Thursday, October 18, 2012

Genetics, Twins, and Autism

I think I may have mentioned this once or twice in the past but, at the risk of repeating myself, let me say it again.  I am somewhat skeptical about the idea that autism is a primarily genetic disorder consisting of hundreds of extremely rare mutations.

I understand the evidence that suggests that autism is "genetic".  I understand that twin studies suggest that something about being a twin greatly increases the risk of autism and that the risks are greater for identical than fraternal twins.  But what I don't understand is why the presumption is that genetics is the cause of the greater risk.  I don't understand why the other things that twins have in common - such as a shared prenatal and early childhood environments - are often ignored or overlooked.

For example, the existing evidence suggests that a fraternal twin has a greater risk of autism if their twin has autism than other, non-twin siblings do.  Fraternal twins are no closer genetically than other other siblings so the increased risk over non-twin siblings has to come from shared environmental factors.

And then there is the little fact that the genetics of even identical twins isn't as straight forward as you might think.  Yes identical twins start off with identical genetic material at the moment of conception, but after that point things can get a little murky.

As I think I have mentioned in the past, my identical twin daughters who both have autism each have their own set distinct set of mutations.  Under the standard assumption that identical twins are always genetically identical that shouldn't happen.  But clearly someone forgot to tell them that they shouldn't be like that.

 Just because identical twins start out with the same genetic code when they split from each other a few days after conception doesn't mean they will still be identical when they are born.  But if my little anecdote doesn't make you say hmm, then consider the following abstract of a study that was just published -

Monozygotic twins discordant for submicroscopic chromosomal anomalies in 2p25.3 region detected by array CGH.
Although discordant phenotypes in monozygotic twins with developmental disorder are not an exception, underlying genetic discordance is rarely reported. Here, we report on the clinical and cytogenetic details of 4-year-old female monozygotic twins with discordant phenotypes. Twin 1 exhibited global developmental delay, overweight and hyperactivity. Twin 2 had an autistic spectrum disorder. Molecular karyotyping in twin 1 identified a 2p25.3 deletion, further confirmed by FISH analysis on leukocytes. Interestingly, array-CGH was normal in twin 2 but FISH analysis using the same probe as twin 1 showed mosaicism with 1/3 of cells with a 2p25.3 deletion, 1/3 of cells with a 2p25.3 duplication, and 1/3 of normal cells. Genotyping with microsatellite markers confirmed the monozygosity of the twins. We propose that the chromosome imbalance may be due to a mitotic non-allelic recombination occurring during blastomeric divisions of a normal zygote. Such event will result in three distinct cell populations, whose proportion in each embryo formed after separation from the zygote may differ, leading to discordant chromosomal anomalies between twins. We also discuss that the MYTL1L and the SNTG2 genes within the reported region could probably relate to the phenotypic discordance of the monozygotic twins.
Makes perfect sense, right?

OK, the text is a little bit dense, so lets break it down a little bit.  In this study there is a pair of four year old identical twin girls who have different but related developmental disorders -

Twin 1 -
  • has global developmental delay (intellectual disability)
  • is overweight
  • is hyperactive
  • has a deletion at 2.25p3
Twin 2 -
  • has autism
  • did not have an overall mutation at 2.25p3
  • but did show signs of mosaicism, meaning that roughly -
    • 1/3 of her cells had a 2.25p3 deletion
    • 1/3 of her cells had a 2.25p3 duplication
    • 1/3 of her cells had a normal 2.25p3 gene
If that last bit about one twin having a deletion at 2p25.3 while the other has a partial deletion and a partial duplication of the same gene leaves you confused, you aren't alone.  I had to read the abstract a few times and look up what some of the terms meant before it started making sense to me.

I haven't read the full text of the study but there are a few things that jumped out at me.

First, genetics - even in the case of identical twins - is not simple.  Most people hear the word "identical" and jump to the conclusion that identical means exactly the same.  But as this abstract and other results have shown, identical twins aren't exactly the same.  You cannot assume that because one twin has a certain genetic mutation that her twin will have the same.

Second, just because you can point to developmental differences between identical twins and can find a mutation in one twin does not mean that the mutation is automatically the cause.  There has to be a mechanism related to the mutation that could cause the differences.

Third, when you find differences in the genetics of identical twins, the fact that there are differences might be more important than what the actual differences are.  Any mutation that is not shared has to happen after conception but before the fetus gets large enough that a mutation wouldn't be able to spread to the entire body.

Since these mutations cannot be inherited there has to be something in the prenatal environment that causes them (and please don't say random chance).  It seems, at least to me, that any factor that is strong enough to cause permanent genetic change would also be able to play havoc with the delicate developmental process of the fetus.

Finally, you have to look at the overall picture of the twins to see if the mutation is important.  In the study the twins both have different yet related developmental disorders - global developmental delay and autism.  Since these conditions are so closely related and since it can be difficult to tell the difference between the two (especially in four year olds), I think the fact that both twins are both developmentally delayed is more important than the subtle differences between the two.

The bottom line here is that presumption that autism is largely genetic rests heavily on twin studies but the genetics of twins isn't as straight forward as you would think.  When you add in the facts that there isn't an "autism gene" or even a small set of "autism genes" but rather hundreds of mutations that might each account for a very small number of cases and that even identical twins don't always share these rare yet presumed causal mutations, the genetic presumption starts to look a little weak.


Rio M, Royer G, Gobin S, de Blois M, Ozilou C, Bernheim A, Nizon M, Munnich A, Bonnefont JP, Romana S, Vekemans M, Malan CT. Monozygotic twins discordant for submicroscopic chromosomal anomalies in 2p25.3 region detected by array CGH. Clin Genet. 2012 Oct 15. doi: 10.1111/cge.12036. [Epub ahead of print] PubMed PMID: 23061379.


  1. Identical twins are not completely identical, e.g. they have different fingerprints. So, they are not exactly the same and discordance among twins, i.e. one autistic and one nonautistic can exist. There are cases in which autism is clearly genetic, such as tuberous sclerosis and fragile X and neuroligins. Other forms of autism may have a genetic component, but involve multiple genes so the exact etiology is complex. A number of mutations in autism are de novo, not inherited and can occur for various reasons including random chance.

    There are probably environmental factors interacting with genetic so that is the reason for higher concordance among fraternal twins versus siblings.

    This sort of muddies the hypothesis of temple grandin that autism genes contributed to every invention from the spear to the cellphone and Baron-Cohen's assertion that autism has stayed in the population due to evolutionary advantage. Wish I knew more and had read more than I could comment further, however, my own autistic disability makes this difficult.

  2. Hi Jonathan,

    Just a few follow ups to what you said.

    First, while it is certainly true that identical twins are different and that these differences become larger over time, the opposite is also true. The younger the twins are, the more identical they will be. If you take that logic to its conclusion then you get to the point where identical twins *should* be almost completely identical the instant after they separate from each other.

    Second, for a genetic change to be in present in most cells in the body, it has to happen extremely early in the developmental process. The obvious example is when the change is present in the genetic material from the mother or father, then the change will be in every cell in the body. But as the fetus grows it will be harder and harder for something to introduce a change that impacts most cells in the body.

    I am not an expert in this area so I might be getting some details wrong but it seems that the larger the organism, the stronger and more widespread the external factor is going to have to be to introduce a change. So while you wouldn't need that much force to cause a mutation in a two cell organism you would need an extremely strong force to cause a widespread mutation in a newborn baby. And I suspect that a force that strong would be likely to cause more serious problems than just a random mutation.

    Now when you apply that idea to identical twins you see the possible problems.

    If the twins do not share overlapping mutations then that means the mutations had to have occured after they split from each other but before they got large enough that such a widespread mutation is no longer possible.

    This is what had to have happened to my daughters because their mutations do not overlap. They probably split from each other some time in the first three days and after that point something introduced their mutations. I don't think we can call their mutations "random" unless the same "random" thing happened to both of them.

    Another scenario is the one that happened to the twins in the study. Something introduced a change after conception but before they split. However the change did not impact all of the cells equally - some of the cells were normal, some had a deletion, and others had a duplication. And when the twins split from each other one was left with cells that had a deletion while the other had cells that had all three properties.

    In both scenarios, the mutations were caused after conception and both twins were impacted because they shared a prenatal environment. So my point is that the higher concordance among twins could just as easily be caused by a shared prenatal environment (both in the wrong place at the wrong time) as it could be caused by some shared genetic predisposition.