Genetics: Study ties three immune genes to autism
-
Immune interest: Some individuals with autism carry a particular common variant of CD99L2, which encodes a molecule involved in the movement of white blood cells.
Common variants in three genes involved in the immune system are more likely to crop up in people with autism than in typical controls, according to a study published 9 June in Molecular Autism1.
Many genetic and epidemiological studies in the past few years have found unexpected connections between psychiatric disorders and the immune system.
Large genetic screens have shown, for example, that people with schizophrenia tend to carry common glitches in the major histocompatibility complex, a genomic region that contains a slew of genes related to immune function. One report found that some autism risk genes belong to larger networks of genes related to cytokines, which transmit messages in the immune system.
In the new study, researchers used a bioinformatics database called Ingenuity Pathway Analysis to identify 2,012 genes with known roles in the immune system. They then searched for common variants in those genes in blood samples taken from 1,510 trios (two parents and a child with autism) from the Autism Genetic Resource Exchange, a genetic repository that includes families with at least two children who have autism.
Common variants in three genes — CD99L2, JARID2 and TPO — are significantly more common in the individuals with autism than in their unaffected parents, the study found.
All three genes have several biological roles, in the immune system and elsewhere. CD99L2 encodes a molecule involved in the movement of white blood cells during the process of tissue inflammation. JARID2, which helps remove methyl marks from histones, has previously been tied to schizophrenia2 and autism3. TPO encodes an enzyme important for normal function of the thyroid gland.
Interestingly, when looking at the set of immune system genes as a whole, the researchers found no statistically significant enrichment in the autism group compared with controls. This shows the value in investigating the associations of individual genes, they say.
References:
1: Ramos P.S. et al. Mol. Autism 3, 4 (2012) PubMed
2: Pedrosa E. et al. Am. J. Med. Genet. B Neuropsychiatr. Genet. 144B, 45-51 (2007) PubMed
3: Weiss L.A. et al. Nature 461, 802-808 (2009) PubMed
Comments
Didn't they teach you in med school that you cannot have a genetic epidemic. Until some real research is done on environmental factors that affect our health we will ignore this medical catastrophe. Recent studies have shown that when children 0-4 yrs old are exposed to two or more anesthetics they have an increase in developmental disabilities. Duh! Expose a young child to toxins, drugs, chemicals or other environmental factors and what do you get?
I have some fairly serious concerns about the statistical approach taken here. The direct test of the authors' hypothesis is looking to see if immune genes AS A CLASS are enriched for association - but that isn't the case. The fact that a handful of the variants are significant after correction for multiple testing is not particularly interesting - it would be possible to define the category "immune genes" in many different ways, such that at least one of those potential groupings provided a nominally statistically significant outcome.
But most seriously, the authors have made no attempt to see if their reported significant variants actually hold up in separate cohorts of people with autism. Without independent replication (a standard requirement for most modern genetic association studies) this study's findings are deeply suspect.
Overall this study provides no compelling support for an immune component to autism.
Thanks for your comments, Dan. These are all really important things to consider when evaluating autism genetic studies, especially as they become ever more popular.
From my reading of the study, it seemed that the authors defined "immune genes" according to that Ingenuity Pathway Analysis program, which is supposed to group genes by their known biological functions. But you don't buy into those categories? I ask because I'm pretty sure I've seen that program used in other studies, too.