U researchers, in national study, find a method of predicting autism

Tracking the brain growth of infants can predict the likelihood that they will be diagnosed with autism in their toddler years, according to new research that could give doctors a head start on treating the developmental disorder.

The study, published Wednesday in the prestigious journal Nature, took place at four U.S. hospitals and was co-authored by two University of Minnesota researchers.

Children with autism often show hints of the disorder by their first birthdays, such as failing to make eye contact or respond to their names. But studies haven't proven these early behaviors to be reliable predictors, said co-author Jed Elison, an assistant professor in the U's Institute of Child Development. Most children aren't diagnosed until age 2 or older, when clear behavioral signs emerge.

"What really differentiates this work [is] the accuracy with which we can make a prediction," Elison said. "We're generating a prediction before the signs of autism can be observed, which is really groundbreaking."

Autism, or autism spectrum disorder, refers to a broad range of brain development abnormalities that cause patients to struggle with learning, social interaction and verbal and nonverbal communication.

The U.S. Centers for Disease Control and Prevention estimate that one in 68 children meet criteria for ASD, a rate that has increased sharply in recent years due partly to rising prevalence but also to a broadening clinical definition of the disorder.

Studies have shown that earlier intervention can reduce autism's severity because of the "plasticity" of the brain at an early age, which can make it particularly receptive to cognitive and communication training, Elison said. "The earlier an intervention is implemented, the better the outcome for kids with autism ... This has been a finding since the '70s."

Identifying those physical differences correctly predicted 80 percent of the children in the second high-risk group who met the clinical criteria for autism.

Excessive growth in the size and surface area of the brain between 6 and 12 months of age proved to be a significant predictor, Elison said.

The study also found that the technique correctly predicted 89 percent of the children who didn't go on to develop autism. That finding is equally important because a test can't be used clinically if it frightens too many people who are healthy into believing they are sick.

Imaging scans for the study took place in other states including North Carolina, where Elison first became involved in 2007 as a graduate student and then became a co-investigator before his move to Minnesota.

Also participating was Jason Wolff, a U assistant professor of educational psychology. The study, Wolff said, "offers the unprecedented possibility of predicting whether or not a child will develop autism based on neurobiological data."

"If we diagnose and don't do anything, what's the point?" she asked.

Luskin also said she worries about diverting resources to costly MRI screenings, when doctors already underutilize cheaper behavior screenings that can be somewhat reliable starting at 18 months of age.

Elison has applied for a grant to conduct a new round of imaging studies in Minnesota because he said this initial finding needs to be replicated with different children and different MRI techniques.

The end result is unlikely to be universal MRI scans for children, because of the extraordinary costs involved, but Elison said he could one day see brain scans being routinely used on the siblings of children with autism.

The study was not without risks, primarily a concern that the infants might suffer hearing damage from the loud noise of MRI machines, Elison said. Babies underwent scans during naps, to make sure they remained still, and wore earplugs and headphones to protect their ears.