Here's a great article from the NY Times: 

http://www.nytimes.com/2012/08/26/opinion/sunday/immune-disorders-and-autism.html?_r=1&src=me&ref=general 

So here’s the short of it: At least a subset of autism — perhaps one-third, and very likely more — looks like a type of inflammatory disease. And it begins in the womb.It starts with what scientists call immune dysregulation. Ideally, your immune system should operate like an enlightened action hero, meting out inflammation precisely, accurately and with deadly force when necessary, but then quickly returning to a Zen-like calm. Doing so requires an optimal balance of pro- and anti-inflammatory muscle.In autistic individuals, the immune system fails at this balancing act. Inflammatory signals dominate. Anti-inflammatory ones are inadequate. A state of chronic activation prevails. And the more skewed toward inflammation, the more acute the autistic symptoms.Nowhere are the consequences of this dysregulation more evident than in the autistic brain. Spidery cells that help maintain neurons — called astroglia and microglia — are enlarged from chronic activation. Pro-inflammatory signaling molecules abound. Genes involved ininflammation are switched on.These findings are important for many reasons, but perhaps the most noteworthy is that they provide evidence of an abnormal, continuing biological process. That means that there is finally a therapeutic target for a disorder defined by behavioral criteria like social impairments, difficulty communicating and repetitive behaviors.But how to address it, and where to begin? That question has led scientists to the womb. A population-wide study from Denmark spanning two decades of births indicates that infection during pregnancy increases the risk of autism in the child. Hospitalization for a viral infection, like the flu, during the first trimester of pregnancy triples the odds.Bacterial infection, including of the urinary tract, during the second trimester increases chances by 40 percent.The lesson here isn’t necessarily that viruses and bacteria directly damage the fetus. Rather, the mother’s attempt to repel invaders — her inflammatory response — seems at fault. Research by Paul Patterson, an expert in neuroimmunity at Caltech, demonstrates this important principle. Inflaming pregnant mice artificially — without a living infective agent — prompts behavioral problems in the young. In this model, autism results from collateral damage. It’s an unintended consequence of self-defense during pregnancy.Yet to blame infections for the autism epidemic is folly. First, in the broadest sense, the epidemiology doesn’t jibe. Leo Kanner first described infantile autism in 1943. Diagnoses have increased tenfold, although a careful assessment suggests that the true increase in incidences is less than half that. But in that same period, viral and bacterial infections have generally declined. By many measures, we’re more infection-free than ever before in human history.Better clues to the causes of the autism phenomenon come from parallel “epidemics.” The prevalence of inflammatory diseases in general has increased significantly in the past 60 years. As a group, they include asthma, now estimated to affect 1 in 10 children — at least double the prevalence of 1980 — and autoimmune disorders, which afflict 1 in 20.Both are linked to autism, especially in the mother. One large Danish study, which included nearly 700,000 births over a decade, found that a mother’s rheumatoid arthritis, a degenerative disease of the joints, elevated a child’s risk of autism by 80 percent. Herceliac disease, an inflammatory disease prompted by proteins in wheat and other grains, increased it 350 percent. Genetic studies tell a similar tale. Gene variants associated with autoimmune disease — genes of the immune system — also increase the risk of autism, especially when they occur in the mother.In some cases, scientists even see a misguided immune response in action. Mothers of autistic children often have unique antibodies that bind to fetal brain proteins. A few years back, scientists at the MIND Institute, a research center for neurodevelopmental disorders at the University of California, Davis, injected these antibodies into pregnant macaques. (Control animals got antibodies from mothers of typical children.) Animals whose mothers received “autistic” antibodies displayed repetitive behavior. They had trouble socializing with others in the troop. In this model, autism results from an attack on the developing fetus.