Menu
Close
.svg)
Earlier this month, I penned an essay about my cat, Vivi, and the increasingly sophisticated lengths to which veterinary medicine is going to help cats live longer. The response was considerable, although hardly surprising. There are certainly an awful lot of us who love our cats and dogs as though they were members of the family, because, of course, that is precisely what they have become.
Anyone still harboring doubts need only spend a work day on Teams or Zoom. People use synthetic backgrounds, of an office, of a beach backgrond, etc. to reveal as little as possible about their domestic lives, only for their cat to promenade imperiously across the desk or a dog to materialize over somebody's shoulder. Our beloved animals, apparently, have declined to sign our employers' confidentiality agreements.
Vivi's cameo in my writing got me thinking about some very different animals I encountered a long time ago.
My father worked for the Food and Drug Administration (FDA), and when I was growing up I occasionally went to work with him. Sometimes that meant the old Parklawn Building in Rockville, Maryland; sometimes it meant an HHS office facility in downtown Washington, DC within easy walking distance of the Capitol and the museums of the National Mall. On one of those visits, my father managed to get me inside an animal research laboratory. Of all my childhood experiences and excursions, this one remains unmistakably vivid.
I remember rooms filled with white albino rabbits, their noses twitching as their temperatures were taken simultaneously. I remember monkeys in cages. Most indelibly, I remember watching a researcher perform surgery on a cat's brain. I was fascinated, although even then I found the experience disquieting. Yet there was nothing particularly outré about it. This seemed to be how biomedical science worked. A few years later, dissecting a cat was practically a rite of passage in my high-school biology class.
Fast forward several decades later, something remarkable is happening behind the scenes of pharmaceutical development. The laboratory I visited as a boy is not disappearing, but some of the scientific assumptions that sustained it are being reconsidered.
This week, HHS announced another tranche of initiatives intended to accelerate human-based biomedical research and reduce unnecessary animal testing. NIH is putting more than $88 million into ten infrastructure projects supporting new research approaches. FDA, meanwhile, has changed its regulations to make explicit that scientifically appropriate non-animal methods can be used to generate evidence about the safety of drugs and biologics. The agency's regulations are replacing references to "animal tests" and "animal studies" with the broader "nonclinical tests" and "nonclinical studies." Bureaucratic nomenclature rarely sets one's pulse racing, but this particular bit of semantic housekeeping is consequential. It changes the presumption.
For nearly a century, animal testing has been deeply embedded in pharmaceutical development for good reasons. Before an investigational medicine could be administered to a human being, regulators quite reasonably wanted evidence that it was unlikely to poison one. Animals offered something a petri dish could not: an intact biological system, with organs, circulation, metabolism, immunity and all the bewildering interactions among them.
Animal models contributed enormously to pharmacology, immunology, vaccines and modern medicine. The ethical framework surrounding their use evolved as well. Since 1959, researchers have worked under the principle of the "3Rs": replace animals where possible, reduce the number required, and refine experiments to minimise suffering.
But a sensible scientific practice can, over time, ossify into a default. And defaults are particularly tenacious when regulators are involved. A pharmaceutical company may be intrigued by a novel experimental model, but nobody wants to arrive at FDA with a billion-dollar asset only to discover that the agency is unpersuaded by the evidence dossier.
The other difficulty is embarrassingly obvious: a mouse is not a diminutive human, nor is a beagle, rabbit or macaque. Evolution has endowed us with enough biological commonality to make animal models useful, but also enough difference to make extrapolation precarious. A compound can behave impeccably in an animal and fail in humans because of efficacy or safety. Conversely, an animal may produce a toxicity signal that consigns a potentially useful human medicine to oblivion.
Overall drug-development success rates remain strikingly low. FDA scientists writing in JAMA this year noted that animal studies can fail to predict human responses and argued that newer human-centered models may improve predictions of pharmacokinetics, pharmacodynamics, efficacy and toxicity. This does not mean that animal testing "doesn't work", nor that clinical attrition can simply be laid at the paws of the laboratory mouse. It means something subtler: the best model depends upon the question being asked.
What has changed is not merely our discomfort with imperfect translation. Scientists now possess alternatives that would have seemed fantastical when I wandered through that government laboratory as a boy.
They are collectively known by the rather unlovely acronym NAMs, for New Approach Methodologies, and include human-derived cell cultures, three-dimensional organoids, spheroids, microphysiological systems known as organs-on-chips and increasingly sophisticated computer models. Artificial intelligence can integrate immense quantities of biological, chemical and clinical information to predict toxicity and drug behavior. Researchers can grow miniature approximations of human organs from human cells. A liver-on-a-chip can reproduce aspects of human hepatic physiology without requiring a liver to reside inside an animal at all.
FDA has already qualified an AI-based drug-development tool for use in assessing disease activity in MASH clinical trials. It has established review infrastructure for NAMs, created a public database showing developers where alternative approaches have been acceptable, and issued guidance describing how a new method should demonstrate biological relevance, technical robustness and fitness for its regulatory purpose.
NIH is moving in parallel, building infrastructure to develop, validate and scale human-based technologies. HHS's latest initiatives include plans for a laboratory at the NIH Clinical Center combining standardized human organoids, robotics, AI and advanced data capabilities. The laboratory of the future may (and hopefully, will) therefore look rather different from the one I visited with my father.
Yet an organoid is not an organism. A liver chip does not possess kidneys, hormones, an immune system, a cardiovascular system or a brain. Scientists working in cancer and neuroscience have pointed out that interactions across tissues, organs and behavior remain extraordinarily difficult to reproduce outside an intact living system. NIH and FDA themselves acknowledge that validated alternatives do not yet exist for every application. Some researchers worry, reasonably enough, that enthusiasm could outrun validation and expose humans to risks that animal studies might have revealed.
The sensible destination, therefore, is not an edict declaring animal research obsolete. It is a world in which an animal is used because it is the best available model for answering a particular scientific question, rather than because it has always been the model used to answer it.
This may be the most interesting part of the story.
Congress altered federal law in 2022 to recognize alternative nonclinical approaches. FDA followed with a roadmap in 2025, guidance on particular drug classes, a database of accepted NAM applications and, now, regulatory language reflecting the broader scientific toolbox. The changes have accumulated rather than arrived with a single thunderclap.
FDA scientists writing in JAMA described the emerging paradigm particularly well: instead of reflexively asking for animal data, determine what information is necessary to establish human safety and then identify the best way to obtain it.
That is a deceptively profound inversion. Instead of beginning with the model, begin with the question.
There are already tangible consequences. FDA guidance on monoclonal antibodies, for example, describes circumstances in which lengthy toxicology studies in non-human primates may no longer be warranted. Such six-month studies can use more than 40 primates apiece; FDA authors estimate that eliminating unnecessary extended studies could spare several hundred to more than 1,000 non-human primates annually. Individual animals can cost as much as $50,000, meaning that the implications are economic as well as ethical.
The science is evolving alongside something less quantifiable: our relationship with animals.
Gallup found in 2017 that 44% of Americans considered medical testing on animals morally wrong, compared with 26% in 2001. A different survey, conducted by Morning Consult in 2024 for the Physicians Committee for Responsible Medicine, found that 80% of 2,205 adults agreed that the federal government should commit to a plan to phase out animal experiments, while roughly 85% supported prioritizing non-animal research methods. The questions and methodologies differ, so these figures should not be stitched together into a spurious trend line. But substantial public unease is difficult to miss.
Advocacy groups such as PETA go considerably further, arguing that animal experimentation is both ethically objectionable and scientifically unreliable, and pressing for replacement with human-relevant approaches. Some of those arguments remain vigorously contested within biomedical science. What is less contestable is that animal welfare has acquired a cultural salience it did not possess to the same degree several generations ago.
Nor is this solely an American phenomenon. Britain has published its own program for reducing animal testing, including plans to replace particular tests and reduce the use of dogs and non-human primates. The European Union is developing a roadmap of its own. This increasingly looks less like a parochial regulatory experiment than an international evolution in biomedical science.
Perhaps that should not surprise us. We have learnt considerably more about animal cognition, emotion and social behavior. At the same time, millions of us have developed extraordinarily intimate relationships with the animals living in our homes. When I look at my Instagram feed, half or more of the pages show dogs or cats who live stunningly luxurious lives. They are family. We buy them medications, arrange specialist consultations, treat their cancers, manage their chronic diseases and increasingly wonder how science might give them not merely healthier lives but longer ones. My interest in the science of feline longevity comes from precisely this impulse. I would like Vivi to be around for as long as biology and good medicine will permit.
There is something rather heartening about these two scientific currents occurring simultaneously. On one side, researchers are asking how technology might extend and improve the lives of animals we love. On the other, scientists and regulators are asking whether human cells, organoids, chips, computational models and AI can spare other animals from being used unnecessarily in the development of medicines for us.
For those of us who work in pharmaceutical marketing, commercial strategy, and I&A, preclinical toxicology can feel like a distant province of the pharmaceutical empire. By the time we encounter a medicine, years of laboratory work may already be buried in its biography. Yet what happens upstream eventually changes everything downstream.
Better predictive models could prevent doomed compounds from entering expensive clinical programs. Human-derived systems may eventually help researchers understand differences among patients earlier. AI could increasingly stitch together nonclinical, clinical and real-world evidence. More efficient testing could affect development time, development cost and ultimately which medicines survive long enough to reach patients.
There is also a broader lesson for our industry. Public expectations do not remain neatly confined to the consumer world. They eventually find their way into laboratories, regulatory agencies, corporate priorities and investment decisions. A society that increasingly regards animals as beloved companions rather than disposable instruments will inevitably ask harder questions about when their use in research is genuinely necessary.
I still picture those rabbits surprisingly clearly. I can see the rows of white animals and their perpetually twitching noses, an image preserved from a DC laboratory of another era. The scientists working there were not villains. They were using the tools available to them to answer difficult questions and protect human beings from dangerous medicines. Animal research has contributed immeasurably to modern medicine, and in some areas it will remain necessary for some time.
But perhaps an age of greater scientific enlightenment is one in which progress allows us to become both cleverer and kinder. The purpose of preclinical research has not changed: understand biology, anticipate danger and protect the people who will eventually receive a medicine. What is changing, rather wonderfully, is how much ingenuity we can bring to doing it, and how many animals we may eventually be able to leave out of the bargain.
