Twin studies have suggested that there is a very significant genetic contribution to autism, with monozygotic (identical) twins sharing the disorder 92 percent of the time, compared to only 10 percent of other twins. The difference in rates also suggests that more than a single gene is responsible; otherwise, the 10 percent figure would be much higher. Indeed, a variety of studies have now identified individual genes that are associated with a tendency to autism. The new study expanded on this work, using a collection of over 900 families in which more than one child has been diagnosed with autism, provided by the Autism Genetic Resource Exchange. Results were also validated against an independent collection of 859 autistic individuals, and about 2,500 healthy controls.
The authors examined the DNA of these individuals, looking for what are termed copy number variations, or CNVs. CNVs can result from duplications, where an individual's chromosomes carry an extra copy of one portion of the genome, or deletions, where one of the two copies of a stretch of the genome is missing. Most individuals appear to carry a number of CNVs compared to the average human genome, so the authors focused on those that are more likely to have consequences: CNVs that include all or part of a gene.
They actually found quite a lot, even under fairly stringent conditions (at least three cases in two families in the experimental population). For deletions, they came up with 72 genes at 55 independent locations on the chromosome; duplications added 200 genes at 63 locations—the higher number of genes comes from the fact that duplications and deletions can affect more than one gene. A number of these genes had been identified previously as being associated with autism, and several of the genes showed up in the non-familial autism cases, both of which validated the approach.
But the big picture turns out to be in the little details. The absence of these CNVs in the unaffected population suggests that, individually, they're probably rare in the general populace. Some also may represent recent changes in the genome—the CNV in one family often wasn't identical to the one found in another, even when they contained the same gene. The other notable thing was that some of these CNVs showed up in unaffected children in the family pedigrees. So, there may be a steady background of new mutations that predispose individuals to autism, which explains its erratic appearance within families.
The authors note that "unaffected" siblings of those diagnosed with autism sometimes carry the same CNVs, but other research has indicated these individuals often have similar but less severe symptoms—"subclinical" in their words. This suggests that an individual CNV isn't enough to be causal. Instead, many may contribute, and the number and nature of genetic changes can combine with environmental influences to push individuals to different points along the spectrum of autism.
The other thing about the results that stuck out was the nature of the genes involved. The authors don't provide an exhaustive catalog, but there's a bias towards genes that mediate interactions among cells. These genes probably won't influence the basic structure of the brain—the right parts will be produced in the right places—but they may influence how those different parts establish connections among themselves, or how those connections are rearranged following learning or experience. A few of the other genes would appear to influence how nerve impulses are transmitted. Again, the impulses will probably still happen regardless of how many copies of these genes are present, but they may occur with different efficiency or dynamics.
Putting everything together, the results seem to point to a situation where genetics plays a big role in autism, but individual genes probably aren't that important. Instead, the genes found here seem to influence the function of the nervous system in subtle ways, and it's only the cumulative impact of these changes, and their interactions with environmental factors, that determine if and where an individual will fall within the autistic spectrum. That's not a convenient answer—we'll have a lot of work to do to sort things out further—but it may go a long way towards explaining some of the complexity of the disorder.