Their findings, published today, offer both good and bad news in the war against HIV. The study of 41 apparent cases of "transient" HIV infection indicates that transmission of the virus from mother to child occurs at a slightly lower rate than estimated - as the researchers reckon that the children were never in fact infected.
Their analysis, based on detailed genetic examinations of the samples, shows that the positive results were instead caused by mislabelling or by laboratory contamination.
However, the work also squashes the idea that some virologists had been developing, which was that something in children's metabolisms might be able to defeat the virus, which gradually overwhelms the body's immune system. Using that theory as a basis, they suggested, could lead to a cure for the disease.
"I don't think people were hanging their hopes on this being the answer," said Clive Loveday, professor of retrovirology at the Royal Free Hospital in London.
He was not involved in the new study, published today in the journal Science. But he commented: "I've never seen any proof of such 'transient' cases, and there's been no real explanation of why the virus should disappear."
The work shows again that where HIV is concerned, laboratory work is occasionally suspect. In 1995 The Independent revealed that the claim that a Manchester man in 1954 was the "first case" of Aids was false, because the version of the virus found was a modern mutation. That was subsequently blamed on contaminated instruments in the pathology laboratory.
Today's finding offers slim hope for HIV-infected mothers. Roughly 14 per cent of babies born to HIV-infected mothers are themselves infected. But in 1995, a paper published in the New England Journal of Medicine claimed that in a number of cases, HIV-positive babies later tested HIV- negative.
The new paper re-examines 41 cases of "transient" infection which turned up in five separate studies of 1,561 HIV-infected infants and their mothers. By examining the genetic components of the HIV samples in detail, it shows that there are no cases where the virus identified in mother and child is unambiguously identical - as it should be if it was transmitted through the blood.
Instead, the scientists think that the cause is either mislabelling in the laboratory - so that HIV-positive samples taken from others were mistakenly allocated to a child - or contamination when the "amplification" process, used to identify the genetic material of HIV, was carried out.
DNA amplification is one of the most powerful diagnostic methods available to scientists performing gene testing, but it is so sensitive that it is easily contaminated by tiny amounts carried in the air or on laboratory instruments. That could give rise to "false positives". But when the child was examined subsequently, no HIV would be found - giving the appearance of "transient" infection.