Friday, September 24, 2010

Genetic Mutations in Autism - Cause or Effect?

Whenever you read anything about autism you will quickly run into the belief that autism is mostly "genetic".  The idea is that something, somewhere is not quite right with a person's genetic code and it is this difference that leads to the symptoms that make up what we call autism.  This belief is largely based on twin studies that show that identical twins are more likely to both have autism than non-identical twins and based on findings that children whose siblings or parents have autism are more likely to have autism themselves.

Based on this theory, millions upon millions of dollars have been spent looking for the genetic "causes" of autism.  But even though a large amount of time and money has been spent looking, there has been very little success.  Many studies have found genetic mutations that appear in a very small number of cases - typically much less than 1 percent. But whenever other researchers look for the these mutations in other groups, they almost always come up empty.

What this leaves us with is a large number of genes that might be involved in some way shape or form with a very small number of cases.  If we add up all of these cases - and make the rather large assumption that they don't overlap - we might reach a total of 10 to 20 percent of all autism cases.  That means we would have 10 to 20 percent of cases made that might be caused by twenty to thirty distinct mutations.  And that is just talking about the mutations that involve a sizable population (even if it is under 1 percent of the whole).  Under this sort of model, the other cases would also have mutations but they could almost be unique to the individual case.

At the end of the day, we would be likely talking about a hundred or so unique mutations that could play a role in causing autism.

To make matters worse, even when we can find these mutations, rarely do we have any real understanding of how these mutations could lead to something like autism.  And then even in the extremely rare cases when we do understand what these mutations might be doing, we have no clue of what can be done to fix the problem.

When you consider the failure of science to identify the genetic underpinnings of autism - or even to find any underlying mechanism that could cause autism - you have to ask yourself whether the mostly genetic model of autism is right.  Is autism really caused by a huge number of almost individualized mutations or is something else going on?

I have a idea that I have been thinking about for a while now.  What if all of these rare genetic mutations that are found in autism are not the cause of autism but rather a side effect of autism or collateral damage from whatever process caused the autism?

I know that this is pure conjecture on my part, so take it for what it is worth.  But consider the following.

First, a large number of the genetic mutations that have been found to be associated with autism are de novo, meaning that the mutations are seen in the children but not in their parents.  While this does not mean that autism isn't genetic, it does put a rather large dent in the idea that autism runs in families. It also makes you wonder what is going on with twin studies, but I will get to that point in a minute.

Second, when you look at current estimates on how often identical twins both have autism, you will see that some of the earlier studies were overly optimistic.  More recent research is showing that identical twins will only both have autism 80 to 90 percent of the time and, even when they do both have autism, one can be more severe than the other.  While it is true that identical twins will start to differ over time on a genetic level, before the age of three there should not be any major genetic differences.  As a result, if autism were solely based on genetics this number would be a lot close to  100 percent and the severity would almost always be the same.

Third, when you look at fraternal twins, they are much more likely than typical siblings to both have autism.  Recent research has shown that male male fraternal twins have up to a 40 percent chance of both having autism while other types of siblings have only an 2 to 8 percent chance.

For those of you who don't understand the details of the different types of twinning, the simple way of thinking about it is as follows.  Identical twins alway start out as one fertilized egg - one organism - that splits into two separate organisms sometimes in the first nine days after fertilization.  Depending on how late the identical twins split they will share either a placenta, amniotic sac, or both.  If they split early enough then they will share nothing.  Fraternal twins are two fertilized eggs and are no more alike than other siblings would be.

The thing that all twins have in common is that they share, to some degree, the prenatal environment and tend to have very similar post-natal environments as well (just take my word for it).  Identical twins tend to share more of the prenatal environment while fraternal twins share less.  Normal siblings, of course, do not share any of the prenatal environment.

This leads me to my last, and mostly anecdotal, set of reasons why I think that the small mutations are an effect of autism rather than a cause.  For those of you who haven't been reading here for a while, two of my daughters with autism are identical twins.  While they are both about the same severity and share the same core deficiencies (primarily communications), it is striking how different their autism looks.  For example, Twin A has great imaginative play skills which Twin B lacks but Twin B is much more in tune socially with other children and adults than Twin A.

It is easy to tell when something about them is being driven by genetics because they are almost identical.  Even when you have something that you wouldn't think is genetic, they still stay very close to each other.  For example, they have had different eating habits for several years now but have stayed the same height and almost the same weight as each other for five years now.  This year is the first time that they differed by more than an ounce or two but they are still the same height.

What I find striking is how different their autism looks and the different areas that they have problems with. For example, Twin A does not like noise and gets upset when there is too much background noise.  But Twin B wants to have constant background noise and gets upset when there isn't enough.  It is almost like their autistic behaviors are more like personality traits rather than something predetermined by genetics.

More important, however, is the genetics of these two young girls.  Earlier this year we have a microarray test done on both of them to look for genetic mutations that could cause their autism.  We know that they are in fact identical twins because we had genetic testing done that confirmed this.  So imagine our surprise when the tests came back with different, non-overlapping mutations.

Twin A has two smaller copy number mutations (copy loss) located on two chromosomes while Twin B has one larger copy number mutation (copy loss) on a different chromosome.  All three mutations have appeared in various studies that looked for rare mutations in autism, although none of them was a major finding.

Let me say that again - identical twins who have different genetic mutations.  Clearly these mutations cannot have been inherited.

Just a few other, more technical points.  The twins were dichorionic-diamniotic twins which is just a fancy way of saying that they split relatively early (1 to 3 days) and had separate placentas and separate amniotic sacs.  In other words, they shared about the same amount of their prenatal environment as fraternal twins do.

Furthermore, their mutations do not show any evidence of being in only some of their cells not more others because the test we used would have picked up on that (ok, not strictly true, but true enough for this discussion).  This means that the mutations had to have happened very near to conception when there were very few cells so that the mutation would effect all of their cells.

So when you put these facts together you are left with the conclusion that these mutations happened at some point after they split off from each other (1 to 3 days) but early enough that whatever caused the mutations hit them both and could change most of their cells  (first 9 or so days).  I have no idea what that event was but I would guess that something strong enough to cause permanent genetic damage would also be able to cause autism.

So what do you think, do these mutations cause autism or does whatever causes autism also cause these mutations as a side effect?


  1. two issues here:
    1. From what I understand, one of the problems with genetics research in autism is lack of sample size. To date, it has been infeasible to get enough subjects in studies to find more genetic defects that can cause autism.

    2. Autism is probably a term to describe a variety of different conditions with somewhat similar (though not identical) phenotypes, so it is unlikely that one common pathway genetic type of cause will be found in autism. For example, I have above average or even superior skills (in vocabulary) that Stephen Wiltshire is deficient if not retarded in. Wiltshire is a savant in drawing. I have very serious fine motor impairments and am basically retarded in drawing and have printing that is not entirely legibile and is difficult for me in handwriting. I do poorly on block design and object assembly tests on wechsler that many others with ASD diagnosis do well on. Yet we have the same diagnosis, but clearly a different disease.

    It is possible you are correct about some cases of autism due to the phenomenon of epigenetics in that something could have happened to alter the genes that might be a side effect.

    I knew there was a large difference between fraternal twins and non twin siblings, but I was not aware that it was as large as you said.

    Some cases are inherited via autosomal dominence such as tuberous sclerosis. They have found abnormalities on chromosomes such as 22 such as in DiGeorge syndrome which is associated with autism. Some cases arise from spontaneous mutations. So you could be right in some instances about autism but not all.

  2. I agree with your observations.
    I still think vaccines are the culprits. The toxins in vaccines, particularly mercury, can cause mutations.

  3. It is a legitimate and an interesting question to ask.

    I ask another, whether all of these genetic studies are providing any useful information at all, or are they merely feeding an industry of academic publication and justifying various university departments existence in terms of the funding they attract?

    No that's two questions I will come in again :)

  4. Jonathan,

    I don't know whether sample size would play that much of a role. There have been several studies recently were hundreds of people with autism were involved and that should have been sufficient to find anything that a sizable percentage of them had.

    I would say that one possible confounder would be the idea that there are many different subtypes of autism and each would have its own genetics. It might be hard to see an effect if the each subgroup only made up ten to twenty percent of your sample.

    Which goes into your second point, would you expect the strengths and weaknesses of autism to be determined by genetics? My twins are very similar in some features of their autism but complete opposites in others. Assuming that they share the same underlying cause of autism, how and why could they differ? There is nothing about their individual mutations that could explain the differences.

    Why can one not tolerate noise while the other seeks it out? I realize that the behaviors are two sides of the same coin but presumably there are differences in how their brains work that are causing them to behave in opposite ways.

    The only explanation I can come up with is that the specifics of autism are not controlled by genetics and takes me to the idea that autism is an ongoing disruption like Fragile X and less like something like Downs. But unlike Fragile X, there is no known common genetic variant which suggests that the disruption isn't limited to a single biological process like it is in Fragile X.

    In a way this is good because it should be easier to correct problems that aren't written into the genes. But on the flip side it is worse in that we have to find the disruptions unique to many different subtypes of autism.

  5. Author,

    I think the genetic studies do serve a purpose, although part of that purpose is to feed the academic process. Publish or perish.

    Another purpose would be that researchers get to chance to play with some of the cool new scientific toys instead of the old boring ones.

    But seriously, these studies are showing that our ideas about autism are wrong. These studies are showing that autism is not primarily a genetic disorder and are pointing towards the fact that there is some sort of process, ongoing or one-time, that is "causing" autism.

    But since we have no real idea what that process is, we can either spend money on the psych 101 style research (how does looking at this picture of a cartoon face make you feel) or we can spend it on genetics and possibly get lucky when we find a biological pathway that is disrupted.

  6. It's an interesting idea, that the variations are chickens and not eggs... but I think Jonathan makes a good point about individual differences within ASD. As I'm sure you know, genes rarely influence behavior directly, they can only turn on and off production of proteins. One recent study demonstrated that the "high risk" variation in CNTNAP2 influenced the development of neural connectivity in the frontal lobe and other areas related to attention and processing. This gene has been associated with increased risk of autism (and other developmental and communication disorders), but subclinical symptoms also appear in individuals who have no diagnosis. This is likely to explain why some siblings have subclinical symptoms without meeting any of the other criteria.

    Anyway, I think this suggests that the influence of genes cannot be considered without considering the dynamic interplay between different genes, prenatal and postnatal environment, which includes exposure to teratogens and physical experiences that influence connectivity in the brain. Add additional complexity by introducing the variable of timing, and you've got a recipe for an infinite number of individual variations (even between otherwise genetically identical twins).

    What's awesome about this, as you suggested in your comments, is that thinking about this as a dynamic process, one can, in theory, be modified through well orchestrated experiences!

    (By the way, thank you for your blog. It's a fantastic combination of science and personal perspective.)

  7. Rebecca,

    Thanks, I'm happy to hear that you find my site interesting.

    Regarding the genetics of autism, I think it is safe to say that whatever is going on, the genetic pieces of autism are not going to be simple.

    One one hand you have these rare mutations that seem to give you a greater risk of autism and on the other you have recent results like this that seem to say that the problem might be more about the expression of the genes rather than the genes themselves.

    And then you have the idea that if there were a single mutation that caused the majority of the cases of autism it should have been found by now.

    Putting all of that together leads to one big, fat question mark.