We do not currently know. Autism is usually classified as a severe form of 1 of the 5 so-called pervasive developmental disorders (PDDs). Children who display milder symptoms may have an autism spectrum disorder (ASD). Asperger syndrome is often separated from classic autism because there is usually no delay in language development. These terms are frequently used interchangeably, unfortunately, which reflects the fluid nature of the diagnostic observations. One of my favorite categories decries a form of diagnostic surrender: PDD, not otherwise specified.
(Just so you know, terms like these drive behavioral geneticists nuts!)
But there are genes
All this has not stopped us from researching the phenomenon, of course. And the results after years of looking are clear: there is a tantalizing and substantial genetic component to the disorder regardless of how it is classified.
Initial studies that confirmed a genetic role came from the traditional family and twin heritability studies, some of which are now decades old. Some of the best recent work comes from assessing sibling recurrence risk. Usually described as a percentage, sibling recurrence risk is the formal probability that a younger sibling of a child with autism will also have the disorder. When autism is defined narrowly, the normal rate in unrelated populations is about 1 affected child per 500 (0.002%). When you look at sibling recurrence risk, the rate rises to about 1 in 6 (15%). Thus, there is ample reason to pursue genetic research in this area.
But that is where the easy stuff ends. Although many genes over the years have been nominated as the source of the behavioral anomalies, few studies have been successfully replicated. Specific chromosomal inversions, large deletions, chromo- somal translocations, and changes in copy number of individual genes (Figure) have been observed as risk factors for autism. In fact, you can practically name any type of mutation and find that it has been associated in the past decade, at least to some extent, with some part of the autism spectrum.
It is now clear that multiple genes expressed in specific combinations are involved differently in creating specific autistic behavioral profiles. It is also clear that wide nucleotide variations within these candidate genes exist that are undoubtedly more capable of predicting discrete autistic behaviors than others.
Looking at first cousins
Given the large number of potential genes in autism, one might expect that the research method of choice would include the deployment of large, Deadliest Catch–like gene fishing protocols. Recent progress has indeed been made using one of these larger screening technologies, a technique called homozygous mapping. What follows is a brief description of the technique.
Human genetic disorders that have complex, multigene origins have 2 overall causes. The first arises from the random roll of the meiotic dice—presenting cases that show no pattern of previous inheritance. In rare cases, heritable forms of what appears to be the same disease also exist. Mutations in these patients clearly show a pattern that can be transferred from one generation to the next. Homozygous mapping is capable of identifying these rare, heritable (invariably recessive) disease forms. The technology takes advantage of the presence of consanguineous families . . . which should probably be explained before we go further.