Julia Yellow

A panel of genes expressed in blood can distinguish toddlers who will later be diagnosed with autism from those without the disorder 83 percent of the time.

The finding, reported 4 March in JAMA Psychiatry, suggests that signaling pathways involved in inflammation, immune responses and protein synthesis are at play in autism¹. It also hints at the still-distant possibility of a blood test for the disorder.

“This is not the test, but it says that such a test seems feasible now,” says lead researcher Eric Courchesne, professor of neurosciences at the University of California, San Diego.

Others are more equivocal about the promise of the approach for predicting autism. “This study tells us that there’s a real signal there, but by itself it may not be good enough to make a diagnosis,” says Isaac Kohane, professor of pediatrics and health sciences technology at Boston Children’s Hospital, who was not involved in the study. “I think the most important contribution of this study is to cause us to focus even more tightly on these particular sets of pathways, to understand what’s causing them.”

Clinicians currently rely on a battery of behavioral evaluations to diagnose autism. But the average age of diagnosis hovers stubbornly around 4 years, long after symptoms emerge and behavioral interventions are believed to be most effective.

To close this gap, Courchesne and others are looking for biomarkers that can identify individuals with the disorder before symptoms emerge.

“What we’re looking for is something that might be an earlier, more objective indicator,” Courchesne says. “It’s a whole line of research aimed at identifying babies at risk for autism to get them into evaluation and treatment at early ages.”

Gene screen:

Courchesne and his team measured gene expression in blood samples from 142 boys between the ages of 1 and 4. The researchers selected the boys from community pediatric clinics that had flagged them using the Infant-Toddler Checklist — a 24-item screening test for developmental disorders.

Of the 142 boys, 87 were later diagnosed with autism. The remaining 55, including 14 with developmental delays, made up the control group.

The researchers found that 2,765 genes were expressed at different levels in the boys with autism than in the controls. The genes cluster into 12 distinct networks, dubbed ‘modules.’ Each module consists of genes that are turned on or off at the same time and tend to have a shared biological function.

Expression levels of four modules comprising a total of 762 genes predict which toddlers will receive an autism diagnosis with 83 percent accuracy, the researchers found. In a separate group of 44 boys with autism and 29 controls (including 4 with developmental delays), the panel predicted autism status with 75 percent accuracy.

Accuracy encompasses two important metrics: sensitivity, or the test’s ability to correctly identify children who have autism, and specificity, its ability to pinpoint those who do not have it.

The panel’s sensitivity — as measured in the separate group of 73 boys — is only 77 percent, meaning it misses 23 percent of cases. Its ‘specificity’ is 72 percent, so it incorrectly flags 28 percent of children.

Those figures are too low for the test to be of practical use, experts say, especially for screening large populations.

“The real question is: Can you get the sensitivity and specificity to the point where it’s useful clinically?” says Dan Arking, associate professor of genetic medicine at Johns Hopkins University in Baltimore, Maryland, who was not involved in the research.

The panel also contains many genes, making it too expensive for a diagnostic test, says Valerie Hu, professor of biochemistry and molecular biology at George Washington University in Washington, D.C., who was not involved in the study.

Beyond its predictive potential, the profile implicates protein synthesis, immune response and inflammation in autism. It is “pulling out common, shared pathways that are disrupted,” Courchesne says.

The new study is not the first to look for blood-based biomarkers, though its participants are among the youngest ever studied. The largest previous study, led by Kohane and published in 2012, included 170 children with autism and 115 controls between the ages of 4 and 13.

The two studies flag few of the same genes, however, a difference that might result from demographics. Courchesne’s team focused on young boys, whereas the earlier study included older children of both genders. Age and gender are known to influence gene expression patterns.

Perhaps more importantly, the two studies analyzed their data differently. Kohane’s study distinguished cases from controls based on the expression of individual genes. But finding reliable differences in single genes is difficult in such small samples, says Arking.

By contrast, Courchesne’s study separated the groups using expression of gene networks.

Both studies, however, point to pathways involved in inflammation. “That seems to be a recurring theme emerging in autism,” says Arking, whose work in brain tissue also implicates inflammatory pathways.

References:

1. Pramparo T. et al. JAMA Psychiatry 72, 386-394 (2015) PubMed