By Jonathan Amos
BBC News Online science staff
The use of genetically modified animals in UK labs continues to grow, official statistics released on Tuesday show.
Rodents do most of the work in the lab
In 2003, these animals were employed in 764,000 "experiments", which represent just over a quarter of all procedures.
Overall, total experiments are also up but only marginally, by 2.2% to 2.79 million - still a significant reduction on what was happening in the 1970s.
GM animals have become increasingly important to science as researchers try to understand how human genes work.
In studying predominantly mice which mimic human illnesses, scientists hope to make sense of these afflictions at the genetic level.
The human genome project found that there were about 30,000 genes directing the functioning of our bodies - but precisely what most of them do is a complete mystery.
By adding or knocking out genes in rodents, scientists believe they can gain an insight into the molecular flaws in humans that lead to disease.
"We have to put this into the context that every cat kills about 40 mice per year; [but that] 4,000 people have artificial heart valves fitted per year, 180,000 people are treated with insulin for diabetes and 30 million people get prescriptions for asthma each year," Professor Chris Higgins, director of the Medical Research Council's Clinical Sciences Centre, told the BBC.
"None of those treatments could possibly have been developed without animal research."
He cited the UK's "very tight" regulatory process which he said kept the number of animals and the amount of suffering to a minimum.
"I wish animals were not used as well but, unfortunately, if it came to the health of my child or the health of a rat I would choose my child," he added.
But this position is questioned by animal welfare groups who are worried the recent downward trend in procedures - the number is nearly half what it was in the 1970s - is now showing signs of a significant reverse.
The British Union for the Abolition of Vivisection (Buav) called on the government to investigate "other means" of testing.
David Thomas, the legal adviser for Buav, said: "There is a large and increasing debate about whether some of these experiments are necessary.
"Only last week a survey of GPs showed that four out of five had comparatively little faith in the results of pharmaceutical tests on animals."
And Andrew Tyler, the director of Animal Aid, told BBC News Online: "There is a wealth of evidence demonstrating that animal 'models' cannot be relied upon for safety testing and disease research.
"This means that - animal suffering aside - human patients of today and tomorrow are being betrayed by the continued use of animals."
Nearly all (85%) of the experiments done in British labs are performed on rodents; mostly on mice and rats. Fish (6%) and birds (4%) account for much of the remainder. Dogs, cats, horses and primates (usually macaques and marmosets) combined are used in less than 1% of all procedures.
This said, the number of primate procedures in 2003 jumped to 4,799, up 822 from 2002. The Royal Society for the Protection of Animals criticised this rise in particular.
"Much is made of the additional justification required for using primates, but the numbers have still gone up significantly," said the society's Dr Penny Hawkins. "An increase in experiments on primates should not be acceptable in a humane society."
More than three-quarters of all the experiments are for research and drug development; safety testing accounts for much of the rest.
For the GM figures, the recorded procedures also count the breeding events that maintain the numbers of animals for labs to work on - not merely the scientific tests performed on those altered animals.
However, the way the annual statistics are compiled comes in for regular criticism by many groups, not just the welfare lobby.
The figures give little information on the real level of suffering experienced by the animals involved. They reveal only that just over a third required some form of anaesthesia.
After rodents, fish are the single biggest group to be experimented on
The stats also take no account of "wasted" animals - animals bred for their tissues and then discarded or animals rejected because their genetic modifications did not work.
If these were included in the annual statistics, the figures for animal use would be considerably higher.
The Animal Procedures Committee, the government's watchdog in this area, is asking "stakeholders" for their views on an improved statistics system.
In May, the government announced that it would establish a National Centre for the Replacement, Refinement and Reduction of Animals in Research.
The so-called 3 "R's" are supposed to underpin laboratory rules and culture.
"The statistics are useful, but they are only part of the picture," said Dr Vicky Robinson from the new national centre.
"For example they don't give us a measure of how animal research has actually been replaced through in-vitro techniques, computer modelling, the use of human volunteers, and so on.
"And neither do they tell us about how much the animals have suffered and what has been done to reduce that suffering."
A recent comparison of almost 3,000 research papers published over 30 years in major biomedical journals found a 30% fall in the number of studies using animals.
The analysis by a Swedish team suggested scientists were trying to use alternatives where possible, and when animals had to be used, to try to use them more efficiently.