sulfanilamide disaster,” as the incident came to be known, took the lives of more than 100 people, many of them children. Congress responded with the Federal Food, Drug and Cosmetic Act, a set of laws designed to ensure that no company would ever again unknowingly sell a toxic drug. Among other things, the act required that new drugs in development be tested on animals before being given to humans. “An early success of animal models was to keep these horrible products off the market,” Locke says.
Today animal models are still considered the standard for pharmaceutical and drug-discovery research, partly because many people in the scientific community still get value from them and partly because they’re the status quo. Yet the full extent of animal use in the U.S. is unknown. Federal laws do not require researchers to make public the number of rats, mice and birds—the three species that make up more than 95 percent of testing subjects—bred for research purposes. Likewise, no comprehensive analyses have tallied the amount of U.S. government–funded research that uses animal models, according to Tagle.
People for the Ethical Treatment of Animals, or PETA, has estimated that just under half of NIH research funding goes toward animal-based studies. Organizations outside of government and academia use animals for research as well. The Humane Society of the United States estimates that more than 50 million animals are used for research purposes every year in the U.S. alone.
Locke and others say it’s not a question of whetheranimal testing will be phased out of most research but when. “Everyone recognizes that the goal is to eventually try to replace animals,” says Naomi Charalambakis, associate director of science policy at the Federation of American Societies for Experimental Biology, a nonprofit group that represents 22 scientific societies and more than 110,000 researchers worldwide. But animals are not going to disappear from research soon. “We’re still very much in the nascent phases,” Charalambakis says.
The Interagency Coordinating Committee on the Validation of Alternative Methods (ICCVAM), which operates under the auspices of the U.S. National Institute of Environmental Health Sciences, is a group of 18 research and regulatory agencies working together to promote new scientifically valid methods that are rooted in human biology and that reduce and replace animal tests. According to Nicole Kleinstreuer, a computational toxicologist and executive director of ICCVAM, the team prioritizes “scientific projects that will ultimately result in regulatory translation and implementation.”
How quickly that happens, though, will vary. Locke suspects discovery research—which seeks to understand basic mechanisms of biological systems—will probably take the longest because it is the most complex domain that scientists use animals to explore. For this research, animals offer the advantage of being living creatures with complete organ systems that interact in a coordinated fashion—something that in vitro approaches cannot yet do.
The cosmetics and personal products industry is the furthest along in doing away with animal testing, primarily because of consumer demand. More than 2,500 North American cosmetics, personal care and household product companies are certified as animal-free. Twelve U.S. states and 45 countries have banned animal-tested cosmetics, and legislation reintroduced in the House in September 2023 could add the full U.S. to that list.
Phasing animals out of some types of toxicity testing—carried out to establish a substance’s potential to cause harm—is probably next in line. Numerous studies have shown that, in many cases, artificial-intelligence-based algorithms trained with preexisting data are as reliable as or more reliable than animals in predicting toxicity for various chemicals. In 2016 President Barack Obama signed an amendment to the Toxic Substances Control Act directing the Environmental Protection Agency to begin reducing the use of vertebrate animals in toxicity testing and replacing it with alternative methods if scientifically feasible.
The EPA has made some progress toward this goal. In 2018, for example, the agency granted 62 waiver requests for reductions in animals undergoing certain toxicity testing, sparing around 16,500 animals and resulting in savings of about $8.9 million in its first year. In 2024 the EPA published a new framework for assessing eye irritation or corrosion through alternative test methods.
Some environmental groups have opposed a complete transition away from animal testing. As Susanne Brander, an ecotoxicologist at Oregon State University, summarized last year in work published by the Environmental Defense Fund, “the looming concern” is that new methodologies may miss negative effects that animal models would have caught, “potentially allowing toxic chemicals to appear in consumer products or end up in our environment.”
Kleinstreuer says she understands why consumer protection groups might be wary about these changes. But she emphasizes that the EPA’s motivation in phasing out animal testing “is to actually provide better human health protection using the best science that is fundamentally rooted in human biology.”
To see what the future of human-centric models might look like, I visited Emulate, a biotechnology company in Boston. Emulate specializes in organ chips: flexible polymer platforms, about an inch long, that duplicate human cell and tissue microenvironments.
Sushma Jadalannagari, a tissue engineer, let me play lab tech. Working in a biosafety cabinet, I sucked up trypan blue dye and inserted a pipette tip into a tiny divot in the top of a pristine chip. As I released the fluid, a thin, inky line appeared along a hollow channel that crisscrossed the chip and ended at another opening at the opposite end. A second channel ran below that one, separated from it by a porous membrane.
Real researchers seed this kind of chip’s channels not with dye but with human cells. Multiple chips, each lined with different types of organ-specific cells and tissues, can be linked to mimic multiorgan systems, and researchers can run experiments on one or more chips by flowing fluid or air across the cells, exerting mechanical forces on them or adding things such as pharmaceuticals, cigarette smoke, chemicals, viruses or bacteria.
Emulate’s chips can stand in for liver, kidney, colon and duodenum, and the company also offers blank chips that can be customized. Outside researchers have used the chips to create about 70 additional models with cells from their labs. Emulate’s customers are pharmaceutical firms, academic labs and government facilities.
Emulate spun out of the Wyss Institute at Harvard University, based on the work of Don Ingber, an animal-loving cell biologist and bioengineer, who began developing in vitro models 40 years ago because he didn’t like experimenting on living creatures. “I used to joke that I was raised by dogs, but now a cat rules my life,” Ingber says. His work on organ chips was fast-tracked in 2012 when he and his colleagues received a $37-million grant from the Defense Advanced Research Projects Agency to develop the technologies. Ingber is now regularly approached by agencies, foundations and companies with offers to apply for funding.
Advances in so-called organ chips could accelerate the phasing out of animal testing in laboratories—and lead to better results.
Emulate is one of a growing number of businesses pursuing alternative methods that scientists can use in their research, a space only five to 10 companies were exploring a decade ago. These tests and devices are not designed to be exact replicas of human organs in healthy or disease states. Nor are they meant to serve as one-to-one replacements of animal models. Rather the goal is to recapitulate functions and features that scientists need to study for a particular problem. A liver chip doesn’t have to perfectly simulate a human organ; it just needs to accurately answer a question that researchers are using it to address.
Emulate’s liver chip is the company’s most popular organ model. That’s because one of the primary reasons new drugs fail is that animal models don’t reveal human hepatotoxicity—a condition that occurs when the liver is damaged by exposure to harmful substances.In a study published in 2022 in Communications Medicine, an Emulate-led team evaluated 870 liver chips across a blinded set of 27 known hepatotoxic and nontoxic drugs. The chips correctly identified 87 percent of hepatotoxic drugs—none of which were detected with animal models. Further, the chips did not inaccurately label any safe drugs as toxic, as animal models commonly do. Based on these findings, the authors calculated that the pharmaceutical industry could generate an additional $3 billion or more a year if it routinely used liver chips.
Increasing so-called predictive capability saves time, money and animal lives, according to a 2022case study by Moderna. The pharmaceutical company used Emulate’s liver chips to screen 35 drug-delivery molecule candidates. The liver chips allowed it to complete that task in a year and a half at a cost of $325,000. If the company had performed the same tests in nonhuman primates, Moderna says, it would have cost more than $5 million and taken five years.
Academic labs are also inventing new tools. Vasiliki Machairaki, a molecular biologist at the Johns Hopkins School of Medicine, has been creating a nonanimal model of Alzheimer’s disease, inspired by her grandmother’s diagnosis. Machairaki uses blood samples collected from people with Alzheimer’s to make stem cells, which she differentiates into brain cells and brain organoids, self-assembled three-dimensional tissue cultures that look a bit like trays of pale Dippin’ Dots. The organoids begin to show signs of their donors’ pathology within about four months, enabling the researchers to test the effectiveness of various pharmaceuticals against Alzheimer’s. “This is a personalized model that could eventually tell you the best drugs to use for different patients,” Machairaki says.
In a Johns Hopkins lab headed by biomedical engineer Deok-Ho Kim, researchers culture human heart tissue across plates holding 24 dime-size wells. Electrodes stimulate the heart tissue with an electric current, and magnetic sensors allow the researchers to measure the twitching force of the beating muscle. Some lab members are testing the cardiotoxicity and effectiveness of new chemotherapy drugs—many of which fail in humans because they are unsafe or don’t work—and others are screening new therapies for muscular dystrophy, a group of wasting diseases. Treatments have been found to improve symptoms of some forms of the disease in mouse models, but they don’t work in human patients, many of whom die from heart failure in their 30s.
The engineered heart models, like the brain organoids, are derived from stem cells that carry the genes of their donors. This opens up opportunities for studying patients who have been traditionally overlooked in research, including ones with rare diseases for which “no [animal] model exists at all,” Tagle says. “Rare diseases are poorly studied, and there’s little interest in developing animal models for them because it takes a lot of time and effort.”
Molecular biologist Anicca Harriot says the ability “to do experiments that are directly relevant to the patient” was a motivating factor in her decision to join Kim’s lab as a postdoctoral fellow. Harriot was diagnosed with a rare blood-clotting disorder when she was a child, and doctors couldn’t tell her anything about her prognosis, because the small number of patients with her condition had precluded clinical trials. Conducting research with human stem cells rather than animal models “helps to shift this work toward equity,” she says.
Conferences dedicated to exploring alternatives to animal-dependent methods, such as the Microphysiological Systems World Summit, attract 1,000 or more attendees. Many of them are doctoral students and postdocs who are looking to build a scientific career using what they see as the tools of the future.
In November 2022 U.S. federal prosecutors unsealed an unusual indictment: felony charges against eight people for allegedly running an international monkey-trafficking ring. Until 2020 China was the world’s largest supplier of captive-bred lab monkeys. But wildlife trade bans during the COVID pandemic triggered an international shortage in lab monkeys—around 70,000 of which are used every year in the U.S. alone. According to the government, the trafficking group had used false documents to smuggle hundreds of illegally captured wild long-tailed macaques—an endangered species—from Cambodia to Florida and Texas to supply the research industry.
Locke sees this “hot mess” as something that should have been a glaring message to the scientific community about the need to be more proactive in its pivot away from lab animals and toward human-centric alternatives. By and large, though, that wasn’t the reaction. Instead “the research community screams, ‘We need more macaques!’” Locke says—a shortsightedness that he equates to “asking in the 1950s, ‘How do we get more slide rules?’”
The cultural shift away from animals has not been easy. “To say you want to have a research career but don’t want to do animal work, you’re still a little bit laughed at,” says Antonia Egert, a physician at the University Medical Center Freiburg in Germany.
On top of that, it’s difficult for researchers to make the transition because regulators have yet to clearly spell out what is needed for an alternative model to replace an existing animal-based test, says Breanne Kincaid, a doctoral student in environmental engineering at Johns Hopkins. Although the FDA and EPA broadly state that they will accept nonanimal toxicity data, their regulators have not “put pen to paper to say these are the accepted standards you need to meet to use your model,” she continues. This means scientists who use alternative methods have no guarantee that findings they submit to regulators “won’t simply be met with a vote of no confidence and a request for additional animal data.”
Policymakers have also sent mixed signals about whether labs should be investing in alternative methods. When President Biden signed the FDA Modernization Act 2.0 in 2022, the new law did away with a 1938 mandate that animal testing be a part of every new drug application and authorized the use of the best nonclinical model instead. The Modernization Act is “a really encouraging sign of change,” Locke says, but it has yet to filter down to specific guidelines, standards or policies.
In response to questions about the agency’s plans for translating the Modernization Act to real-world decision-making, an FDA spokesperson wrote, “While the FDA is committed to doing all that it can to reduce the reliance on animal-based studies in the broad context of human drug development, animal testing is scientifically necessary in most circumstances because the current state of science does not support replacing all animal studies with alternative methods.”
That stance could change if the FDA Modernization Act 3.0, introduced by legislators in February 2024, is passed. It would require the FDA to establish a process for qualifying new methodologies so they can be used for drug development. This year the NIH distributed upward of $30 million to several academic centers that will work with input from the FDA over the next five years to qualify some organ and tissue chips as sanctioned drug-development tools. If the bill passes, “I’d anticipate a surge in demand for these things, largely coming from the pharmaceutical industry,” Tagle says.
Tagle and his colleagues recently secured around $300 million from the NIH Common Fund for a 10-year project to develop and validate new methods for biomedical research. The Complement Animal Research in Experimentation program will encourage interdisciplinary teams that include engineers, computational scientists and physicists to advance scientific development and regulatory acceptance of nonanimal methods. Reaching the next generation of scientists through workshops and conferences is another integral part of the plan, Tagle says, “so that when they start their own labs, it becomes natural for them to employ those new technologies.”
