One of the most interesting ASD studies to emerge this year came out while I was in the midst of my Australian speaking tour. Consequently, Im a bit late reporting on it, but better late than never. The study is titled, The relationship of Asperger’s syndrome to autism: a preliminary EEG coherence study. It’s brought to us by a team from Boston Children’s Hospital: Frank Duffy, Aditi Shankardass, Gloria B McAnulty, and Heidelise Als.
Popular media accounts say this research distinguishes Asperger’s from autism. If that’s so, they raise the question: Are Asperger’s and autism two distinct and separate conditions?
I’d like to discuss that very question in this article. I’ll do so by explaining the methods behind the study, and what I think the findings mean. The intrepid among you may want to see the original paper, which you can read here:
You may also want to read the principal author’s 2012 paper, A stable pattern of EEG spectral coherence distinguishes children with autism from neuro-typical controls - a large case control study. It’s the foundation for the work discussed here.
What they did
In 2012, the researchers published a first paper (the second one cited above) in which they described analyzing EEG data from almost 1,000 children – half of whom had been diagnosed with autism and half who were NT. That analysis was able to separate the autistic kids from the NT controls with an accuracy of more than 90%. That in itself was striking. But there’s more, as released in this newest paper . . .
They continued with an additional premise and a question: If our first study showed that EEG data can distinguish people on the autism spectrum from the NT population, could further EEG analysis separate people with Asperger’s from the general autism population?
To answer that question they re-evaluated the data for 430 autistic kids from their first study and compared it to fresh data for 26 kids with anAsperger diagnosis. That was compared to data for 554 neurotypical controls. All kids in the study ranged from 2-12 years in age; recently diagnosed in the Harvard hospital system using current best practice methods. None of the kids in the study had other disorders (epilepsy, for example) that would alter or confound EEG data collection.
To gather the data for analysis researchers placed 24 electrodes in a grid pattern on each subject’s head. Those electrodes collected EEG waves for a minimum of 8 minutes while the kids sat there, awake. Having experienced this process myself I can say EEG collection is not painful or stressful; it’s just boring – sitting still for 10 minutes at a time. If you have every tried to get a five-year-old to sit still 10 minutes you know how challenging that must have been for researchers!
EEG uses electrical sensors to pick up millivolt-level signals generated on the surface (cortex) layers of the brain. These signals are often described as “brain waves,” and indeed they are like waves in that they look sort of like a “side view” of the ocean surface, and they propagate over the whole brain, being strongest near their origin and weakest in the most distant neural areas.
Most brain waves are in the frequency range of 1-20 cycles per second, or 1-20HZ (hertz). A “cycle” is defined as one complete wave; from peak to trough and back up to peak again. Just as the ocean has many complex wave patterns, the brain can have multiple waves laid over one another. The different waves have distinct peak-to-trough periods (said another way, they have different frequencies) that overlay each other to make a complex interwoven pattern.
The waves we see on a monitor are formed from billions of tiny impulses from the constant firing of cortical neurons; the processes of neuronal activity never ends (except when our brain dies) though it does change insleep or unconsciousness.
This particular study looked at what’s called spectral coherence of those wave patterns across the brain. Simply put, spectral coherence is a measure of how well connected two brain areas are. Areas with high coherence will rise and fall at the same time to a particular wave pattern. Areas with low coherence will rise and fall at different – even opposite – times.
When that happens we can say those areas are decoupled from one another, or marching to the beat of different drummers.
We might draw an analogy with ocean waves, as we watch them in a harbor. Waves in a coherent pattern reinforce each other, becoming smooth and strong. Waves in an incoherent pattern are at odds, breaking and jagged. Anyone who has stood at a pier or ridden a ferry and watched the water has seen these patterns of smooth rolling water contrasted with rough unsettled seas.
In this study, the researchers compared signals from those 24 sensors and did some very complex mathematical analysis to find patterns of coherence for waves at several different frequencies. The mathematical technique they used is called discriminant function analysis (DFA.)
What did they find?
In their original 2012 study the researchers found that patterns of coherence could separate autistic people from NT controls with an accuracy of well over 90%. When compared to the NT population, the autistic people had less coherence over short distances, but more coherence over long distances. By way of explanation, the authors theorize that, “predominantly reduced short-distance coherences may indicate poor local network function. The increased long-distance coherences may represent compensatory processes or reduced neural pruning. The wide average spectral range of factor loadings may suggest over-damped neural networks.”
That was a very interesting discovery.
How about the comparison of Asperger people to the broader pool of people with autism?
Their first analysis (2012) sorted subjects into control or autism spectrum groups. The newer analysis (2013) classified 25 of 26 patients with Asperger’s as belonging to the autism spectrum community. They said: “This indicates that subjects with Asperger’s are neurophysiologically closer to the autism spectrum population than to the neurotypical control population.”
Next, they dug into the differences between the Asperger people and the others on the spectrum.
Using discriminant function analysis, 24 out of 26 subjects with Asperger’s were separated from the general autism population. Going in the other direction, the majority of the autism population scored different from the 26 Asperger subjects. It’s worth noting that the general autism population in this study had not included people with an Asperger diagnosis. Had the population been mixed this result would have been mixed too.
What was different between autism and Asperger people?
In their words, The pattern of coherence difference demonstrated that the Asperger population showed even more reduction of left lateral anterior-posterior coherence than the autism group. . . . the Asperger group had markedly increased left mid temporal to central parietal-occipital coherence. It is speculated that this increased left temporal connectivity may partially compensate for the language deficiency suggested by the first difference.
It is also proposed that the postulated compensation may not completely facilitate all aspects of normal language development, and may result in the several, readily identifiable, higher level differences of language use observed in subjects with Asperger’s such as excessive pedantic formality, verbosity, literal interpretation devoid of nuance and prosodic deficiency, to name a few.
This is very interesting, because it suggests a reason why the therapies that work for Asperger kids and kids with autism are so very different. While the researchers still seem to think Asperger’s and autism look like two points on a spectrum, they suggest more study could answer the question:
At this point, current study results are consistent with Asperger’s forming one end of the autism spectrum population. . . . The small size of the tested Asperger population limits definitive determination of whether Asperger’s is a separate entity to autism. Study of a larger Asperger population is necessary to assess this important question in a more conclusive manner.
They also say, Inclusion of the Asperger sample with the overall autistic population did not result in a statistically significant [change] as would be seen if the ASD and ASP populations represented completely differing clinical entities. That too suggests traditional autism and Asperger’s are two points on a curve.
What do I think, as a person on the spectrum?
First of all, I want to stress that these two studies are only looking at children. One of the reasons the DSM5 work group combined the various autistic conditions under one heading was that conditions that look very different in childhood become indistinguishable in adults. An EEG study of adults on the spectrum has yet to be done. We don’t know the extent to which these differences persist into adulthood and through the lifespan. I hope a follow up adult study is done soon.
Second – and this is very important – the study does not suggest that Asperger’s is fundamentally different from other autisms. Rather, it identifies subtle differences while finding the major differences that set people on the spectrum apart from the NT majority are present in both sets of people (the autism and Asperger groups.) We already know autism affects people very differently, and this study may be highlighting a particular subgroup; other such groups may be identified in the future. That could prove valuable by helping us understand how to help autistic individuals with differing needs.
Third – the study’s findings of less coherence over short distance, and greater coherence over long distances seems to tie in with the connectivity theories of Nancy Minshew and Marcel Just at CMU/Pitt. That is fascinating to me.
Finally, in my opinion, this study reinforces the concept of a broad autism spectrum. By identifying a biomarker (the EEG signatures) that separates Asperger kids from kids with traditional autism, it also points the way to a possible future where we develop and use an EEG test to separate kids with different “autisms” and deliver different optimized therapies to each group.
If we can separate Asperger kids today we may also be able to separate other subgroups tomorrow.
By doing so we may then develop better targeted interventions that help more precisely defined population subgroups. That could be life changing.
We may also gain valuable insight from the extension of this study into the adult population. Do we become “normalized” in some ways? Do we merge or stay distinct? No one knows. I eagerly await the answers.
John Elder Robison
John Elder Robison is the author of Raising Cubby, Look Me in the Eye, My Life with Asperger’s, and Be Different - adventures of a free range Aspergian. John’s books on life with autism are sold in ten languages in over 65 countries. He is a member of the Interagency Autism Coordinating Committee of the US Department of Health and Human Services, and he serves as an advisor to many autism related boards and foundations, both public and private. The opinions expressed here are his own.