Sunday, July 10, 2011

Autism in Twins : The World Isn't Flat

Earlier this month a paper1 was published that took (yet) another look at autism in twins and found that autism might not be as "genetic" as once thought.

In a nutshell, this paper found that the number of identical twins who both have autism is lower and the number of fraternal twins who both have autism is significantly higher than once thought.  When you consider the findings separately or together, it all points to the fact that environmental factors play a far larger role in autism than was previously thought.

I think the most surprising thing about this study for me is how many people seem to be taken off guard by this finding.  This isn't the first recent twin study to show at least some of these results.  In fact, I wrote about one that suggested that autism was far more common in fraternal twins than was thought almost two years ago.

But even ignoring recent twin studies, the purely genetic model has, even after decades of searching, failed to find any mutations that could account for more than an minority of cases of autism.  As each passing study not only failed to find any single mutation that could account for more than a small handful of case but also (mostly) failed to confirm the small mutations from the last study, it became clear (at least to me) that genetics alone could not determine whether someone developed autism.

Sure, you can take all of the prior studies together, put on your optimistic glasses, and say that you might be able to point to the genetic "cause" for up to 30% of all cases2 of autism.  But that presumes that each of the individual "causes" that have been identified are in fact causes and not simply random chance.  It also assumes that a given person would only have one of these "causes".  Since most of these "causes" have been identified in only an extremely small group of people, I think it is far more likely that most of these mutations are simply chance and have nothing to do causing autism.

But, for the sake of argument, lets say that you could take a random group of 100 people with autism and find the genetic "cause" in 30 of them.  That would still leave the overwhelming majority of cases without a known cause.  What does it say when about the correctness of a theory when it fails to account for the majority of cases?

In most of science, if you had a hypothesis that has failed to be confirmed by hundreds of studies over several decades, that would be called a failed hypothesis.  But when it comes to autism, that is called the prevailing theory.  And the reason that this theory has survived is the results of past twin studies that found that autism seemed to be mostly genetic.

Enter the current study that very meticulously shows that identical twins often don't share their autism and that fraternal twins do so more than other siblings.  These findings have effectively taken out the foundation that the genetic only theory of autism was based on.  So when you consider that not only has the genetic theory failed to find any major results but also that it might rest on a flawed foundation, I think it is very clear that the theory is dead in the water.

Now, before anyone takes this the wrong way, it is still obvious that there are genetic causes of autism and that genetics does still play a large role in developing autism.  However, we can now say (with an almost certainty) that environmental factors also play a large role.

The world of autism is no longer flat.


1. Hallmayer, Joachim et al. 2011. “Genetic Heritability and Shared Environmental Factors Among Twin Pairs With Autism.” Archives of general psychiatry 1-8.
DOI : 10.1001/archgenpsychiatry.2011.76

2. Schaaf, Christian P, and Huda Y Zoghbi. 2011. “Solving the autism puzzle a few pieces at a time.” Neuron 70:806-8.
DOI : 10.1016/j.neuron.2011.05.025


  1. You seem to suggest there may be some cases that are purely genetic and some that are purely environmental. I think it's fair to say that all cases are a combination of both. Some genes carry a risk for autism but they don't make it inevitable. For example, lots of people with Fragile X meet criteria for autism, but many don't.

    For a thorough critique of this overhyped paper, I'd recommend Kevin Mitchell:

  2. Hi Jon,

    Actually I wasn't trying to saw that it was an either or scenario. I am sure that there are some cases that are purely genetic, some cases that are purely environmental, and the rest (probably the majority) falling somewhere in between.

    What I was trying to get across, perhaps poorly, that the idea that autism is predominately a genetic condition is most likely wrong.

    I have to wonder about the idea that the lack of inevitability implies an environmental element. It depends on just how you defined environment, I suppose.

    To me it seems that simple random chance during development could dictate the appearance of autism without any special environmental trigger.

    Whenever you have a genetic condition that influences the body's biological makeup it seems like the wrong chemical being out of balance at the wrong moment during what would otherwise be "normal" development could push someone into id plus autism or just id.

    But then again, if you look at the actual symptoms of autism in identical twins who seem to have the same "cause" of autism (such as my daughters), you would see that there can be substantial variation between them. Maybe that would be the "environmental" element that you are referring to?

    Although if you go down that road to the ultimate end, there is no condition that is purely genetic.

  3. Here is an interview with an autism expert discussing the findings of the study... interestin­g stuff and it's certainly food for thought for pregnant women and those planning to become pregnant, particularly if they plan to use fertility treatments...

  4. Jon,

    I read Kevin's critique of the paper and I think he has some good ideas.

    But I also think he is neglecting the difference in co-morbidity that this study shows between DZ twins and non-twin siblings. The differences between DZ twins and non-twin siblings almost has to come from the shared environment of twinning (or possibly other twin risk factors).

    I left a comment over at the other site if you were interested in a better explanation of that point.

  5. Rubella and valproic acid are two proven environmental causes that are seldom brought up.

    Is it genetic susceptability that makes the fetus vulnerable to these teratogens?

    My son (adopted) says he feels it is genetic. Just an aside.

  6. I am happy to see that Jon Brock now is clear in saying that all cases of autism are a combination of genetic and environmental risks.

    He has not always been so clear about the environmental risk component of autism disorders as indicated in his Cracking the Enigma blog post of November 13 2010 "Genes for autism or genes for connectivity?'. In that post Jon Brock stated:

    "Autism is a genetic disorder. We've known this ever since the 1970s when studies by Susan Folstein and Michael Rutter showed that genetically identical twins are much more likely to both be autistic than non-identical twins. These findings were incredibly important at the time and fundamentally changed the way people think about autism. But they didn't tell us which genes cause autism or, perhaps more importantly, how they do it."

    Nowhere in the comment did Jon Brock indicate that environmental factors played, or might play, a role in causing autism.

  7. Excuse the brevity but...

    "So when you consider that not only has the genetic theory failed to find any major results but also that it might rest on a flawed foundation, I think it is very clear that the theory is dead in the water."

    " However, we can now say (with an almost certainty) that environmental factors also play a large role."

  8. Hi Anonymous,

    The simple answer to your why question is that autism is presumed to be a strongly "genetic" (think inherited from the parents) because of studies of autism in twins. But since the latest research is showing what we thought about twins and autism might not be true, that suggests that the earlier conclusion (autism is "genetic") might not be true either. And when you add to that the fact that researchers can't find any common genetic modification in more than a very small subset of cases (under 1 percent), I think it becomes obvious that there is not some common inherited genetic mutation that causes autism.

    That doesn't mean that genetic or epigenetic modifications don't play a role in autism. But it does seem to suggest that the modifications are not coming from the parents. So either the sperm or egg develop the mutation because of some external factor or something happens during the prenatal period that causes the mutation.

    Articles on CNVs in identical twins like the you quoted seems to imply the latter, that something happens to the child during pregnancy. Because identical twins always start out with identical genetic material, any mutation that one has but the other doesn't has to caused by something in the environment.

    The flip side here is that if one identical twin has a condition such as autism and has a mutation, you can't assume that the mutation is responsible for the condition. My identical twin daughters who both have autism have different CNVs from each other. That suggests (to me at least) that the CNVs might be a side effect of whatever is "causing" their autism. Or that the CNVs happen all of the time and we just don't know it.

  9. I think in your initial blog and this reply to my questions there are a couple statements that could be termed underdetermined and somewhat of an unwarranted generalisation. The observations that you make can certainly be partially explained using your logic but are not the only explanation and, if you forgive me, seem to have a reluctance to involve direct hereditary causation.

    The "clustering" in families to me provides perhaps an overwhelming argument that a spectrum condition is hereditary not some spontaneous event or random and inconsistent extraneous effect from unknown and inexplicable cause.

    Having said that you do seem admit to the possibility that CNV's happen all the time and even though your twin's have different CNV's (do you have their full sequence?) this may still explain why they are both experiencing autistic changes.

    I may be wrong but you read to be selecting some of the genetic information to suit your argument and not taking a more open minded approach, admitting that although there are many pointers to a genetic cause (and CNV's explain many of the variations including both dizygotic and monozygotic twins) there is still much that is not known. Simply putting these inconsistencies down to an environmental explanation and not a gap in our knowledge seems somewhat convenient and almost predictably unscientific to me.

  10. There are many things about autism that are unknown or where there is insufficient evidence to definitely draw a conclusion. In fact, the things that we don't know about autism far outweigh the little that we do know. The idea that autism is an inherited disorder as opposed to environmental one is definitely one of those areas.

    Although, to split some hairs, the real idea here that autism is heritable condition, not necessarily one that is inherited. And I think that the both either side of the argument support the idea that the changes that result in autism are heritable even if they don't come directly from the parent's dna.

    However, I don't think that I am overstating the case that the idea that autism is an inherited disorder is rather weak to start with. This idea that autism is genetic comes straight from several decades worth of twin studies and, if you go back and read these studies to see what they actually say (which I did), you would see that the current study is superior to almost every study that has come before it.

    Please note that I am talking about studies of twins and autism - not studies of autistic traits in twins. The former is directly on point where while the studies about "traits" tend to be poor or too vague.

    So since the current study is better and directly contradicts the earlier studies I think that calls into question the entire idea. If you think that is unscientific, well, that is certainly your opinion. But when I see evidence that's directly contracts the foundation that another idea is built on, I tend to question the second idea.

    The rest of the available evidence (that I have seen) is questionable and can be interpreted differently depending on where you start out from. For example, if autism is primarily an inherited condition then the clustering in families would make sense. However, if autism has a prenatal cause, such as the mother having too high a level of certain endocrine disruptors such as BPA, then the clustering would also make sense.

    The fact is that siblings of children on the spectrum appear to have a higher risk of having autism than a child who doesn't have a sibling on the spectrum. What that fact means or how to interpret it is up for debate.

    There are also other data that would only support the idea that autism environmental, such as if the actual prevalence of autism has risen. After reading everything that I could over the past several years and spending a good amount of time looking at the actual numbers, it is my opinion that at least chunk of the increase is real.

    But who knows, you could be right and I am ignoring evidence. A lot of my opinions about the genetics of autism have been formed by reading the literature but they have also been shaped by observing what autism looks like in my identical twin daughters. Seeing what autism actually looks like in identical twins when both twins have the same severity is very interesting.

    The core problems are the same in both but the exact form that the problem takes can be different. And then when you compare that to the other area that are known to be very strongly "genetic" and see that those sort of differences aren't present, it makes you wonder how much of their autism is being caused by some genetic mutation and how much is being caused by some biological disruption.

    And then you have the identical twins where only one has autism or where the severity is different. In these cases there has to be something other than inherited genetic mutations that is causing these differences.

    I could go on, but I think you get where I am coming from here. The inherited model of autism can't explain all of the facts any more than the environmental only model can. There is going to be some element of both in the final answer, I just happen to think that the environment plays a larger role that is currently thought.

  11. Regarding the CNVs, we only have the results of a microarray test, not a full sequence on them. However, the microarray was a fairly detailed and hit most of the major areas of their genome that had any clinical significance.