A Tale of Two Genes

An unusually gripping genetics paper from Biological Psychiatry: Pagnamenta et al.

The authors discuss a family where two out of the three children were diagnosed with autism. In 2009, they detected a previously unknown copy number variant mutation in the two affected brothers: a 594 kb deletion knocking out two genes, called DOCK4 and IMMP2L.

Yet this mutation was also carried by their non-autistic mother and sister, suggesting that it wasn't responsible for the autism. The mother's side of the family, however, have a history of dyslexia or undiagnosed "reading difficulties"; all of the 8 relatives with the mutation "performed poorly on reading assessment".

Further investigation revealed that the affected boys also carried a second, entirely separate, novel deletion, affecting the gene CNTNAP5. Their mother and sister did not. This mutation came from their father, who was not diagnosed with autism but apparently had "various autistic traits".

Perhaps it was the combination of the two mutations that caused autism in the two affected boys. The mother's family had a mutation that caused dyslexia; the father's side had one that caused some symptoms of autism but was not, by itself, enough to cause the disorder per se.

However, things aren't so clear. There were cases of diagnosed autism spectrum disorders in the father's family, although few details are given and DNA was only available from one of the father's relatives. So it may have been that the autism was all about the CNTNAP5, and this mutation just has a variable penetrance, causing "full-blown" autism in some people and merely traits in others (like the father).

In order to try to confirm whether these two mutations do indeed cause dyslexia and autism, they searched for them in several hundred unrelated autism and dyslexia patients as well as healthy controls. They detected the a DOCK4 deletion in 1 out of 600 dyslexics (and in his dyslexic father, but not his unaffected sister), but not in 2000 controls. 3 different CNTNAP5 mutations were found in the affected kids from 3 out of 143 autism families, although one of them was also found in over 1000 controls.

This is how psychiatric genetics is shaping up: someone finds a rare mutation in one family, they follow it up, and it's only carried by one out of several hundred other cases. So there are almost certainly hundreds of genes "for" disorders like autism, and it only takes a mutation in one (or two) to cause autism.

Here's another recent example: they found PTCHD1 variants in a full 1% of autism cases. It seems to me that autism, for example, is one of the things that happens when something goes wrong during brain development. Hundreds of genes act in synchrony to build a brain; it only takes one playing out of tune to mess things up, and autism is one common result.

Mental retardation and epilepsy are the other main ones, and we know that there are dozens or hundreds of different forms of these conditions each caused by a different gene or genes. The million dollar question is what it is that makes the autistic brain autistic, as opposed to, say, epileptic.

The "rare variants" model has some interesting implications. The father in the Pagnamenta et al. study had never been diagnosed with anything. He had what the authors call "autistic traits", but presumably he and everyone just thought of those as part of who he was - and they could have been anything from shyness, to preferring routine over novelty, to being good at crosswords.

Had he not carried the CNTNAP5 mutation, he'd have been a completely different person. He might well have been drawn to a very different career, he'd probably never have married the woman he did, etc.

Of course, that doesn't mean that it's "the gene for being him"; all of his other 23,000 genes, and his environment, came together to make him who he was. But the point is that these differences don't just pile up on top of each other; they interact. One little change can change everything.

Link: BishopBlog on why behavioural genetics is more complicated than some people want you to think.

Pagnamenta, A., Bacchelli, E., de Jonge, M., Mirza, G., Scerri, T., Minopoli, F., Chiocchetti, A., Ludwig, K., Hoffmann, P., & Paracchini, S. (2010). Characterization of a Family with Rare Deletions in CNTNAP5 and DOCK4 Suggests Novel Risk Loci for Autism and Dyslexia Biological Psychiatry, 68 (4), 320-328 DOI: 10.1016/j.biopsych.2010.02.002

A Tale of Two Genes

An unusually gripping genetics paper from Biological Psychiatry: Pagnamenta et al.

The authors discuss a family where two out of the three children were diagnosed with autism. In 2009, they detected a previously unknown copy number variant mutation in the two affected brothers: a 594 kb deletion knocking out two genes, called DOCK4 and IMMP2L.

Yet this mutation was also carried by their non-autistic mother and sister, suggesting that it wasn't responsible for the autism. The mother's side of the family, however, have a history of dyslexia or undiagnosed "reading difficulties"; all of the 8 relatives with the mutation "performed poorly on reading assessment".

Further investigation revealed that the affected boys also carried a second, entirely separate, novel deletion, affecting the gene CNTNAP5. Their mother and sister did not. This mutation came from their father, who was not diagnosed with autism but apparently had "various autistic traits".

Perhaps it was the combination of the two mutations that caused autism in the two affected boys. The mother's family had a mutation that caused dyslexia; the father's side had one that caused some symptoms of autism but was not, by itself, enough to cause the disorder per se.

However, things aren't so clear. There were cases of diagnosed autism spectrum disorders in the father's family, although few details are given and DNA was only available from one of the father's relatives. So it may have been that the autism was all about the CNTNAP5, and this mutation just has a variable penetrance, causing "full-blown" autism in some people and merely traits in others (like the father).

In order to try to confirm whether these two mutations do indeed cause dyslexia and autism, they searched for them in several hundred unrelated autism and dyslexia patients as well as healthy controls. They detected the a DOCK4 deletion in 1 out of 600 dyslexics (and in his dyslexic father, but not his unaffected sister), but not in 2000 controls. 3 different CNTNAP5 mutations were found in the affected kids from 3 out of 143 autism families, although one of them was also found in over 1000 controls.

This is how psychiatric genetics is shaping up: someone finds a rare mutation in one family, they follow it up, and it's only carried by one out of several hundred other cases. So there are almost certainly hundreds of genes "for" disorders like autism, and it only takes a mutation in one (or two) to cause autism.

Here's another recent example: they found PTCHD1 variants in a full 1% of autism cases. It seems to me that autism, for example, is one of the things that happens when something goes wrong during brain development. Hundreds of genes act in synchrony to build a brain; it only takes one playing out of tune to mess things up, and autism is one common result.

Mental retardation and epilepsy are the other main ones, and we know that there are dozens or hundreds of different forms of these conditions each caused by a different gene or genes. The million dollar question is what it is that makes the autistic brain autistic, as opposed to, say, epileptic.

The "rare variants" model has some interesting implications. The father in the Pagnamenta et al. study had never been diagnosed with anything. He had what the authors call "autistic traits", but presumably he and everyone just thought of those as part of who he was - and they could have been anything from shyness, to preferring routine over novelty, to being good at crosswords.

Had he not carried the CNTNAP5 mutation, he'd have been a completely different person. He might well have been drawn to a very different career, he'd probably never have married the woman he did, etc.

Of course, that doesn't mean that it's "the gene for being him"; all of his other 23,000 genes, and his environment, came together to make him who he was. But the point is that these differences don't just pile up on top of each other; they interact. One little change can change everything.

Link: BishopBlog on why behavioural genetics is more complicated than some people want you to think.

Pagnamenta, A., Bacchelli, E., de Jonge, M., Mirza, G., Scerri, T., Minopoli, F., Chiocchetti, A., Ludwig, K., Hoffmann, P., & Paracchini, S. (2010). Characterization of a Family with Rare Deletions in CNTNAP5 and DOCK4 Suggests Novel Risk Loci for Autism and Dyslexia Biological Psychiatry, 68 (4), 320-328 DOI: 10.1016/j.biopsych.2010.02.002

Autistic Toddlers Like Screensavers

Young children with autism prefer looking at geometric patterns over looking at other people. At least, some of them do. That's according to a new study - Preference for Geometric Patterns Early in Life As a Risk Factor for Autism.

Pierce et al took 110 toddlers (age 14 to 42 months). Some of them had autism, some had "developmental delay" but not autism, and some were normally developing.

The kids were shown a one-minute video clip. One half of the screen showed some kids doing yoga, while the other was a set of ever-changing complex patterns. A bit like a screensaver or a kaleidoscope. Eye-tracking apparatus was used to determine which side of the screen each child was looking at.

What happened? Both the healthy control children, and the developmentally delayed children, showed a strong preference for the "social" stimuli - the yoga kids. However, the toddlers with an autism spectrum disorder showed a much wider range of preferences. 40% of them preferred the geometric patterns. Age wasn't a factor.

This makes intuitive sense because one of the classic features of autism is a fascination with moving shapes such as wheels, fans, and so on. The authors conclude that

A preference for geometric patterns early in life may be a novel and easily detectable early signature of infants and toddlers at risk for autism.

But only a minority of the autism group showed this preference, remember. As you can see from the plot above, they spanned the whole range - and over half behaved entirely normally.

There was no difference between the "social" and "geometrical" halves of the autism group on measures of autism symptoms or IQ, so it wasn't just that only "more severe" autism was associated with an abnormal preference.

They re-tested many of the kids a couple of weeks later, and found a strong correlation between their preference on both occasions, suggesting that it is a real fondness for one over the other - rather than just random eye-wandering.

So this is an interesting result, but it's not clear that it would be of much use for diagnosis.

Pierce K, Conant D, Hazin R, Stoner R, & Desmond J (2010). Preference for Geometric Patterns Early in Life As a Risk Factor for Autism. Archives of general psychiatry PMID: 20819977

Autistic Toddlers Like Screensavers

Young children with autism prefer looking at geometric patterns over looking at other people. At least, some of them do. That's according to a new study - Preference for Geometric Patterns Early in Life As a Risk Factor for Autism.

Pierce et al took 110 toddlers (age 14 to 42 months). Some of them had autism, some had "developmental delay" but not autism, and some were normally developing.

The kids were shown a one-minute video clip. One half of the screen showed some kids doing yoga, while the other was a set of ever-changing complex patterns. A bit like a screensaver or a kaleidoscope. Eye-tracking apparatus was used to determine which side of the screen each child was looking at.

What happened? Both the healthy control children, and the developmentally delayed children, showed a strong preference for the "social" stimuli - the yoga kids. However, the toddlers with an autism spectrum disorder showed a much wider range of preferences. 40% of them preferred the geometric patterns. Age wasn't a factor.

This makes intuitive sense because one of the classic features of autism is a fascination with moving shapes such as wheels, fans, and so on. The authors conclude that

A preference for geometric patterns early in life may be a novel and easily detectable early signature of infants and toddlers at risk for autism.

But only a minority of the autism group showed this preference, remember. As you can see from the plot above, they spanned the whole range - and over half behaved entirely normally.

There was no difference between the "social" and "geometrical" halves of the autism group on measures of autism symptoms or IQ, so it wasn't just that only "more severe" autism was associated with an abnormal preference.

They re-tested many of the kids a couple of weeks later, and found a strong correlation between their preference on both occasions, suggesting that it is a real fondness for one over the other - rather than just random eye-wandering.

So this is an interesting result, but it's not clear that it would be of much use for diagnosis.

Pierce K, Conant D, Hazin R, Stoner R, & Desmond J (2010). Preference for Geometric Patterns Early in Life As a Risk Factor for Autism. Archives of general psychiatry PMID: 20819977

Is Your Brain Autistic?

There's been a lot of buzz and some scepticism about the

New brain scan to diagnose autism

Here's a quick overview. Autism is believed to be a disorder of brain development. If so, it should be possible to diagnose it based on a brain scan. Unfortunately, it's not. You can't tell, from a scan, whether someone has autism or not. Not even if you're a world expert.

There are reports of various differences between autistic and non-autistic brains - a bit smaller here, a bit bigger there - but there's a lot of overlap. So at present, diagnosis of autism is purely based on symptoms.

Ecker et al, a team based at the Institute of Psychiatry in South London, made use of a mathematical technique called a Support Vector Machine (SVM) to try to spot differences that the naked eye can't. An SVM is a learning algorithm: you "teach" it to spot differences by showing it lots of examples. In this case, they showed it 20 autistic brains, and the brains of 20 healthy controls matched for age, gender, and IQ.

How does an SVM work? Imagine that there are two kinds of, say, fruit. Both are kind of round but A's are more spherical than B's. So you could draw a plot of sphere-ness, and find a line separating A and B:
An SVM is an automatic method of finding that line. How? It's complex, but fortunately you don't need to know (I don't). Of course, that's easy, but imagine that things got more tricky. As well as the variable of roundness, there's colour. Fruit B can be either spherical and dark, or non-spherical and light (maybe it's two different stages of ripeness).

An SVM could do this easily too:
Now suppose that there's 1000 different variables, and you want to find the "line" - actually a 1000-dimensional "hyperplane" (a line is 2D, a plane is 3D, anything with 4D or more is a hyper-plane) - dividing the "space" of possibilities into two.

For a human that's impossible, but not for an SVM. This is essentially what Ecker et al did. Each dimension of their "space" was the amount of grey matter at a particular point in the brain. So, they were training the SVM to distinguish between autistic brains and non-autistic brains, based on their shape, but in a much more complex way than a human could.

Did it work? Surprisingly well. Here's the end result (the multi-dimensional space has been helpfully compressed into 2D by the SVM):
It wasn't perfect, but the best approach, based on the cortical thickness in the left hemisphere, managed 90% accuracy, which is pretty awesome. Focussing on the headline 90% result is cherry-picking a bit, because using other variables, like cortical curvature, wasn't as good, but even the worst ones managed 70-85%, much better than chance (50%). Importantly, they also tried the system on 20 adults with ADHD, and it classified them as non-autistic. This shows that it's not just measuring "normality".

*

Now the question everyone's asking: is this going to be used for diagnosis in the real world any time soon? The first thing to remember is that this is a scientific paper, and this result is first and foremost of research interest: it provides clues towards the biology, and ultimately the causes, of autism.

But let's suppose you're a clinician and you have someone who you suspect may have autism, but you're not sure. They're a tricky one, a borderline case. You use this system on their brain and it says they are autistic. Should that factor into your decision? It depends. The fact is that rather than an either-or result, the SVM returns a distance from the hyperplane for each brain. You can see this clearly in the plot above.

In my opinion, if you have a borderline case, and the machine says he's borderline, then that's not much help, and it doesn't matter if he's just over the line, or not quite over it. You already knew he was borderline.

But if the machine says that he's deep into the autism space, then I think that is something. It tells you that his brain is very typical of people with autism. Interestingly, Ecker et al found that distance from the hyperplane correlated with symptom severity for "social" and "communication" symptoms (though not "repetitive behaviours"). That's a pretty cool result because the SVM wasn't trained to do that, it was trained to decide on an either-or basis.

What needs to happen next? As it stands, this system only works for adults: it would fail for children or teenagers, because their brains are a very different size and shape. Exactly the same SVM approach could be used in younger age groups, though, so long as the patients and the controls were the same age.

We also need to make sure that the SVM can tell the difference between autism and other conditions; Ecker et al showed that it could distinguish autism from ADHD, but that's only one comparison and it might not be the hardest one: I would want to see it tested against things like epilepsy, mental retardation, and dyslexia as well.

Overall though, this is very exciting work, and certainly a cut above most "Brain Scans To Diagnose Mental Illness" studies that make it into the headlines.

Full Discloser: I know some of the researchers involved in this work.

Links: The same team had a paper out a few months back, using a slightly less sophisticated SVM approach, which managed 80% accuracy. I wrote about another application of SVMs previously: How To Read Minds. This study has been blogged about at The New Republic and Dormivigilia.

Ecker C, Marquand A, Mourão-Miranda J, Johnston P, Daly EM, Brammer MJ, Maltezos S, Murphy CM, Robertson D, Williams SC, & Murphy DG (2010). Describing the brain in autism in five dimensions--magnetic resonance imaging-assisted diagnosis of autism spectrum disorder using a multiparameter classification approach. Journal of Neuroscience, 30 (32), 10612-23 PMID: 20702694

Is Your Brain Autistic?

There's been a lot of buzz and some scepticism about the

New brain scan to diagnose autism

Here's a quick overview. Autism is believed to be a disorder of brain development. If so, it should be possible to diagnose it based on a brain scan. Unfortunately, it's not. You can't tell, from a scan, whether someone has autism or not. Not even if you're a world expert.

There are reports of various differences between autistic and non-autistic brains - a bit smaller here, a bit bigger there - but there's a lot of overlap. So at present, diagnosis of autism is purely based on symptoms.

Ecker et al, a team based at the Institute of Psychiatry in South London, made use of a mathematical technique called a Support Vector Machine (SVM) to try to spot differences that the naked eye can't. An SVM is a learning algorithm: you "teach" it to spot differences by showing it lots of examples. In this case, they showed it 20 autistic brains, and the brains of 20 healthy controls matched for age, gender, and IQ.

How does an SVM work? Imagine that there are two kinds of, say, fruit. Both are kind of round but A's are more spherical than B's. So you could draw a plot of sphere-ness, and find a line separating A and B:
An SVM is an automatic method of finding that line. How? It's complex, but fortunately you don't need to know (I don't). Of course, that's easy, but imagine that things got more tricky. As well as the variable of roundness, there's colour. Fruit B can be either spherical and dark, or non-spherical and light (maybe it's two different stages of ripeness).

An SVM could do this easily too:
Now suppose that there's 1000 different variables, and you want to find the "line" - actually a 1000-dimensional "hyperplane" (a line is 2D, a plane is 3D, anything with 4D or more is a hyper-plane) - dividing the "space" of possibilities into two.

For a human that's impossible, but not for an SVM. This is essentially what Ecker et al did. Each dimension of their "space" was the amount of grey matter at a particular point in the brain. So, they were training the SVM to distinguish between autistic brains and non-autistic brains, based on their shape, but in a much more complex way than a human could.

Did it work? Surprisingly well. Here's the end result (the multi-dimensional space has been helpfully compressed into 2D by the SVM):
It wasn't perfect, but the best approach, based on the cortical thickness in the left hemisphere, managed 90% accuracy, which is pretty awesome. Focussing on the headline 90% result is cherry-picking a bit, because using other variables, like cortical curvature, wasn't as good, but even the worst ones managed 70-85%, much better than chance (50%). Importantly, they also tried the system on 20 adults with ADHD, and it classified them as non-autistic. This shows that it's not just measuring "normality".

*

Now the question everyone's asking: is this going to be used for diagnosis in the real world any time soon? The first thing to remember is that this is a scientific paper, and this result is first and foremost of research interest: it provides clues towards the biology, and ultimately the causes, of autism.

But let's suppose you're a clinician and you have someone who you suspect may have autism, but you're not sure. They're a tricky one, a borderline case. You use this system on their brain and it says they are autistic. Should that factor into your decision? It depends. The fact is that rather than an either-or result, the SVM returns a distance from the hyperplane for each brain. You can see this clearly in the plot above.

In my opinion, if you have a borderline case, and the machine says he's borderline, then that's not much help, and it doesn't matter if he's just over the line, or not quite over it. You already knew he was borderline.

But if the machine says that he's deep into the autism space, then I think that is something. It tells you that his brain is very typical of people with autism. Interestingly, Ecker et al found that distance from the hyperplane correlated with symptom severity for "social" and "communication" symptoms (though not "repetitive behaviours"). That's a pretty cool result because the SVM wasn't trained to do that, it was trained to decide on an either-or basis.

What needs to happen next? As it stands, this system only works for adults: it would fail for children or teenagers, because their brains are a very different size and shape. Exactly the same SVM approach could be used in younger age groups, though, so long as the patients and the controls were the same age.

We also need to make sure that the SVM can tell the difference between autism and other conditions; Ecker et al showed that it could distinguish autism from ADHD, but that's only one comparison and it might not be the hardest one: I would want to see it tested against things like epilepsy, mental retardation, and dyslexia as well.

Overall though, this is very exciting work, and certainly a cut above most "Brain Scans To Diagnose Mental Illness" studies that make it into the headlines.

Full Discloser: I know some of the researchers involved in this work.

Links: The same team had a paper out a few months back, using a slightly less sophisticated SVM approach, which managed 80% accuracy. I wrote about another application of SVMs previously: How To Read Minds. This study has been blogged about at The New Republic and Dormivigilia.

Ecker C, Marquand A, Mourão-Miranda J, Johnston P, Daly EM, Brammer MJ, Maltezos S, Murphy CM, Robertson D, Williams SC, & Murphy DG (2010). Describing the brain in autism in five dimensions--magnetic resonance imaging-assisted diagnosis of autism spectrum disorder using a multiparameter classification approach. Journal of Neuroscience, 30 (32), 10612-23 PMID: 20702694

Autism And Wealth

We live in societies where some people are richer than others - though the extent of wealth inequality varies greatly around the world.

In general, it's sad but true that poor people suffer more diseases. Within a given country almost all physical and mental illnesses are more common amongst the poor, although this isn't always true between countries.

So if a certain disease is more common in rich people within a country, that's big news because it suggests that something unusual is going on. Autism spectrum disorders (ASDs) have long been known to show this pattern, at least in some countries, but this has often been thought to be a product of diagnostic ascertainment bias. Maybe richer and better-educated parents are more likely to have access to services that can diagnose autism. This is a serious issue because autism often goes undiagnosed and diagnosis is rarely clear-cut.

An important new PLoS paper from Wisconsin's Durkin et al suggests that, while ascertainment bias does happen, it doesn't explain the whole effect in the USA: richer American families really do have more autism than poorer ones. The authors made use of the ADDM Network which covers about 550,000 8 year old children from several sites across the USA. (This paper also blogged about here at C6-H12-O6 blog.)

ADDM attempts to count the number of children with autism based on

abstracted data from records of multiple educational and medical sources to determine the number of children who appear to meet the ASD case definition, regardless of pre-existing diagnosis. Clinicians determine whether the ASD case definition is met by reviewing a compiled record of all relevant abstracted data.

Basically, this allowed them to detect autism even in kids who haven't got a formal diagnosis, based on reports of behavioural problems at school etc indicative of autism. Clearly, this is going to underestimate autism somewhat, because some autistic kids do well at school and don't cause any alarm bells, but it has the advantage of reducing ascertainment bias.

What happened? The overall prevalence of autism was 0.6%. This is a lot lower than recent estimates in 5-9 year olds in the UK (1.5%), but the UK estimates used an even more detailed screening technique which was less likely to leave kids undetected.

The headline result: autism was more common in kids of richer parents. This held true within all ethnic groups: richer African-American or Hispanic parents were more likely to have autistic children compared to poorer people of the same ethnicity. So it wasn't a product of ethnic disparities.

Crucially, the pattern held true in children who had never been diagnosed with autism, although the effects of wealth were quite a bit smaller:

The difference in the slope of the two lines suggests that there is some ascertainment bias, with richer parents being more likely to get a diagnosis for their children, but this can't explain the whole story. There really is a correlation with wealth.

So what does this mean? This is a correlation - the causality remains to be determined. There are two obvious possibilities: to put it bluntly, either being rich makes your kids autistic, or having autistic kids makes you rich.

How could being rich make your children autistic? There could be many reasons, but a big one is paternal age: it's known that the risk of autism rises with the age of the father, maybe because the sperm of older men accumulates more genetic damage, and this damage can cause autism. In general richer people wait longer to have kids (I think, although I can't actually find the data on this) so maybe that's the cause.

How could having autistic kids make you richer? Well, unfortunately I don't think it does directly, but maybe being the kind of person who is likely to have an autistic child could. Autism is highly heritable, so the parents of autistic children are likely to carry some "autism genes". These could give them autistic traits, or indeed autism, and autistic traits, like being intensely interested in complex intellectual matters, can be a positive advantage in many relatively well paid professions like scientific research, or computing. Marginal Revolution's Tyler Cowen recently wrote a book all about that. I hope I will not offend too many when I say that in my experience it's rare to meet a scientist, IT person or, say, neuroscience blogger, who doesn't have a few...

Durkin, M., Maenner, M., Meaney, F., Levy, S., DiGuiseppi, C., Nicholas, J., Kirby, R., Pinto-Martin, J., & Schieve, L. (2010). Socioeconomic Inequality in the Prevalence of Autism Spectrum Disorder: Evidence from a U.S. Cross-Sectional Study PLoS ONE, 5 (7) DOI: 10.1371/journal.pone.0011551

Autism And Wealth

We live in societies where some people are richer than others - though the extent of wealth inequality varies greatly around the world.

In general, it's sad but true that poor people suffer more diseases. Within a given country almost all physical and mental illnesses are more common amongst the poor, although this isn't always true between countries.

So if a certain disease is more common in rich people within a country, that's big news because it suggests that something unusual is going on. Autism spectrum disorders (ASDs) have long been known to show this pattern, at least in some countries, but this has often been thought to be a product of diagnostic ascertainment bias. Maybe richer and better-educated parents are more likely to have access to services that can diagnose autism. This is a serious issue because autism often goes undiagnosed and diagnosis is rarely clear-cut.

An important new PLoS paper from Wisconsin's Durkin et al suggests that, while ascertainment bias does happen, it doesn't explain the whole effect in the USA: richer American families really do have more autism than poorer ones. The authors made use of the ADDM Network which covers about 550,000 8 year old children from several sites across the USA. (This paper also blogged about here at C6-H12-O6 blog.)

ADDM attempts to count the number of children with autism based on

abstracted data from records of multiple educational and medical sources to determine the number of children who appear to meet the ASD case definition, regardless of pre-existing diagnosis. Clinicians determine whether the ASD case definition is met by reviewing a compiled record of all relevant abstracted data.

Basically, this allowed them to detect autism even in kids who haven't got a formal diagnosis, based on reports of behavioural problems at school etc indicative of autism. Clearly, this is going to underestimate autism somewhat, because some autistic kids do well at school and don't cause any alarm bells, but it has the advantage of reducing ascertainment bias.

What happened? The overall prevalence of autism was 0.6%. This is a lot lower than recent estimates in 5-9 year olds in the UK (1.5%), but the UK estimates used an even more detailed screening technique which was less likely to leave kids undetected.

The headline result: autism was more common in kids of richer parents. This held true within all ethnic groups: richer African-American or Hispanic parents were more likely to have autistic children compared to poorer people of the same ethnicity. So it wasn't a product of ethnic disparities.

Crucially, the pattern held true in children who had never been diagnosed with autism, although the effects of wealth were quite a bit smaller:

The difference in the slope of the two lines suggests that there is some ascertainment bias, with richer parents being more likely to get a diagnosis for their children, but this can't explain the whole story. There really is a correlation with wealth.

So what does this mean? This is a correlation - the causality remains to be determined. There are two obvious possibilities: to put it bluntly, either being rich makes your kids autistic, or having autistic kids makes you rich.

How could being rich make your children autistic? There could be many reasons, but a big one is paternal age: it's known that the risk of autism rises with the age of the father, maybe because the sperm of older men accumulates more genetic damage, and this damage can cause autism. In general richer people wait longer to have kids (I think, although I can't actually find the data on this) so maybe that's the cause.

How could having autistic kids make you richer? Well, unfortunately I don't think it does directly, but maybe being the kind of person who is likely to have an autistic child could. Autism is highly heritable, so the parents of autistic children are likely to carry some "autism genes". These could give them autistic traits, or indeed autism, and autistic traits, like being intensely interested in complex intellectual matters, can be a positive advantage in many relatively well paid professions like scientific research, or computing. Marginal Revolution's Tyler Cowen recently wrote a book all about that. I hope I will not offend too many when I say that in my experience it's rare to meet a scientist, IT person or, say, neuroscience blogger, who doesn't have a few...

Durkin, M., Maenner, M., Meaney, F., Levy, S., DiGuiseppi, C., Nicholas, J., Kirby, R., Pinto-Martin, J., & Schieve, L. (2010). Socioeconomic Inequality in the Prevalence of Autism Spectrum Disorder: Evidence from a U.S. Cross-Sectional Study PLoS ONE, 5 (7) DOI: 10.1371/journal.pone.0011551

Oh Crap. More Autism Genes.

There's been much excitement about the latest big genetic study into autism, published in Nature : the grandly titled Autism Genome Project, brought to you by a crack team of no fewer than 177 researchers.

For a good summary of the research take a look here, and for a longer account here. In a nutshell, the authors examined DNA from almost 1000 people with an autism spectrum disorder. They were looking for deletions and duplications of segments of DNA: so-called copy number variations (CNVs). A CNV could be anything from missing half a chromosome, down to having an extra copy of just a small part of a single gene.

It turned out that autistic cases carry more CNVs affecting genes, on average, than controls. The difference was small - just a 1.2-fold increase - but significant, and reassuringly, the extra CNVs were especially common in genes already known to be related to autism. The authors conclude that about 5% of cases of autism are likely caused by a single CNV of the type they studied. In almost all cases it was a different variant in a different gene - in other words, each case is genetically unique. Here's the gory details.

So we have new autism genes - dozens of them. But is that good news? Not really - with genes, it's not a case of the more the merrier. If there's just one gene for a disease, it's pretty easy to work out how it does it. Genes code for proteins. Proteins do stuff in cells. Follow the trail of causality from gene to protein to the impact on the body, and you've understood the disorder. Nowadays, with the help of modern genetics, you can do this in a few years.

There are several hundreds of these nice easy monogenetic diseases. For example, a few months ago I reported on a case report of a guy with complex neurological and psychiatric symptoms, caused by a single mutation in the gene for the enzyme sepiapterin reductase. All of his symptoms followed logically from that mutation. With autism, there are already many known genes; this study has found many more; more are implicated each year. Oh dear.

My suspicion is that a large proportion of all of the genes that control brain development - which is a lot - will turn out to be autism genes. The brain is amazingly complex. Thousands of genes work in synchrony build a "normal" brain. There are an awful lot of things that might not go according to plan.

Sometimes, the outcome is a rare and bizarre condition like holoprosencephaly. More often, the end result falls into one of a few common categories, like epilepsy and mental retardation (intellectual disability). There is no one gene for these disorders: they're just one of the things that happens when a gene goes wrong. I suspect that autism is another.

At present we have no clear idea what is different about autistic brains. If we did, we could probably predict which genes would be autism genes. For example, one theory of autism is that brain cells are too tightly packed. Suppose that's true (it's almost certainly not that simple), and suppose that one day, someone finds a gene, pushy, that causes developing brain cells to make little molecular spikes that push each other away. It would not take a genius to predict that a mutation that stops pushy working might cause autism.

Of course pushy mutations would only account for a small fraction of cases: plenty of other mutations would have the same eventual impact. The point is that if we understood the biology of autism in this way, we'd know which genes to look for; we wouldn't have to fish around the whole genome looking for all kinds of random mutations.


It's true, as the authors of the Nature paper say, that genes can themselves provide clues as to the nature of the disease. They present a neat diagram of the functional relationships between their autism genes - what they do inside cells. But this is painted with a very broad brush - "cell adhesion" i.e. how cells fit together; "central nervous system development"; "cell proliferation".

This is why I'm personally more interested in research into the psychology and the neuroscience of autism than I am by the genetics. Genetic studies are important but there are glaring gaps in our knowledge that probably deserve at least as much attention.

Just for starters, there have been very few studies simply comparing the brains of autistic people to non-autistic controls at autopsy. I think in total there have been published post mortem reports on maybe 30 or 40 autistic brains...ever. Some very interesting results have emerged, but with such small numbers it's impossible to know what's really going on, especially since most of the cases also suffered other conditions, such as - no surprise - epilepsy and mental retardation. We need more autistic people to donate their brains to science, and more scientists to study them.

Pinto, D. et al. (2010). Functional impact of global rare copy number variation in autism spectrum disorders Nature DOI: 10.1038/nature09146

Oh Crap. More Autism Genes.

There's been much excitement about the latest big genetic study into autism, published in Nature : the grandly titled Autism Genome Project, brought to you by a crack team of no fewer than 177 researchers.

For a good summary of the research take a look here, and for a longer account here. In a nutshell, the authors examined DNA from almost 1000 people with an autism spectrum disorder. They were looking for deletions and duplications of segments of DNA: so-called copy number variations (CNVs). A CNV could be anything from missing half a chromosome, down to having an extra copy of just a small part of a single gene.

It turned out that autistic cases carry more CNVs affecting genes, on average, than controls. The difference was small - just a 1.2-fold increase - but significant, and reassuringly, the extra CNVs were especially common in genes already known to be related to autism. The authors conclude that about 5% of cases of autism are likely caused by a single CNV of the type they studied. In almost all cases it was a different variant in a different gene - in other words, each case is genetically unique. Here's the gory details.

So we have new autism genes - dozens of them. But is that good news? Not really - with genes, it's not a case of the more the merrier. If there's just one gene for a disease, it's pretty easy to work out how it does it. Genes code for proteins. Proteins do stuff in cells. Follow the trail of causality from gene to protein to the impact on the body, and you've understood the disorder. Nowadays, with the help of modern genetics, you can do this in a few years.

There are several hundreds of these nice easy monogenetic diseases. For example, a few months ago I reported on a case report of a guy with complex neurological and psychiatric symptoms, caused by a single mutation in the gene for the enzyme sepiapterin reductase. All of his symptoms followed logically from that mutation. With autism, there are already many known genes; this study has found many more; more are implicated each year. Oh dear.

My suspicion is that a large proportion of all of the genes that control brain development - which is a lot - will turn out to be autism genes. The brain is amazingly complex. Thousands of genes work in synchrony build a "normal" brain. There are an awful lot of things that might not go according to plan.

Sometimes, the outcome is a rare and bizarre condition like holoprosencephaly. More often, the end result falls into one of a few common categories, like epilepsy and mental retardation (intellectual disability). There is no one gene for these disorders: they're just one of the things that happens when a gene goes wrong. I suspect that autism is another.

At present we have no clear idea what is different about autistic brains. If we did, we could probably predict which genes would be autism genes. For example, one theory of autism is that brain cells are too tightly packed. Suppose that's true (it's almost certainly not that simple), and suppose that one day, someone finds a gene, pushy, that causes developing brain cells to make little molecular spikes that push each other away. It would not take a genius to predict that a mutation that stops pushy working might cause autism.

Of course pushy mutations would only account for a small fraction of cases: plenty of other mutations would have the same eventual impact. The point is that if we understood the biology of autism in this way, we'd know which genes to look for; we wouldn't have to fish around the whole genome looking for all kinds of random mutations.


It's true, as the authors of the Nature paper say, that genes can themselves provide clues as to the nature of the disease. They present a neat diagram of the functional relationships between their autism genes - what they do inside cells. But this is painted with a very broad brush - "cell adhesion" i.e. how cells fit together; "central nervous system development"; "cell proliferation".

This is why I'm personally more interested in research into the psychology and the neuroscience of autism than I am by the genetics. Genetic studies are important but there are glaring gaps in our knowledge that probably deserve at least as much attention.

Just for starters, there have been very few studies simply comparing the brains of autistic people to non-autistic controls at autopsy. I think in total there have been published post mortem reports on maybe 30 or 40 autistic brains...ever. Some very interesting results have emerged, but with such small numbers it's impossible to know what's really going on, especially since most of the cases also suffered other conditions, such as - no surprise - epilepsy and mental retardation. We need more autistic people to donate their brains to science, and more scientists to study them.

Pinto, D. et al. (2010). Functional impact of global rare copy number variation in autism spectrum disorders Nature DOI: 10.1038/nature09146

DSM-V: Change We Can Believe In?

So the draft of DSM-V is out.

If, as everyone says, the Diagnostic and Statistical Manual is the Bible of Psychiatry, I'm not sure why it gets heavily edited once every ten years or so. Perhaps the previous versions are a kind of Old Testament, and only the current one represents the New Revelation from the gods of the mind?

Mind Hacks has an excellent summary of the proposed changes. Bear in mind that the book won't be released until 2013. Some of the headlines:

• Asperger's Syndrome is out - everyone's going to have an "autistic spectrum disorder" now.
  
• Personality Disorders are out - kind of. In their place, there's 5 Personality Disorder Types, each of which you can have to varying degrees, and also 6 Personality Traits, each of which you can have to varying degrees.
• Hypoactive Sexual Desire Disorder - the disease which failed-antidepressant-turned-aphrodisiac flibanserin is supposed to treat - is out, to be replaced by Sexual Interest and Arousal Disorder.
• Binge Eating Disorder, Hypersexuality Disorder, and Gambling Addiction are in. Having Fun is not a disorder yet, but that's on the agenda for DSM-VI.

More important, at least in theory, are the Structural, Cross-Cutting, and General Classification Issues. This is where the grand changes to the whole diagnostic approach happen. But it turns out they're pretty modest. First up, the Axis system, by which most disorders were "Axis I", personality disorders which were "Axis II", and other medical illnesses "Axis III", is to be abolished - everything will be on a single Axis from now on. This will have little, if any, practical effect, but will presumably make it easier on whoever it is that has to draw up the contents page of the book.

Excitingly, "dimensional assessments" have been added... but only in a limited way. Some people have long argued that having categorical diagnoses - "schizophrenia", "bipolar disorder", "major depression" etc. - is a mistake, since it forces psychiatrists to pigeon-hole people, and that we should stop thinking in terms of diagnoses and just focus on symptoms: if someone's depressed, say, then treat them for depression, but don't diagnose them with "major depressive disorder".

DSM-V hasn't gone this far - the categorical diagnoses remain in most cases (the exception is Personality Disorders, see above). However, new dimensional assessments have been proposed, which are intended to complement the diagnoses, and some of them will be "cross-cutting" i.e. not tied to one particular diagnosis. See for example here for a cross-cutting questionnaire designed to assess common anxiety, depression and substance abuse symptoms.

Finally, the concept of "mental disorder" is being redefined. In DSM-V a mental disorder is (drumroll)...

A. A behavioral or psychological syndrome or pattern that occurs in an individual

B. The consequences of which are clinically significant distress (e.g., a painful symptom) or disability (i.e., impairment in one or more important areas of functioning)

C. Must not be merely an expectable response to common stressors and losses...

D. That reflects an underlying psychobiological dysfunction

E. That is not primarily a result of social deviance or conflicts with society

The main change here is that now it's all about "psychobiological dysfunction", whereas in DSM-IV, it was about "behavioral, psychological, or biological dysfunction". Hmm. I am not sure what this means, if anything.

But read on, and we find something rather remarkable...

J. When considering whether to add a mental/psychiatric condition to the nomenclature, or delete a mental/psychiatric condition from the nomenclature, potential benefits (for example, provide better patient care, stimulate new research) should outweigh potential harms (for example, hurt particular individuals, be subject to misuse)

This all sounds very nice and sensible. Diagnoses should be helpful, not harmful, right?

No. Diagnoses should be true. The whole point of the DSM is that it's supposed to be an accurate list of the mental diseases that people can suffer from. The diagnoses are in there because they are, in some sense, real, objectively-existing disorders, or at least because the American Psychiatric Association thinks that they are.

This seemingly-innocuous paragraph seems to be an admission that, in fact, disorders are added or subtracted for reasons which have little to do with whether they really, objectively exist or not. This is what's apparently happened in the case of Temper Dysregulation Disorder with Dysphoria (TDDD), a new childhood disorder.

TDDD has been proposed in order to reduce the number of children being diagnosed with pediatric bipolar disorder. The LA Times quote a psychiatrist on the DSM-V team:

The diagnosis of bipolar [in children] "is being given, we believe, too frequently," said Dr. David Shaffer, a member of the work group on disorders in childhood and adolescence. In reality, when such children are tracked into adulthood, very few of them turn out to be bipolar, he said.

And the DSM-V website has a lengthy rationale for TDDD, to the same effect.

Now, many people agree that pediatric bipolar is being over-diagnosed. As I've written before, pediatric bipolar was considered to be a vanishingly rare disease until about 10 years ago, it still is pretty much everywhere outside the USA.

So we can all sympathize with the sentiment behind TDDD - but this is fighting fire with fire. Is the only way to stop kids getting one diagnosis, to give them another one? Should we really be creating diagnoses for more or less "strategic" purposes? When the time comes for DSM-VI, and the fashion for "pediatric bipolar" has receded, will TDDD get deleted as no longer necessary? What will happen to all the "TDDD" kids then?

Can't we just decide to diagnose people less? Apparently, that would be a rather too radical change...