The preliminary results of the bloodtest, described Dec. 5 in the online journal PLOS ONE, led toa diagnosis of autism in about two-thirds of patients before clearsymptoms begin to appear. Studies have shown the average age of anautism diagnosis is four-and-a-half years old, although parents beginsuspecting autism at an average age of 19 months.
"Since brain biopsy isn't a viableoption for research, we asked ourselves whether blood could serve asa proxy for gene expression in the brain," says Dr. Isaac Kohane,director of the Boston Children's Hospital Informatics Program(CHIP). "We found that it could, though we and others wereinitially skeptical."
Dr. Sek Won Kong, a CHIP investigator,and his research team analyzed blood samples from 66 male patientswith autism from Boston Children's and other hospitals and theAutism Consortium of Boston, then compared them with 33 age-matchedboys without autism. Using microarrays, they looked for RNAsignatures reflecting differences in gene activity or expressionbetween the two groups.
Analyzing the blood samples, Kong andcolleagues flagged 489 genes as having distinct expression patternsin the autism group, then narrowed this to a group of 55 genes thatcorrectly identified or ruled out autism in 76 percent of samples.They validated their findings in a second group of 104 male andfemale patients with autism and 82 controls, achieving an overallclassification accuracy of 68 percent (73 percent for males and 64percent for females).
The gene signature approach, whichBoston Children's Hospital has licensed exclusively to SynapDx ofSouthborough, Mass., can potentially diagnose autism far more oftenthan the genetic tests currently available. Those tests look forvariety of autism-related mutations—from small "spelling"changes to lost or extra copies of a gene or genes, or wholesalechromosome abnormalities. However, the known mutations account forfewer than 20 percent of autism cases.
Kohane stresses that the test was notintended as a general screening tool, but as an aid to help provideguidance after suspecting autism—and would not supplant behaviortesting.
"It's clear that no single mutationor even a single pathway is responsible for all cases," Kohanesays. "By looking at this 55-gene signature, which can capturedisruptions in multiple pathways at once, we can say with about 70percent accuracy, 'this child does not have autism,' or 'thischild could be at risk,' putting him at the head of the queue forearly intervention and evaluation. And we can do it relativelyinexpensively and quickly."
Kohane also believes autism can developin ways other than genetics. The 55 genes for which expression wasaltered also suggest more than one path to autism, he says. He andhis team found that based on their genetic signatures, subjects withautism clustered into four subgroups marked by changes in differentbiological pathways. Synaptic pathways, specifically long-termpotentiation pathways—which are essential for memory andlearning—and neurotrophic pathways signal neurons to survive,develop and grow.
Most current theories of autism focuson disordered synapses, says Kohane, speculating that braindevelopment in autism may be impaired by abnormal immune responses toinfections and other stressors prenatally or during infancy.
"In our sample, about half of theautism cases had some sort of alteration on immune pathways, synapticpathways or both," Kong states.
For example, chemokines (inflammatorychemical messengers that recruit immune cells to fight infections)have been found to be overexpressed in some neurocognitive disorders,and animal studies have found that prenatal infections are associatedwith autism-like features in offspring, notes Kohane.
"We know that if a mother hasrheumatoid arthritis or the father has type 1 diabetes, there is anincreased risk of autism in the child," Kohane says. "Studieshave also found an increase in autoimmune disease, just as autism isincreasing."
Kohane's next step in studying theefficacy of the RNA blood analysis will be to conduct a siblingstudy.
"Our research is of siblings in theBoston area to test whether the siblings of autism patients have ahigher tendency to develop autism," he says. "We know identicaltwins have a 70 to 90 percent of both being diagnosed with autism.There is believed to be a 20- to 30-percent chance of finding autismin a sibling of a child diagnosed with autism."
The results of this study will beavailable in 2013, he says.