Imaging study suggests possibility of predicting autism by age 1

A standard type of brain imaging may be able to predict whether babies as young as one year are likely to develop autism, according to a new study that for the first time suggests the possibility of diagnosing the disorder before behavioral symptoms appear a year or two.

The research, published Wednesday in the journal Nature, is early, and the findings are based on a very small number of infants. At the moment, doctors don't know how they could use this early warning to prevent or lessen the developmental disorder.

But the findings, if confirmed, could eventually be highly significant, the authors said.

"The ability to accurately predict who will develop autism opens up tremendous new opportunities to develop effective therapies starting in the first year of life," said Robert T. Schultz, director of Children's Hospital of Philadelphia's Center for Autism Research, one of the study sites.

"Being able to identify those needing intervention before the full emergence of autism," said Schultz, a study coauthor, "offers hope for being able to blunt the development of autism and dramatically improve outcomes."

Autism Spectrum Disorder is a complex developmental disorder marked by difficulties in social interactions and verbal and nonverbal communication, and often repetitive behaviors.  Rates have risen dramatically in recent years, with an estimated one out of every 68 school-age children in the U.S. on the spectrum,  although it's unclear whether most or all of the rise  is the result of more awareness and testing.

Children who appear to be intellectually and behaviorally normal start developing symptoms around age two or slightly later. The cause is unknown, although researchers have been mapping where  the brain's development seems to go awry.

The new study focused on babies who are known to be at very high risk:  siblings of children already diagnosed on the spectrum have a one in five chance of developing the disorder.

The researchers used magnetic resonance imaging (MRI) technology to measure the babies' brains at six, 12, and 24 months of age.  As the brains developed, the overall volume, surface area and thickness of in certain regions could be examined.

A computer-generated algorithm predicted which of the infants would develop autism by age two. It was more than 90 percent accurate. A total of 106 high-risk infants and 42 at low risk were studied. All were given the MRI scans.

Previous research had identified enlarged brain size as a risk factor for autism. The imaging in this study found that the surface area of brains in babies who later developed autism had grown much faster and ended up  larger between six and 12 months of age than in babies who didn't show behavioral signs of autism at age two.  The study also showed that the rapid growth pattern originates in specific brain regions – particularly the cerebral cortex --  long before the  brain overall showed  notable enlargement.

The cerebral cortex, the outer layer of the brain, has undergone the biggest changes through humans' evolutionary history.

"Babies that went on to have autism by age two years showed faster expansion of the cortical surface area, particularly in regions of the brain for language, cognitive control, and social perception, such as understanding facial expressions," Schultz said.

More research is needed, he added, to determine if the study's methods can be used to accurately screen low-risk children, which would be the vast majority of cases.

Joseph Piven of the University of North Carolina-Chapel Hill, the study's lead author, said that in the larger field of neuroscience and treatment, the "big push" has been to find the biomarkers of conditions before patients are diagnosed, so that some preventive efforts may still be possible.

By the time autism is diagnosed now, the brain has already begun to change significantly, he said.

"We haven't had a way to detect the biomarkers of autism before the condition sets in and symptoms develop,"   Piven said  in a statement. "Now we have very promising leads that suggest this may in fact be possible."

If the diagnostic technique is eventually proven successful, the next question would be which babies should be scanned. Screening the entire population would be unaffordable.

Besides UNC and CHOP, other clinical sites involved in the study were the University of Washington in Seattle and Washington University in St. Louis. Other key collaborators were New York University, the College of Charleston, McGill University and the University of Alberta.