Animal studies may be poor predictors of drug effects in human clinical trials, according to a systematic review published by the British Medical Journal.

A research team led by Pablo Perel of the London School of Hygiene and Tropical Medicine identified six drug interventions where there had been unambiguous evidence of a treatment effect (whether beneficial or harmful) in clinical trials. They then carried out a systematic review of the corresponding animal experiments to establish the degree of concordance between the effects in animals and humans.

While animal testing was controversial, a poll by the UK’s Medical Research Council had found that most people supported it provided there were benefits to human health, no alternatives existed and there was no unnecessary suffering, the authors noted. However, the usefulness of animal models had been questioned, with some claiming the results could not be extrapolated to humans due to the biological differences between the species and because the outcomes of animal experiments often depended on the type of model used.

The results of the systematic review did not provide any definitive support for these objections, although in some cases there was a significant disjunction between the outcomes of animal and human testing. Corticosteroids, for example, did not show any benefit in clinical trials for traumatic head injury – in fact they were associated with a higher risk of mortality. In some of the animal models reviewed, though, a benefit was seen. And while tirilazad was found in clinical studies to increase the risk of death and dependency in patients with acute ischaemic stroke, in animal models the drug reduced infarct volume by 29% and improved neurobehavioural scores by 48%.

In some of the other interventions there was more consistency between the animal and human results. Thrombolysis with recombinant tissue plasminogen activator, for example, reduced death or dependency in clinical trials despite an increase in intracranial haemorrhage. In animal models, TPA reduced infarct volume by 24%, improved neurobehavioural scores by 23% and increased the probability of haemorrhage.

Interventions with bisphosphonates also tended more towards concordance with animal models. In clinical trials these drugs were found to improve bone mineral density in postmenopausal women with osteoporosis. Where outcomes data were available from animal studies, 11 out of 11 showed an increase in bone mineral density with bisphosphonates, while six out of six studies demonstrated improvements in bone mass.

Little scope for conclusions to be drawn

These varied findings did not offer any scope for general conclusions about the utility of animal research. The researchers did point out that many of the animal studies reviewed were of poor methodological quality. As such, they said, discordance between animal experiments and clinical trials might be due to bias, random error or the failure of animal models adequately to represent human disease. More regular use of systematic reviews could facilitate the translation of research findings from animals to humans, the authors suggested, both by providing insights into the limitations of animal models and promoting better communications between animal researchers and clinical trialists.

Albeit inconclusive, the BMJ report will do little for the argument that animal testing is still the only safe and accurate way to anticipate drug effects in humans. The recent final report by the expert group convened in the wake of the disastrous Phase I study at London’s Northwick Park Hospital was a boon to sceptics in this respect. The report concluded that preclinical studies in monkeys with TeGenero’s monoclonal antibody TGN1412 “did not predict a safe dose for use in humans, even though current regulatory requirements were met." By Peter Mansell