Genetic signatures yield a blood test and back a neuro-immune view of autism

A “heat map” for autism gene expression (click to enlarge). Each row represents one of the 55 genes differently expressed in ASD patients vs. controls; columns show expression profiles for each of the 99 subjects. Genes in red have relatively increased gene activity; green, reduced activity. The bars along the bottom show how ASD patients vs. controls are distributed; overall, the ASD group has more genes over-expressed, while the control group has more down-regulated. The brackets at left connect genes that tend to be expressed together, while those along the top link individuals with similar gene expression patterns.

Though autism can respond well to early behavioral interventions, it’s typically not diagnosed in the U.S. until around age 5, when these interventions are less effective. Autism is diagnosed based on a child’s behaviors and language, which take time to develop to the point where clinicians can reliably assess them. What’s really needed is a fast, objective test when a child is much younger, before symptoms even show up.

In the past decade, researchers have chipped away at the problem, linking more than a dozen genetic mutations to autism—from small DNA “spelling” changes to lost or extra copies of a gene or genes (known as copy number variants) to wholesale chromosome abnormalities. Tests have been created, such as the chromosomal microarray test. But together, the known mutations account for, at best, 1 in 5 autism cases among tested patients.

About seven years ago, Isaac Kohane, MD, PhD, director of the Boston Children’s Hospital Informatics Program (CHIP), and Louis Kunkel, PhD, director of the Program in Genomics at Boston Children’s, became interested in differences in gene activity from tissue to tissue. While the DNA in our genetic code is hard-wired, not all our genes are always turned on in all our cells. They’re expressed selectively, based on what job the cell has to do, but also on factors in the cell’s environment.

Kohane and Kunkel had been working together to discover hidden genetic contributors to degenerative muscular disorders, studying not DNA but RNA—the code that carries information from the DNA and directs production of the specified proteins.


Around this time, Leonard Rappaport, MD, MS, chief of the Division of Developmental Medicine, approached Kunkel and Kohane and suggested that they look at gene expression in autism spectrum disorders (ASDs), pairing their genomics and bioinformatics expertise. Brain biopsies seemed out of the question, but Rappaport wondered whether patterns could be picked up in blood samples from children with ASDs that might reflect abnormalities in brain pathways.

Kohane and Kunkel were initially skeptical. But there was reason to speculate that gene expression in the blood might echo that in the brain, however distantly. “There are lots of genes that play multiple purposes in different organs,” says Kohane.


So they decided to give it a go. Kunkel provided the “wet” lab, where, led by Christin Collin, PhD, patients’ white blood cells were processed and the RNA extracted and loaded onto gene chips—microarrays capable of measuring the activity of 30,000 genes at once. Kohane and colleague Sek Won Kong, MD, crunched the data, comparing blood samples from 66 male patients with ASDs and 33 age-matched boys without ASDs.

On a first pass, the researchers flagged 489 genes as having distinct expression patterns in the ASD group. Eventually, they honed this to a group of 55 genes that correctly identified or ruled out autism in 76 percent of samples. To check their findings, they applied the 55-gene profile to a second group of 104 patients with ASDs and 82 controls, both male and female. The overall classification accuracy was 68 percent (73 percent for males, 64 percent for females).

Published earlier this month in PLOS ONE, this was the largest gene-chip study ever done in autism and one of the few with a well-matched control group.

“By looking at this 55-gene signature, which can capture disruptions in multiple biological pathways at once, we can say with about 70 percent accuracy, ‘this child does not have autism,’ or ‘this child could be at risk,’ putting him at the head of the queue for early intervention and evaluation,” says Kohane. “And we can do it relatively inexpensively and quickly.”

A neuro-immune view of autism?

Aside from a new testing approach, licensed and being developed by SynapDx (Southborough, Mass.), the gene expression findings shed some interesting light on autism.

(Click to enlarge)

Four groups of autism patients emerged. Some—the red dots in quadrant II in the diagram at right—had differences in biological pathways involved in maintaining synapses, or connections between neurons, already recognized as being a factor in autism. But others, those in quadrant IV, instead had alterations of immunologic and inflammatory pathways. Those in quadrant I had a double hit, and those in quadrant III showed no difference from controls on either pathway. (The blue dots are neurotypical controls.)

One thing is clear: There’s no single pathway to autism. The researchers speculate that for some children, autism may result from abnormal immune responses that impair brain development. There’s support for this idea in the medical literature, and certain immune aberrations have already been reported in autistic children.

In animal models, prenatal infections have been associated with autism-like features in offspring. Studies of brain tissue from patients with autism have found significant activation of immune cells in the cerebral cortex and cerebellum, and recent research at Boston Children’s Hospital has shown that microglia and the complement system—elements of the immune system—also play a role in the brain’s development and functioning. Epidemiologic evidence points to a relationship between autism and familial autoimmune diseases like type 1 diabetes, arthritis and celiac disease.

Looking at gene expression, rather than DNA itself, offers a window into possible environmental factors in some children with autism—factors that may switch genes on or off and take a susceptible brain down a different developmental path.

“There might be something intrinsic to the brain that causes it to respond differently to a perturbation in the environment, generating an inflammatory reaction,” Kohane speculates. “Your genes set you up, but there’s some environmental thing that triggers it.”

A study nearing completion in Kunkel’s lab may shed additional light: It’s using RNA sequencing, a newer, more fine-grained approach that can not only tell whether a gene is on or off, but quantify its level of expression. It’s also using looking at a different group of children with autism—those with neurotypical siblings and parents, who may have new mutations that weren’t inherited. “The samples in the PLOS ONE study were from children who often had an affected sibling,” says Kunkel. “The new study will allow us to compare gene expression in these two autism populations.”