The mercury preservative used in some vaccines can cause behavioral abnormalities in newborn mice characteristic of autism, but only in mice with a specific genetic susceptibility, Columbia University researchers report today.
The findings challenge the results of several large studies on autism and bolster the fears of parents who have long believed their children were harmed by the vaccines.
The fact that the preservative, called thimerosal, had an effect on only one strain of mice could explain why researchers had found it so difficult to prove or disprove a link to autism.
"The exciting thing is that this gives us a way forward in understanding why we have not seen more conclusive findings on either side of the fence, and how we need to design studies to pick up gene-environment interactions," said Ellen Silbergeld of the Johns Hopkins School of Public Health, who was not involved in the study.
"I believe this has enormous implications for public health," said Dr. Julio Licinio of UCLA, editor of the journal Molecular Psychiatry, where the report is appearing.
"Showing that genetic background impacts on the outcome of thimerosal exposure is a major breakthrough."
He added that the study clearly showed that there was a link between vaccines and autism "for some groups and not for others."
An Institute of Medicine report released last month concluded that there was no evidence to support a link and suggested that researchers study other possible causes.
Dr. Steven Goodman of the Johns Hopkins School of Medicine, a member of the commission that prepared the report, said those on the commission were aware of the research.
"It's a tantalizing little piece of evidence that requires a lot more work" to overturn the "tremendous amount of human work that doesn't find a clue of a connection," he said.
The researchers have not yet identified the human analog of the mouse gene or genes that confer susceptibility to the effects of thimerosal, so it is not clear what proportion of children could be at risk from vaccinations containing the preservative.
What they do know is that the genes are involved in the immune system and that they make the mice more vulnerable to autoimmune diseases. Researchers already know that as many as a third of families with an autistic child have a history of autoimmune problems.
The researchers do not believe that all cases of autism -- or even a majority of them -- are caused by vaccines, said Dr. Mady Hornig of Columbia, the lead author. "Autism is a constellation of syndromes that almost certainly has many different causes," she said.
But the link to thimerosal may help explain recent increases in the incidence of the disorder, she said.
Thimerosal, which contains ethyl mercury, has long been used as a preservative in vaccines. Critics contend it became a problem in the 1970s, when the number of vaccines given to children increased sharply.
Since 1999, it has been removed from most of the vaccines routinely recommended for infants and children. It is still used in injectable influenza vaccine, though some thimerosal-free flu vaccine is expected to be available this year.
Autism is a severe developmental disorder in which children seem isolated from the world around them.
There is a broad spectrum of symptoms, but the disorder is marked by poor language skills and an inability to handle social relations.
No cure exists, but many problems can be alleviated with intensive behavioral therapy.
Between 1975 and 1985, studies showed the U.S. rate of autism to be about four cases per 10,000. Between 1985 and 1995, the numbers tripled to 12 per 10,000. But researchers now think the actual rate may be much higher, on the order of 20 cases per 10,000.
Several epidemiological studies have failed to find a link between vaccines and the increase in autism, and laboratory studies in mice and other animals have also failed to show a connection.
But researchers may have simply looked at the wrong animals, said Dr. W. Ian Lipkin, in whose laboratory the new work was carried out.
Hornig and her colleagues studied four strains of mice, including one strain -- called SJL/L -- in which mercury had previously been shown to stimulate autoimmune disorders.
Newborn mice of each strain were injected with either thimerosal or a thimerosal-vaccine combination at ages corresponding to those when human infants are typically immunized.
The doses of mercury were also comparable to those used in humans.
The three strains of mice with no autoimmune susceptibility showed no effects from either type of inoculation.
But virtually all of the SJL/L mice developed a variety of problems, including delayed growth, abnormal response to novel environments, decreased exploration of their environments, abnormalities in brain architecture and increased brain size.
All of those are typical of children with autism, Hornig said.
"This is clearly showing that there is an interaction of genes with the environment," said Dr. Daniel H. Geschwind of UCLA, who had been looking for genetic causes of autism and was not involved with the Columbia study. "The strain difference is ... quite fascinating. This will clearly rev the debate [about vaccines] up again."
The researchers are now following up on these findings by trying to determine what other genes, if any, may be involved in the mercury susceptibility.
They are also working with researchers at Brigham Young University to try to find families with a genetic defect comparable to that observed in the SJL/L mice to determine whether they have a higher risk of autism.