Animals in the service of science

A practice defended and attacked as though it were one thing

A The use of animals in research is argued about in terms that neither side quite believes. The defence tends to present the practice as an unfortunate necessity underwritten by strict regulation and steadily diminishing; the objection tends to present it as cruelty sustained by inertia and commercial interest. Both descriptions contain a good deal of truth, and both avoid the question that determines everything else, which is not whether the practice is justified in principle but how much of it actually produces knowledge that could not have been obtained another way.

B The strength of the practice's defenders is that they can point to specific results; the strength of its critics is that they can point to specific animals. Neither is answering the other. The framework that governs the work in most wealthy countries was proposed in 1959 by two British scientists and is memorable enough to have survived unchanged: replace animals with other methods where possible, reduce the number used to the minimum that will answer the question, and refine procedures to lessen suffering. It is a genuinely useful framework and it is not a moral argument. It presupposes that the research will happen and directs attention to how, which is why it is accepted by regulators, by researchers and by a good many of the practice's critics, and why it settles nothing.

C Numbers are harder to establish than one might expect. Rodents account for the overwhelming majority of animals used, fish are a rapidly growing proportion, and species that attract most public concern — dogs, cats, primates — together constitute well under one per cent. Totals have fallen in some countries and risen in others, and much of the variation reflects changes in what is counted rather than in what is done, particularly the treatment of genetically modified animals bred but never used in a procedure. Anyone quoting a confident trend is usually quoting one jurisdiction's definition.

D The scientific case against the status quo is stronger than the sentimental one and is made largely by scientists. Drugs that work in mice fail in human trials at a rate that has not improved in decades; a substantial majority of compounds entering clinical testing on the strength of animal results do not survive it. Some of this is unavoidable, since a trial exists precisely to find out. But some is systematic: inbred rodents housed in identical conditions, of one sex and one age, model a human population poorly, and models of complex conditions such as stroke and neurodegeneration have proved particularly unreliable. The failures are expensive, and they are paid for in human trials as well as in animal lives. A related and better-documented problem is that a great deal of animal work is never published, usually because it produced nothing, which means the literature systematically overstates how well the models perform.

E Alternatives have improved to a point that would have seemed implausible twenty years ago. Cultured human tissues grown as three-dimensional organoids reproduce the architecture of organs well enough to study development and disease; devices that hold living cells under flow, so-called organs-on-chips, reproduce the mechanical environment that flat cell culture cannot; and computational models trained on large biological datasets now predict some toxicity outcomes better than animal assays. Regulators have begun to accept such evidence, and legislation in several jurisdictions now explicitly encourages it.

F The limits of those alternatives are equally real and are frequently understated by their advocates. No organoid has a circulatory system, an immune system or a brain, and a great many questions concern precisely the interaction between organs — how a compound is metabolised in the liver and what the products do to the kidney, how an infection is contained, why a drug that is safe in isolated tissue causes seizures in an organism. Until something reproduces a whole body, the argument that alternatives can replace animal work entirely is an argument about a future technology rather than a present one.

G What follows from this is less satisfying than either side's position and probably closer to the truth. A substantial portion of current animal use is scientifically weak, poorly designed, statistically underpowered and unlikely to produce transferable knowledge, and it continues because it is what reviewers expect to see and what funders have always funded. Another portion answers questions nothing else can currently answer. Treating these as a single practice to be defended or abolished guarantees that the first portion is protected by the necessity of the second, which serves the interests of nobody, including the animals.