For centuries, animals have been used in scientific research because they share organs, genes and biological processes with humans. That sounds reasonable until you ask the question that actually matters scientifically.
Do experiments on other species reliably predict what will happen in humans? Increasingly, the answer is no.
A review by Lauren Hope and Jarrod Bailey examines both the scientific limitations of animal experiments and the barriers preventing greater adoption of human-focused New Approach Methodologies, or NAMs. Their starting point should be uncomfortable for anyone still describing animal experiments as the “gold standard”.
Around 92% of novel therapies fail during human clinical trials, mostly because of safety and efficacy problems that animal experiments failed to predict. We are imprisoning, harming and killing animals to predict human biology, then discovering that the predictions frequently do not work.
That is not merely an animal rights problem. It is a scientific problem.
A Mouse Is Not A Small Human
Mice and humans share around 85% gene homology. That figure can sound impressive until you look beyond it.
Gene regulation matters. Protein structure matters. Metabolism matters.
The CYP enzyme family is responsible for metabolising around 80% of drugs in humans, yet even closely related enzymes can function differently between species. Mice also differ from humans in their skin, gastrointestinal tract, heart, metabolism, immune system and nervous system.
Using primates does not make the problem disappear. Rhesus macaques share around 90.76% genetic homology with humans, but differences in metabolism and immune function can still undermine translation. HIV/AIDS vaccines have been effective in primates without producing an effective human vaccine. Similarity is not identity. And when the differences determine whether a drug works, fails or harms someone, they are not trivial. The consequences are enormous. Bringing a new therapy to market costs an estimated US$1.3 billion to $4 billion, while potentially useful treatments may also be abandoned because they failed in another species.
Approximately 2.68 million animals were still used in UK research in 2023.
The Experiment Changes The Animal
There is another problem. Laboratory animals are not biological machines sitting passively in cages. They are individuals experiencing confinement, handling, artificial lighting, noise, isolation from their natural environments and the distress of other animals around them. That affects their bodies.
In one analysis of pain sensitivity in mice, environmental stressors accounted for 42% of experimental variability. Think about that. Nearly half of the variation in the results could be explained by the stressful environment created by the experiment itself.
The laboratory does not simply observe biology. It can alter it.
We Already Have Another Direction
NAMs include computer modelling, human cell-based systems, organoids and organs-on-chips.
Organoids can recreate aspects of human organs and diseases. Organs-on-chips can combine human cells, tissues and physical conditions to model how organs function. Multiple organ systems can even be linked together. These approaches are not science fiction.
Human liver chips, for example, have already demonstrated species-specific toxicity differences. Fialuridine caused signs of liver injury in human liver chips but not rat liver chips. Another compound produced toxicity in rat liver chips but not human ones.
The review concludes that combinations of advanced 3-D models and computer analysis could ultimately offer greater predictive power than animal models.
So why has science not moved faster?
Partly because science is conducted by institutions, and institutions develop habits. Researchers worry that NAMs will not be funded. They worry journals will demand animal experiments. They are unfamiliar with new techniques. Some simply trust the methods they were trained to use.
One survey found that 77% of US researchers did not believe NAMs could sufficiently replace animals in their own work. In the Netherlands, around 71% of researchers using animals doubted replacement was possible in the near future.
Publication practices can reinforce this.
In one survey, 31% of researchers said they had used animal methods pre-emptively because they expected reviewers to demand them. Another 44% said reviewers had actually requested animal data.
Then there is money.
The US National Institutes of Health has pledged around $18 million a year towards developing and standardising NAMs. The review estimates that roughly $19.6 billion of its budget goes towards animal studies.
We should not be surprised that animal-free methods develop more slowly when the old system receives vastly more support. Animal experimentation is often defended as though advocates are demanding that scientists abandon proven science for ethics. But that framing is backwards.
The scientific case for replacing animals is growing alongside the ethical one.
NAMs still have limitations. They require greater funding, standardisation, validation and expertise. But the existence of challenges is an argument for investing in better methods, not for indefinitely preserving a system with profound limitations of its own. And replacement must mean replacement.
Adding an organ-on-a-chip to an experiment while continuing to imprison and kill animals is not the destination.
The goal should be science built around human biology where human disease is being studied. Better science. Better predictions. And no animal has to suffer for it.