BA festival of science: Many drug trials using animals are improperly run and provide no proof whether the chemical being tested will be useful in humans. Patients on trials have actually become ill after supposedly positive animal tests failed to identify problems with an experimental drug, writes Dick Ahlstromin York
The BA festival of science in York heard evidence from a consultant neurologist that many animal drug trials are badly designed, incorrectly run and deliver biased information.
Dr Malcolm Macleod, who is based at Stirling Royal Infirmary and at the University of Edinburgh, was speaking yesterday on the last full day of the festival.
He presented his published research into how animal trials of drugs useful in stroke were conducted. One study published earlier this year in Trends in Neuroscience noted that 880 drugs were tested in animal models of stroke and 550 seemed to work in animals. Yet when transferred into human trials, of the 114 tested, only one worked in humans.
Dr Macleod questioned why so many animal tests showed promise, but so few were suited to humans. He put it down to grossly improper animal trials that wasted time, money and the animals themselves.
A good trial, whether in animals or humans, attempts to remove hidden biases that would distort the findings. It also follows procedures that make it impossible for the scientist conducting the trial to introduce distortions or assumptions. For this reason, subjects on a trial must be selected randomly and the scientist must be "blind" to which test subject gets the active drug and which receives a dummy or placebo drug. Dr Macleod's new report showed that these techniques are often not applied, particularly where the drug is being tested for "efficacy", ie whether it has a positive effect.
Many of these tests are funded by drug companies attempting to bring the drugs to market.
Dr Macleod looked at how 288 animal trials were conducted. He found that only 36 per cent randomised the animal subjects and only 29 per cent of the researchers were completely blind during the trial.
Calculations relating to the number of animals needed for a trial were also of crucial importance, he said. Too few and the result would be imprecise; too many and the costs and wastage of animals would be too high. Despite this, he found that only 3 per cent of the trials made any reference to these calculations.
Dr Macleod calculated how many animals might be needed per trial, and determined that between 40 and 50 were necessary. He found, however, that on average just eight animals per trial were used, which he said is far too few for any proper conclusions to be drawn.
He said this high degree of falsely positive animal trials could lead to risks in human trials, citing the case of a test drug, Tirilazad. The animal results showed it worked well, but it made humans ill and was quickly withdrawn.