Sunday, August 14, 2011

Ghosts of Autism Studies Past

I have been reading older autism twin studies in an effort to put the latest autism twin studies and the idea that autism is a genetic disorder into a historical context.  And the more I read of these old studies, the more the following line keeps coming to mind -

Those who cannot remember the past are condemned to repeat it.

We all hang on the latest word in autism research but I have to wonder how much of what we read is really new.  As I am reading these old studies, I keep bumping into ideas that are being presented now as new and wonderful "discoveries".  So I find myself wondering just how much is new and how much is simply a regurgitation of an older ideas that simply fell out of scientific fashion.

Lets take a look at some of the new, radical ideas about autism in terms of historical studies.

To start off with, what about the recent idea that a maternal infection during pregnancy might increase the risk of autism?  If you read Steffenburgh et al 1989, you would see infections during pregnancy being mentioned as a possible cause of autism  -
In not a single one of the pairs discordant for AD was there a reversed relationship in this respect. There were no cases of diagnosed rubella infection in pregnancy or of other infections currently thought to be of causative importance in autism.
How about the idea that autism might be a combination of genetic and environment influences?  Folsten & Rutter in 1977 -
In studying genetic factors, it is necessary to bear in mind that autism is probably a behavioural syndrome with multiple aetiologies (Rutter, 1974). Certainly, it is known that the syndrome can develop in association with medical conditions as pathologically diverse as congenital rubella (Chess et al, 1971) and infantile spasms (Taft and Cohen, 1971). Accordingly, the investigation of possible hereditary factors must take account of aetiological heterogeneity.
What about the idea that prenatal/perinatal complications might play a role in autism?  Steffenburgh et al in 1989 had this to say -
Nevertheless, the association with perinatal problems in discordant autism twin pairs shows that sometimes perinatal problems alone can be a major aetiological contributory factor.
How about the idea that identical twins might not both have autism and if they do both have autism the severity might be different? In the granddaddy of all autism twin studies in 1977, Folsten & Rutter found that only 4 out of the 11 pairs of identical twins were concordant for autism and that even if they were concordant they had significant differences -
Two of the concordant MZ pairs (i and 3*) were closely similar in all respects. In each case, the twins were severely retarded and the autism was somewhat atypical in terms of the limited evidence of ritualistic features. However, in both the other two pairs, there were important differences between the twins in spite of concordance for autism. In one (2) there was an 18 point difference in non-verbal I.Q_. and a 24 point difference in Peabody language quotient. The twin with a lower non-verbal I.Q,. but higher verbal I.Q,. made much more progress in both social relationships and use of language. In the fourth pair (4) there was a 39 point I.Q,. difference; in this case, the more intelligent twin was less severely autistic, although the type of behaviour was closely similar in both. It is also notable that the more intelligent twin did not develop autism until 3 years of age, although apart from the late onset the clinical picture was typical of autism. 
But you might reply that that was under an old, strange definition of autism that would be quite different than one in use today.  Nope, here is the definition of autism from that study -
a serious impairment in the development of social relationships of the type characteristic of autism (that is with limited eye to eye gaze, poor social responsiveness, impaired selective bonding, a relative failure to go to parents for comfort, and, when older, a lack of empathy, a lack of personal friendships and little group interaction); together with delayed and deviant language development with some of the specific features associated with autism (namely poor language comprehension, little use of gesture, echolalia, pronominal reversal, limited social usage of language, repetitive utterances, flat or staccato speech and very restricted imaginative play); and also stereotyped, repetitive or ritualistic play and interests (as indicated by an abnormal attachment to objects, marked resistance to change, rituals, repetitive behaviour, unusual preoccupations and restricted interest patterns).
That looks almost like a modern definition of autistic disorder to me.  And as recent results have shown, autistic disorder (i.e. classic autism) makes up somewhere between 30% to 50% of all modern cases.

But it seems like coming into the "modern" era of autism research that we have forgotten some of these ideas. By the time Bailey et al in 1995 was published, much of these ideas were put put out to pasture and autism was declared a genetic disorder.  It is only now, in modern times, that we have rediscovered some of the ideas about autism that were common over 35 years ago.

In closing, let me leave you with two little fun facts about these old twin studies.  

The following twin studies make up the bulk of historical twin studies that looked at actual co-morbidity in twins - Folsten & Rutter 1977, Steffenburgh et al 1989, and Bailey et al 1995 - and all three of these studies have one very important thing in common.

All three of these studies excluded opposite sex twins, otherwise know as half of the fraternal twin population.

Second, even though the numbers of children with a label has increased by a factor of 25 (2400%) over the years, one thing has not changed - the 4 to 1 ratio of boys to girls.  That ratio is noted all the way back in Folsten & Rutter in 1977 and the reference in there is to an older study from 1966.  

So ask yourself just one question, if the definition of autism has changed so radically over the years and we are now calling things "autism" that would never have been called autism in the past, why hasn't this ratio of boys to girls changed?


Folstein, S, and M Rutter. 1977. “Infantile autism: a genetic study of 21 twin pairs.” Journal of child psychology and psychiatry, and allied disciplines 18(4):297-321.

Ritvo, E R, B J Freeman, A Mason-Brothers, A Mo, and A M Ritvo. 1985. “Concordance for the syndrome of autism in 40 pairs of afflicted twins.” The American journal of psychiatry 142(1):74-7.

Steffenburg, S et al. 1989. “A twin study of autism in Denmark, Finland, Iceland, Norway and Sweden.” Journal of child psychology and psychiatry, and allied disciplines 30(3):405-16.

Bailey, A et al. 1995. “Autism as a strongly genetic disorder: evidence from a British twin study.” Psychological medicine 25(1):63-77.


  1. Aren't boys more susceptible to terratogens? I don't recall the reasoning...

  2. If there is an environmental influence that is causing the increase and the numbers are actually growing, then the ratio remaining the same makes sense.

    But, for example, if a large cause for the increase was the expanding/changing definition of autism then I am at a little bit of a loss to explain how the ratio could be preserved.

    Think of it this way, I start out by saying that only apples are fruit and notice that most are red (4 to 1 ratio). A few years later I decide to exapand the definition and start calling pears fruit as well. A few years after that I include organges. Would you still expect that 4 to 1 ratio of red to non-red to be there?

  3. It is the early European twin studies that led to the notion that autism is the most 'heritable' of the developmental disorders.

    The idea that you can calculate a 'heritability' estimate by the difference in concordance rates between MZ and DZ twins can't explain Downs Syndrome which is caused by a de novo mutation (Trisomy 21)and is not inherited.

    In Downs Syndrome, which is also associated with high rates of co-occuring autism, all MZ are concordant for Downs Syndrome and all DZ twins are discordant for Downs Syndrome.

    Calculating a 'heritability' estimate for Downs Syndrome would make Downs Syndrome, not autism the most 'heritable' of the developmental disorders which is throwing all logic and reasoning out the window.

  4. RAJ,

    As I understand it, and I could easily be wrong, heritable doesn't mean the same thing as inherited. As "random" mutation such as Downs can be highly heritable while not being inherited.

    So, while you wouldn't expect DZ twins to both have Downs (although it would occur at the same rate as other siblings), you would expect that is be a good chance that the mutation that causes Downs could be passed to any potential children.