A Terrible Mystery

Thomas Insel spent years training as a psychiatrist in the 1970s, and in all that time he saw not one child with autism. In 1985, curiosity sent him searching; it took several phone calls to find a single patient. His only prior exposure to the disorder was a lecture in which Bruno Bettelheim "explained that it was due to evil mothers." The '70s were, he says, "an era of psychiatry that had no science."

Today's psychiatry has science--and it is science--and increasingly, it is offering hope for patients with autism. As director of the National Institute of Mental Health, Insel now heads an agency that funds autism research all over the nation and also conducts projects of its own. Thanks to revolutions in neuroscience and genetics, scientists are starting to unravel the shroud of mystery that has hung over autism since it was first described in 1943. But with each new discovery, more questions arise.

That includes the most fundamental question of all: what is autism? Although the basic symptoms are well defined, researchers are now trying to categorize the secondary ones, a suite so varied that Insel's colleagues have started referring to the disease as "autisms." Some children with the disorder never speak. Others "are so fluent that you can't shut them up," says Sarah Spence, a pediatric neurologist at the NIMH. About 20 percent of kids with autism hit early developmental milestones but regress around 15 to 18 months; the rest don't make it that far. What binds them all together is largely unclear.

But autism is known to be highly heritable, and last month, in what was viewed as a major breakthrough, Vanderbilt University's Pat Levitt identified the first common gene that plays a role. The MET gene helps build the brain in utero and in childhood. A faulty variant appears in 47 percent of the population, the vast majority of whom are healthy--but a child who carries that variant also carries more than double the risk of the disease. Another, rarer gene, also implicated in brain development, was identified in August, and mutations on almost every chromosome have been suggested as possible culprits, including some implicated in rare disorders related to autism, such as Rett's Disorder and Fragile X. "There are perhaps hundreds of different causes, and I think the field is finally coming to grips with that," says UCLA neurogeneticist Dan Geschwind.

The NIMH is also newly interested in environmental factors that might set off the disorder in patients who are already genetically prone to it. U.S. scientists recently teamed up with counterparts in Denmark and Norway to screen samples of blood and amniotic fluid for possible toxins. And research is underway to see if viruses might be involved.

Scientists have also recently found several regions of the brain that differ from the norm in patients with autism (graphic), but none of them appears to be the sole problem. The key to the disorder likely lies not in one region but in the way the brain is wired. Some researchers pin the problem on defects in the brain's decentralized "mirror neuron" network, which allows healthy people to feel empathy. Other work shows that distant parts of the autistic brain are connected by too few fibers, while areas close to each other are connected by too many, tangled in thickets of "white matter."