NEW YORK – Mice have served as living proxies for testing the activity of investigational drugs before starting human clinical trials for more than 85 years. But now, the US Food and Drug Administration (FDA) wants to phase out animal testing in the development of monoclonal antibodies and other drugs.
The rise of precision medicine has coincided with the development of nonanimal preclinical testing tools that attempt to mimic human biology. For the companies marketing those tools, the FDA policy is a welcome — if not lagging — change. Experts, however, predict it'll be quite some time before animal testing fully disappears from drug development.
At Children's Mercy Research Institute in Kansas City, Missouri, Scott Younger, the director of disease gene engineering there who develops organoids, noted that he and his colleagues have spent a lot of time developing and refining these newer preclinical models but still urged caution. "We need to be thoughtful about how we proceed," he said.
Last week, the US National Institutes of Health (NIH) said it will establish a new Office of Research Innovation, Validation, and Application to expand funding and training in nonanimal approaches to drug testing to align itself with the FDA's initiative. The policy could affect the future development of precision drugs, many of which are likely to be monoclonal antibodies, given the market success of Merck's Keytruda (pembrolizumab) and Genentech's Herceptin (trastuzumab).
The shift to alternative approaches to animal testing is made possible by the bipartisan FDA Modernization Act 2.0, which Congress passed in 2022, allowing drugmakers to submit nonanimal preclinical data in applications seeking the FDA's permission to proceed to human clinical trials. On April 10, the FDA presented a roadmap for implementing the policy over three years, during which time it will look at existing international data, encourage researchers to submit nonanimal data, and reduce minimum timelines for animal tests when combined with new technologies.
In five years, the agency hopes to make animal safety and toxicity testing the exception, not the norm.
"For too long, drug manufacturers have performed additional animal testing of drugs that have data in broad human use internationally," said FDA Commissioner Martin Makary in a statement. "This initiative marks a paradigm shift in drug evaluation and holds promise to accelerate cures and meaningful treatments for Americans while reducing animal use."
A long-anticipated shift
Companies developing new technologies for toxicity testing — including organoids, organs-on-a-chip, and complex cell models — say it's about time.
Mary Duseau, the president of Molecular Devices, which develops complex 3D biological models for drug discovery and development, called the policy "a game changer."
"The technology has been there, and we're all getting better at this, but the industry has really been nervous about adopting them, because there wasn't a stance by the FDA," Duseau said.
The shift away from preclinical animal testing is happening at a time when drugmakers are developing more precision medicines like biomarker-targeted drugs and autologous cell therapies. The implications for precision medicine research may be "profound," believes Carolina Lucchesi, a principal scientist at BioNexus, a program within the American Type Culture Collection, a century-old nonprofit that provides biological materials to the life sciences community that has moved into developing patient-derived organoids and 3D spheroid kits in recent years. Lucchesi said ATCC has seen a "significant" increase in demand for these technologies since the FDA's announcement.
"These three-dimensional organoids or spheroids better represent the biological complexity of tumors compared to traditional monolayer models, mimicking the architecture and heterogeneity of the original tumor and providing a more realistic system for preclinical study," Lucchesi said, adding that new methodologies capture broader diversity in patients.
Though Lucchesi believes concerns about the accuracy of organoids are "overstated," skeptics maintain that the science behind new methodologies isn't iron-clad, and that it's premature to shift completely away from animal testing. "If these technologies were ready for prime time to fully replace animal models, they'd already be doing it," said Matthew Bailey, the president of the National Association for Biomedical Research, which has long represented researchers who work with animals. "The profit motive is there. It's that [these tools] are not completely validated yet."
There is precedent for the FDA's move. Gradual replacement of animal models has been championed by the European Medicines Agency since at least 2010, when a European Union directive instructed drug manufacturers to use alternative methods when possible, but a total shift hasn't happened yet. Based on the most recently released data, between 2018 and 2022, the number of animals used in experiments and for other scientific research purposes decreased by 5 percent in the EU. There were 16 percent fewer animals used for regulatory purposes in 2022 compared to the prior year and 32 percent fewer compared to 2018.
For this reason, the FDA's announcement didn't really shock those in biomedical R&D, said Vanessa Ott, the senior director of life sciences portfolio strategy and marketing at Promega, which offers a range of cell assays designed for 3D cell models. Promega's assays have been used to support artificial intelligence-based modeling and organs-on-a-chip for years, which Ott claims can more accurately predict drug toxicity and efficacy than animal models, especially for antibody therapeutics.
"We get requests pretty regularly from researchers that are looking to develop in vitro assays that really highly reflect the biology of how these drugs work," she said.
Higher rates of clinical trial success?
An oft-cited statistic in drug development is that 90 percent of all investigational therapies fail in clinical trials. Indeed, McKinsey analysts tracked drug development progression from 1996 to 2014 to compute that figure, finding slightly higher success rates — about 18 percent — for biologics. Another, more recent analysis by McKinsey shows that as of 2021, around 4,000 clinical trials are launched in the US each year, but only 13 percent succeed.
These failures increase the cost of drug development. In research published in 2023 in the Journal of the American Medical Association, a group of consultants and analysts determined that across the industry, pharmaceutical companies spend an estimated $50 billion to $60 billion on failed oncology trials. In a different study, the mean cost of developing a new drug jumped from $172.7 million to $515.8 million when the cost of studying therapies that failed to reach the market was included.
Supporters of alternative preclinical technologies posit that human-derived 3D cell and AI models could improve the proportion of drugs that successfully reach the market. Duseau said Molecular Devices' "bold ambition" is to reduce the length and cost of the drug development process by 50 percent.
"There's just incredible waste in the market. They're just not coming out with new therapeutics at the rate that they should for the investment being made," she said. "These models should unlock that."
The new approaches could be particularly useful for advancing precision medicines that rely on biomarkers and other information to identify best-responder populations, Duseau said. Even though the FDA previously required animal testing prior to clinical trials, organoids are being used in earlier stages of drug development to help researchers identify biomarkers to target, gain a better understanding of the genetic features of tumors, and identify drugs that they should abandon.
For n-of-1 treatments, organoids have helped preclinical researchers test the efficacy of CAR T-cell therapy for cancers like renal cell carcinoma and glioblastoma, as well as establish methods for "off-the-shelf" CAR T cells.
Others aren't so sure that doing away with preclinical animal testing will necessarily improve the drug failure rate. "There are many, many reasons why drugs fail, not [only] because animal models were used," Bailey said.
Addressing challenges in the lab
Despite the potential financial incentives, researchers who work with alternative technologies to develop precision medicine treatments warn they're not ready to take on the full burden of efficacy and toxicity testing quite yet.
Joseph Wu, the director of the Stanford Cardiovascular Institute, works with patient-specific and disease-specific induced pluripotent stem cells (iPSCs) to develop "clinical trials in a dish." Wu's company, Greenstone Biosciences, provides iPSC lines to academic institutions around the world for free with the goal of accelerating preclinical development timelines and reducing reliance on animal testing.
Wu's research has shed light on patient-specific toxicity caused by certain cancer drugs using in vitro models for cardiovascular disease, including cardiotoxicity associated with HER2-targeted cancer drugs such as Herceptin. In one case, his team was able to use CRISPRi/a screens of iPSC-derived cardiomyocytes to determine that a CA12 antagonist could protect cancer patients from cardiotoxicity after taking the commonly prescribed chemotherapy doxorubicin.
Despite his embrace of new methodologies for developing targeted therapies, Wu still works with mice. In the doxorubicin study, for instance, he corroborated his iPSC findings within an in vivo mouse study.
In some cases, researchers learn critical information about the safety and efficacy of drugs via animal testing that they would not have using alternative tools. In one of his studies from 2022, Wu found that marijuana smoking activated inflammatory cytokines in people with cardiovascular disease, which could be tempered with genistein, a phytoestrogen found in soybeans that binds to cells' CB1 receptors. The new drug worked great in organoids, but an animal study revealed that low stomach pH degraded the drug. Wu's team was able to address the pH issue to make the drug effective again, but they wouldn't have known to do that without animal testing, he said.
Many new preclinical technologies have yet to show their full potential, Wu said. As of January, no AI-developed drugs had progressed beyond Phase II trials, he found, in part because of the lack of high-quality training data and a propensity for errors and biases. The process of training AI systems does involve a lot of failure when determining targets for precision drugs, he said, but that is also how the models are trained. "Most of the predictions turn out to be false," Wu said. "You then take this drug, go back, test it on the cells, and if it doesn't work, you tell the computer, 'Hey, your prediction was lousy. Can you retrain yourself?'"
As such, he advocates a mixed approach — reducing the emphasis on mouse models and substituting in organoids and AI models where they can help identify drug targets and speed up multiomic sequencing. "I'm a proponent of using everything at our disposal," Wu said. "There's no one-size-fits-all."
Children's Mercy's Younger agrees. "There's no time in the near future where we'll just be able to skip over animal models," he said. "And even if there were ways to build those models using human cultures, they would be so complicated that you wouldn't be able to do it in a scalable way."
In his lab, Younger works extensively with organoids and patient-derived cell models to determine the best treatments for rare genetic pediatric diseases, especially those involving seizures. He has been successful in screening drugs the FDA has approved for other conditions and predicting how they might work in patients with the specific disease he's studying before giving it to them.
"We can really robustly recapitulate clinical responses to therapeutics using patient-derived organoids," he said.
There's a big asterisk, though. Younger is mostly focused on finding new applications of FDA-approved drugs and believes toxicity testing in animal models is still crucial during preclinical development. "We're not taking novel drugs and trying to say, 'Hey look, we tested this once in an organoid, let's go put it into people,'" he said.
"A balanced approach"
Promega's Ott said the FDA's guidance is less a hammer and more an incentive, in part because researchers are still ironing out a few challenges that may require animal models to be used alongside alternatives in some situations, for now. Standardizing models and scaling them is difficult, especially for large molecule therapeutics where drug-specific models might be necessary, she said, but remains optimistic that, eventually, alternative technologies will surpass animal testing.
There are many good reasons to consistently evaluate the use of animals in preclinical modeling, said Nadia Rosenthal, the scientific director of the Jackson Laboratory, a nonprofit biomedical research institute that also provides mouse models to more than 2,400 organizations in 68 countries.
"But making that leap from testing confidence in a mouse or a dish or even an algorithm to predicting how humans are going to respond comes with a substantial degree of uncertainty," Rosenthal said.
She views the Modernization Act as a "wake-up call" to find better ways of developing effective drugs, especially cancer drugs, to which she said an "unnervingly" low number of patients respond. "The way to address this is complicated, but it's not going to be solved by simply eliminating animals," Rosenthal said.
Rosenthal's solution is to "think more cleverly" about how animals are used in drug development, which at Jackson Lab has meant the development of fewer, more genetically diverse mice that can be used in combination with human-derived cell models. She doesn't see the need for animal testing going away any time soon.
"Probably not in my lifetime," she said. "It does not give us joy to experiment on animals. It's simply a necessity."