Immune function: a critical role in autism?

The headline to this post is not my own but rather taken from the conclusions of the study published by Vishal Saxena and colleagues* (open-access) who undertook some computational analyses of some of the various genomic findings detailed in cases of autism spectrum disorders (ASDs).

Rather than me bore you with all the details of their study, I'll instead refer you to the ScienceDaily entry for this study (see here) which gives a far better overview than I possibly could. In essence it all boiled down to systems biology (looking at genes and their various pathways) and the application of something called Linkage-ordered Gene Sets (LoGS) which "takes pre-existing gene sets and ranks them in terms of their importance in autism".

As per their paper and the accompanying press release suggest, genes involved in immune function featured heavily in their results. In particular was the suggestion of some involvement for genes related to cytokine activity (see this post), the interferons (see this post) and the genetics of virus response (see here). Also appearing in the top 20 pathways turned up was an old friend, glutathione transferase activity, which one has to wonder might fit in with some other results on glutathione and autism which were published in recent months (see here and here). Just speculating...

There's other possibilities arising from the data produced by Saxena et al, some of which the authors start to discuss with reference to things like autoimmunity, which at least one of the co-authors has kinda hinted at before in the famous "significantly over-represented" paper discussed not so long ago. That and mention of the words "in-utero infections" and "mouse models of autism" which 'might' take us back to the work of Paul Patterson and colleagues (see here).

Regular readers might know that I'm not exactly enamoured by the results of the multitude of genetics studies which have been completed with autism in mind given both the heterogeneity of the condition and also the seeming lack of appreciation that autism does not exist in a vacuum but instead might significantly elevate the risk of other conditions co-occurring. As we also found out recently, we all carry quite a few mutations around with us, whether diagnosed with autism or anything else and even those don't necessarily mean anything. That and the rise of epigenetics... What I am however drawn to in this paper is the focus on the 'bigger picture' outside of just genes encoding for one protein; as the study authors put it in terms of pixels and zooming out. The question now is: who are these results most pertinent to on the autism spectrum and is this yet more evidence for the autisms?

But enough for now, aside from a final quote from the paper: "The results presented in this paper show that immune function may play a critical role in the genesis, development, or manifestation of autism". Who am I to argue?

----------

* Saxena V. et al. Structural, genetic, and functional signatures of disordered neuro-immunological development in autism spectrum disorder. PLoS ONE. 2012; 7: e48835.

----------

Vishal Saxena, Shweta Ramdas, Courtney Rothrock Ochoa, David Wallace, Pradeep Bhide, & Isaac Kohane (2012). Structural, genetic, and functional signatures of disordered neuro-immunological development in autism spectrum disorder PLoS ONE

Autism: the sum of its SNPs?

I chanced upon a new piece of research published by Lambertus Klei and colleagues* (open-access) on single nucleotide polymorphisms (SNPs) and autism, and how lots of little mutations might additively contribute to increasing autism risk.

Calculate this @ Wikipedia 

OK, stop.. hammertime.. a few descriptions and details are needed first as well as a caveat about my amateur status when it comes to all things genetics.

SNPs and mutations. A short description is included here but basically think small changes to the letters of the genetic code. Suffice to say that I've covered SNPs and those copy number variations (CNVs) and autism on more than one occasion on this blog as per posts like this one on no single SNP being associated with the universality that is the autism label and this one on just how complex autism seems to be from a genetics perspective.

I hope I've not been too unkind to genetic mutation research (sorry for the cold science term) specifically with autism in mind, but like other developmental conditions such as attention-deficit hyperactivity disorder (ADHD), the whole 'genetics is absolute ruler' argument has not been borne out by the data produced so far. Each and every one of us, whether diagnosed with autism or anything else, are the product of mutation and autism is a mighty diverse condition.

With all this in mind, the Klei paper whilst open-access has a few interesting points worth noting:

• This was a study aiming to look at a few important issues. Not only the question of simplex (one child) vs. multiplex (more than one child) genetic risk of autism and whether they might be different, but whether despite not one SNP being universally applied across all autism, there may be some argument for more [important] SNPs cumulatively = a greater risk of autism.
• Based on genotype data derived from two autism-related datasets representing simplex and multiplex families respectively, the Simons Simplex Collection (SSC) and the Autism Genome Project (AGP), comparisons were made with control data (HealthABC) across several hundred thousands of SNPs for something called narrow-sense heritability - in effect additive genetic variance. There was also a further test group used to assess the robustness of findings (Neurogenetics Research Consortium, N=1986).
• Results: "For simplex families, who have only a single affected individual in multiple generations, approximately 40% of liability traces to additive effects whereas this narrow-sense heritability exceeds 60% for ASD individuals from multiplex families". The actual results were: simplex (39.6%) and multiplex (65.5%); this last figure reflective of the AGP cohort splitting, with some simplex and some multiplex families included; simplex heritability in the AGP group being estimated at 49.8%. These estimates also bearing in mind the "unrealized multiplex potential" of taking snapshots of families.
• Another result: "These results suggest that AGP parents carry a greater load of additive risk variants than SSC parents and thus are, on average, closer to the threshold of being affected". Translation: parents of participants in the AGP cohort, bearing in mind this included both simplex and multiplex families, showed more additive risk SNPs and hence were closer to the diagnosis of autism or the broader phenotype than in the SSC cohort taking into account differences in screening for autism in parents between the two cohorts.
• And another result: "A curious observation from AGP multiplex families was that fathers generate larger heritability than mothers". Translation: lots of potential reasons for it (including the issue of mums, dads and differing proportions of sex hormones) but dads seemed to generate more 'risk' than mums in multiplex autism.
• And a final result: "Our results suggest that common variants affecting liability do not cluster on chromosome X". An interesting observation given the focus on the X chromosome as a result of the reported male domination of autism.
• Mention is also made of assortative mating as being involved in the results but as yet no new data is presented on this possibility.

I like the quote from one of the chief authors on the paper, Prof. Bernie Devlin reported here: "The genetic components alone are far more complex than many imagined a decade ago, including the additive effects we have found, rare inherited mutations, and new mutations arising spontaneously before conception." To me this embodies everything about the current state of knowledge on the genetics of autism in that (a) there is probably no universal 'autism' gene, (b) mutations are likely to be an important part of autism risk and (c) new mutations opens the door up to environment as playing a hand (spontaneous eh?). Indeed the only thing missing from this sentence is discussion about the relative newcomer, epigenetics, and how looking at the function of genes might be able to plug some of that gap in the risk for developing autism.

As per my recent post on systems biology and autism, one also has to ask whether there might be some common biochemical pathways to be derived from these additive genetic effects. Not so long ago, Skafidas and colleagues** were discussing SNPs in relation to gene and biochemical functions, with some interesting observations detailed (see this post). I'd be very interested to see what might come from similar analyses on the Klei data and whether there is overlap or differences based on simplex or multiplex cases.

Whilst an interesting paper, there still remains some significant holes in our knowledge about autism and the risk factors for receiving a diagnosis of autism. Those 40% and 60% risk figures are probably what many people are going to focus on, but my question is what about the remaining 60% or 40% of risk and indeed the question of 'where did the heritability go?' based on previous twin studies lining the autism research trail (see here). Where does that risk come from? and should we perhaps be looking at environmental factors with as much assiduity as we do genetic ones?

----------

* Klei L. et al. Common genetic variants, acting additively, are a major source of risk for autism. Molecular Autism. 2012; 3: 9. doi:10.1186/2040-2392-3-9

** Skafidas E. et al. Predicting the diagnosis of autism spectrum disorder using gene pathway analysis Molecular Psychiatry. September 2012; DOI: 10.1038/mp.2012.126

----------

Klei, L., Sanders, S., Murtha, M., Hus, V., Lowe, J., Willsey, A., Moreno-De-Luca, D., Yu, T., Fombonne, E., Geschwind, D., Grice, D., Ledbetter, D., Lord, C., Mane, S., Lese Martin, C., Martin, D., Morrow, E., Walsh, C., Melhem, N., Chaste, P., Sutcliffe, J., State, M., Cook, E., Roeder, K., & Devlin, B. (2012). Common genetic variants, acting additively, are a major source of risk for autism Molecular Autism, 3 (1) DOI: 10.1186/2040-2392-3-9

Carnitine & autism: genes, biochemistry & intervention ideas

Carnitine. Mention of carnitine has graced this blog before with regards to how a deficiency in carnitine might show some relationship with some cases of autism spectrum conditions and the knock-on effects with regards to areas such as mitochondrial dysfunction. The whole relationship between carnitine and autism is still very much a work in progress but the data so far opens up some interesting avenues.

A recent study by Celestino-Soper and colleagues* (full-text) continues the interest in carnitine and autism with the suggestion that issues with a gene, trimethyllysine hydroxylase, epsilon (TMLHE) involved in the biosynthesis of carnitine, might be associated with some cases of autism.

I was drawn to this paper on several levels, not least further evidence that amino acid chemistry might show some involvement to cases of autism and the authors' suggestion of another in-born error of metabolism to be added to a growing list in general medicine. The authorship content of the current paper is a veritable who's who in autism genetic research (here and here for example) bearing in mind some overlap with the recent Nature de novo papers.

The paper is full-text but here are a few highlights:

• Based on previous findings of potential issues with TMLHE in cases of autism (here), the frequency of mutations in TMLHE was analysed in cases of autism and controls. Various sources were used to acquire participants including the Simons Simplex Collection (SSC) and the Autism Genetic Resource Exchange (AGRE) covering both simplex (one child in the family with autism) and multiplex families (more than one child diagnosed with autism).
• Deletions of exon 2 seemed to be relatively common in both autism and control groups. So of the control male participants examined, roughly 1 in 366 showed deletions of exon 2. Looking at simplex participants in the autism group, the rate was 1 in 323 presenting with a deletion. When it came to looking at multiplex families, the rate of deletions in the autism group became stronger, estimated at 1 in 130. 
• Based on this cumulative data, the authors write "The frequency of TMLHE deficiency is startling" suggesting that this in-born error of metabolism is relatively common, approximately 20 times more frequent than phenylketonuria (PKU) in males.
• When it came to looking at the functional biological effects of exon 2 deletions, based on the activity of 6-N-trimethyllysine dioxygenase (TMLD), enzyme activity was low or undetectable for those carrying the deletion.
• Urinary and plasma analysis for related metabolites, 6-N-trimethyllysine (TML) 3-hydroxy-6-N-trimethyllysine (HTML) and 4-N-trimethylaminobutyric acid [γ-butyrobetaine (γBB)] suggested alterations in the levels detected according to the presence of TMLHE exon 2 deletions. Some of these metabolites were even suggested to have 'diagnostic potential' for TMLHE deficiency.
• Cognitive function did not seem to determine TMLHE deletions, varying widely among those with deletions.
• The authors conclude that TMLHE deficiency is likely to be a risk factor for autism although with low genetic penetrance.

As per just about every genetics paper that I have ever attempted to read, this is quite a complex story to follow if you are not a molecular biologist or at least with some detailed interest in the ways and means of genetics research. It's not quite the dark art of EEG reading, but pretty close.

The authors make some interesting remarks in the manuscript discussion relating to their findings.

First, they talk about the disparity in their results when it came to simplex and mulitplex family findings. Without making too sweeping a generalisation, the argument goes something like this: having more than one child with autism in the family is more likely to represent a stronger genetic component to aetiology as being involved than those cases of a simplex child, bearing in mind the snapshot and 'what if' scenarios of future offspring attached to looking at simplex families. I can see the logic in this argument about single and multiple instances, and how, when it comes to the eternal question of genes vs. environment as per Trading Places and the $1 bet, the heterogeneity of autism might, to some degree become more homogeneous if autism research were to start looking at simplex vs. multiplex autism rather than bundling it all together.

Second, the authors talk about whether or not the 'risk' of autism could potentially be modified by dietary carnitine intake from birth in the early years, perhaps even prenatally and during gestation when it comes to mum's diet. With some degree of caution, I find myself interested in this suggestion, particularly with all the recent interest in epigenetics and how issues like maternal diet might have the propensity to modify offspring risk for lots of different conditions not just autism. I know epigenetic modification is not a large part of the current study but still, genetics is an awfully complicated business where genes being switched on or off by means other than mutation or deletion might be pertinent.

Finally, this is probably not the last we are going to hear about this area of research by this group. Two studies are being initiated we are told, one looking at carnitine metabolites in cerebrospinal fluid (CFS) of infants with autism with or without TMLHE deficiency; a second following on from other research by Geier and colleagues** on the supplementation of L-carnitine or γBB for cases of autism again with and without TMLHE deficiency. A case of watch this space.

Regular readers to this blog will perhaps know that the genetics of autism are becoming more and more complicated as more research emerges. The concept of an 'autism gene' is becoming an ever more distant memory as a result of the cumulative findings and a realisation that autism is an extremely nebulous condition. What I do like about the current paper is that not only does it detail results based on genetic findings, but more than that, it goes through how those findings relate to biochemistry (expression of the gene) and, in this case, the possibility that there may be ways to 'rescue' some of that biological functioning. I also wonder if through epigenetic means, issues with carnitine biosynthesis might also be present in cases of autism without the underlying genetic structural issues and what those all-important modifying variables might be. All that and the need to screen cases for those ever-so important in-born errors of metabolism, I'll be keeping a look out for further results from this group.

To finish something a little less mind-boggling but nevertheless with some bite: The Clash (the law won y'know and don't they keep telling us).

* Celestino-Soper PBS. et al. A common X-linked inborn error of carnitine biosynthesis may be a risk factor for nondysmorphic autism. PNAS. May 2012.
DOI: 10.1073/pnas.1120210109

** Geier DA. et al. A prospective double-blind, randomized clinical trial of levocarnitine to treat autism spectrum disorders. Medical Science Monitor. 2011; 17: 15-23

Dust settled: autism, de novo mutations and older dads

On purpose I have left it a few days to post about the trio of studies published in Nature by Sanders and colleagues*, O'Roak and colleagues** and Neale and colleagues*** which have created big headlines like this one and this one on quite a few de novo mutations - as in not present in parents but present in children - being linked to the diagnosis of autism. A fourth study picked up by Time magazine (here) by Kerin and colleagues**** on moesin adds to the flavour. This last study coincidentally being in the same publication that contains an article suggesting that the predictive power of personal genome sequencing is likely to be a little bit more limited than many initially suspected. Who'd have thought it?

Back to the Nature articles, I let the dust settle a little for two main reasons: (1) just about everyone has an opinion about this work and its meaning, and (2) my first instinct was to say 'add them to the 2193 genes, 2806 SNPs/VNTRs, 4544 copy number variations, etc' already found and discussed in relation to autism.

I've talked about mutation and genes already quite a bit on this blog and how, whether diagnosed with autism or anything else or nothing at all, we are all a product of mutation and individually carry our own store of genetic mutations. Quite by chance I just saw one of the latest Marvel film adaptations, 'Thor' complete with cameo performance from the magnificent Stan Lee, who has whether knowingly or not, popularised mutation and in some respects relieved it of its quite negative connotations. After all, every kid wants to be Wolverine or Spiderman don't they?

Anyway a very, very short summary of the papers in question:

• Sanders and colleagues* discussed exome sequencing for nearly a thousand individuals, including 200 people with a diagnosis of autism. Exome sequencing, I am reliably informed, relates to the analysis of exons, the parts of DNA which get translated into functional proteins. There is quite a good background description here. Their results identified several de novo mutations in brain-expressed genes to be present, with one mutation in the same gene present in 2 unrelated participants with autism but not in asymptomatic controls among other findings.
• O'Roak and colleagues** carried out similar exome analysis for a couple of hundred parent - child trios (trios implying mum, dad and child with autism, N=677). Their analysis suggested quite a few of the 248 de novo mutations, 126 classified as 'severely disruptive', they found were paternal in origin - from dads - and showed a positive correlation with paternal age consistent with other work on older dads perhaps being a risk factor for autism. I was interested in some elements of the last sentence of this paper abstract which talked about 'extreme locus heterogeneity' but at the same time providing ".. a target for future discovery, diagnostics and therapeutics". Make of that what you will.
• Neale and colleagues*** again sequenced exomes in 175 trios. They reported finding de novo mutations in less than half of their cases (46.3%) stating that ".. the overall rate of mutation is only modestly higher than the expected rate". Nevertheless, some clever proteomics work looking at how these mutation might fit together revealed some interesting interactions between the proteins encoded by the genes being looked at. They also suggested that carrying these mutations might up the risk of autism between 5 - 20 times compared with not having them. I was also interested in one of the gene candidates identified in this study, KATNAL2, which has been tentatively correlated with 'conscientiousness' as a personality trait (assuming you believe that personality is genetic). Dr Ben Neale, the lead author, has also summarised his team's results on a guest blog post here and done to my mind, quite a good job outside of all the hype.

I have probably not been able to do justice to the complexity and obvious work that has gone into these studies with this very short summary. There was some overlap in the genetic areas of interest across the studies which, given the statistical odds involved, might be potentially very important. I must admit that I raised an eyebrow when these papers all came out pretty much simultaneously and only a few days after the CDC reported another increase in their estimates of prevalence of autism in the United States, up from 1 in 110 to 1 in 88 8-year olds - indeed 1 in 32 boys apparently in the State of Utah. I'm sure that the timing was just coincidence though.

I've said it before and will say it again, I am not a molecular biologist or anything related, so am very much an amateur when it comes to decoding the precise meaning of exomes and de novo mutations outside of some background reading. With this in mind, I translate these collective works as suggesting a few things (but don't quote me on this):

1. The genetics of autism are getting more and more complicated with every study published. It probably doesn't help that the diagnosis of autism is a subjective experience from a clinical viewpoint and autism is often surrounded by other comorbidities which are also likely to exert an effect on results. Exactly how this might change if and when the new 'sliding scale' DSM-V autism diagnostic criteria comes out will be interesting.
2. These were studies on de novo mutations meaning that they did not originate from parents. I read one comment on a blog (here) by a commentator who drops by this blog now and again, RAJ, asking the very important question: where did all the heritability go? Does this mean that the autism of today is different from the autism of yester-year or is it all a question of technology, participant numbers, etc.?
3. How and why do these de novo mutations appear is a question that should be on everyone's lips. Random is a word that crops us time and time again in relation to these mutations but with due respect, if we are talking about these mutations showing cause and effect in relation to autism risk, saying they just randomly appear leaves quite a big gap in the knowledge base. Again with my amateur status as a caveat, I do wonder about the involvement of environment and judging by the latest blog from Tom Inset at the NIMH, I'm not the only one. The paternal spotlight and in particular the older dads suggestion from the O'Roak study has put sperm in the cross-hairs, and lets face it, that opens up a myriad of possible environmental factors outside of just 'old sperm' as potentially showing some association (something again that RAJ has commented on in other posts on this blog).
4. As per other studies on mutation in areas such as ADHD, whilst these are impressive studies utilising some impressive technologies, one perhaps needs to ask how common these mutations were in terms of autism as a whole. Yes, there were some interesting areas coming to light and yes, these should be priorities for future studies. As the paper from Neale and colleagues reported however, the overall rate of mutation was only marginally higher than what would be normally expected and one should perhaps not discount the influence of things like intellectual disability as accounting for at least some results (see here).

With the onset of epigenetics (changes to gene expression without changes to the genome), things are changing in the world of genetics. Autism research at the same time continues its fascination with this area of endeavour. The concept of an 'autism gene' covering everyone with autism (and the BAP?) is a distant memory (as are the millions of pounds/dollars/other currency pumped into this area down the years) to be replaced by an altogether more complicated picture emerging of spontaneous mutations, differing genetic profiles for individuals and genes and environment potentially acting variably but synergistically.

To finish, the UK is awash with Britain's Got Talent at the moment, and aside from asking 'where me keys, where me phone' a group called the Zimmers got me reminiscing about an old Beastie Boys classic... fight for your right to [fill in the blank].

* Sanders SJ. et al. De novo mutations revealed by whole-exome sequencing are strongly associated with autism. Nature. April 2012.
DOI: 10.1038/nature10945

** O’Roak BJ. et al. Sporadic autism exomes reveal a highly interconnected protein network of de novo mutations Nature. April 2012.
DOI: 10.1038/nature10989

*** Neale BM. et al. Patterns and rates of exonic de novo mutations in autism spectrum disorders. Nature. April 2012.
DOI: 10.1038/nature11011

**** Kerin T. et al. A noncoding RNA antisense to moesin at 5p14.1 in autism. Science Translational Medicine. April 2012.
DOI: 10.1126/scitranslmed.3003479

Genes, SNPs, CNVs and autism

I open this post with a quote from a relatively new article:

"We show that as many as 2193 genes, 2806 SNPs/VNTRs, 4544 copy number variations (CNVs) and 158 linkage regions have been associated with ASD by GWAS, genome-wide CNV studies, linkage analyses, low-scale genetic association studies, expression profiling and other low-scale experimental studies".

The paper in question is this one (open-access) from Li-Ming Xu and colleagues*.

The first thing that struck me when I read this abstract is wow... autism is complicated. OK, nothing new in that statement and for those regular readers who quite often see me gently tugging at the loose threads of various genetic findings, I suppose summaries like this go to show that the concept of an 'autism gene' is pretty well past distant memory seemingly heading the same direction as Voyager 1. In fact reading through this quite complicated paper, perhaps too complicated for me, several things popped into my mind. Stop me if you heard these before:

• If there was a prize for really knuckling down, collecting and collating data, Li-Ming Xu and colleagues might very well be in the running for it. I say this not only for their paper but also the database they developed on the genetic findings related to autism which is quite brilliant (here). If you have a few hours free (don't we all!), I would definitely recommend browsing the dataset which covers everything including population differences, gender differences and even how the diagnosis of autism was arrived at.
• Yep, lots of genetic findings related to lots of different autisms. When you see the the various studies all laid out and summarised like this, I tend to wonder what similar analyses might look like for other developmental conditions like ADHD, dyslexia, learning disability, etc. How about if researchers conducted genome wide analyses of a group of children randomly selected from your average town or city, how would they compare with these results? How much overlap would there be and how many genes might be associated with this random group? I would wager that there would be quite a bit of overlap in terms of the number of genes, SNPs, CNVs et al related to our random group simply because we are all the product of mutation.
• Heterogeneity is something most people with a connection to autism will know about. If you've met one person with autism, you've met one person with autism is a common saying. Further complicating that heterogeneity is the issue of comorbidity; so like everyone else, people with autism carry varying risks for other conditions. Whether these conditions are more or less likely to be related to some of the collected genetic findings... [fill in the blank].
• Given the wide range of genetic findings in "autism", is this perhaps a sign that the field of genetics needs to start (continue) diversifying; looking at sub-groups or phenotypes for example, a little more closely if it wishes to more ably describe autism or at least certain types of autism. There are already some signs of this happening as exemplified by the paper a few years back looking at paraoxonase gene SNPs in Italian vs. US cohorts.
• As per previous posts, there is a new sheriff coming to town with regards to genetics; epigenetics: changes to gene function not due to changes to DNA. "Hey Pilgrim, you forgot your pop-gun".

Still today some people talk about an autism gene as if right on chromosome X [fill in the blank], findings are going to emerge to say that 100% of people with autism will show issues with gene X compared with 0% of the non-autism population (however you define this as) and that genetic tests for autism are only years away. The problem is that there is no evidence of this either happening now or going to happen any time soon. Indeed as with everything, the more you look at it, the more complicated technology you apply to it, the less you seem to know about it and the more questions start to be asked. Yes, our genes are a big part of us and control lots and lots and lots of different things. But they don't do it in isolation. Crikey, even our gut bacteria might be involved in some processes!

To end a song for Mrs Robinson.

* Li-Ming Xu. et al. AutismKB: an evidence-based knowledge base of autism genetics. Nucleic Acids Research. December 2011.

CNVs and intellectual disability

A very quick post based on an interesting paper published in PLoS Genetics by Girirajan and colleagues* on our old friends the copy number variations (CNVs).

The paper is open access so no need for me to summarise too much but I will say:

• Samples from over 1200 individuals were analysed for the frequency of large CNVs in genomic hotspots. Individuals were grouped into those with dyslexia, autism and intellectual disability (ID) (with an asymptomatic control group also included).
• The appearance of various large CNVs between the groups seemed to suggest that the presence of intellectual disability was the primary correlate. So in the ID group (n=501) the odds ratio of possessing a large CNV was 13.71 with a p-value (significance) going down to about 17 decimal places (highly significant). The autism group (n=350) also showed greater odds of possessing a large CNV (odds ratio = 2.99) although nowhere near as significant as in the ID group (p=0.012). The dyslexia group also relative to controls showed... nothing.
• Dividing the autism group up into those with (n=97) and without ID (n=253) continued the trend towards a relationship between large CNVs and cases with ID, although this was not found to be significantly different between the groups (p=0.102).
• When looking at rare CNVs (less than 50% overlap of CNV length with those found in controls) there was more of an association with autism (10% of participants carrying 36 rare CNVs, odds ratio=6) and also continuing the relationship with ID (16% of participants carrying 77 rare CNVs, odds ratio=10). Dyslexia with regards to rare CNVs again showed nothing compared to asymptomatic controls. Nearly half of the rare CNVs reported in autism or ID groups seemed to be de novo (so not passed from parents).

There is a lot more data in this paper on what was found CNV-wise and in which group. Interesting was the degree of overlap between the results from the conditions looked at; so for example, CNVs in the region coding for the AUTS2 variant being found in two cases of dyslexia. Interesting also that the numbers of cases of autism carrying CNVs is about the same as picked up in other studies (10-11%).

I was struck by the relationship suggested in this study between large CNVs and intellectual disability. Very recently I posted an entry regarding CNVs and ADHD which seemed to arrive at a similar conclusion on the involvement of intellectual disability and CNVs. This combined with the current study and the fact that dyslexia is not usually viewed as an intellectual disability (as opposed to a learning or reading disability) adds further weight to the relationship.

The bottom line I suppose from this study is that autism is a very complicated condition, and the whole really is greater than the sum of its parts in terms of the various comorbidity which can and do appear alongside. CNVs may represent one strand of that complexity but with 90% of cases from this study showing no rare CNVs, perhaps we should be looking elsewhere, perhaps to epigentics and environment for some of the answers?

* Girirajan S. et al. Relative burden of large CNVs on a range of neurodevelopmental phenotypes. PLoS Genetics. November 2011.

ADHD with and without CNVs

As per previous posts on this blog, the shift from the 'one gene condition' hypothesis to that of a more 'lots and lots of genes involved probably with some variable environmental contribution' hypothesis in cases of autism and related conditions seems to have taken hold with some vigour over the last few years.  Lots of fancy acronyms like SNPs and CNVs, combined with words like pleiotropy and epigenetics, have sprung up accompanying our realisation that no matter what your diagnostic label or not, we are all seemingly a product of our various mutations and any interplay between our genes, our mutations, and environment is likely to be complex.

I say all this because of a recent article appearing by Langley and colleagues* which very interestingly looked at cases of attention-deficit hyperactivity disorder (ADHD), comparing those who carried various CNVs against those who didn't and finding, well, very little difference between the cases in terms of things like symptom presentation. Many of the authors on this paper are no stranger to ADHD research, looking at things like comorbidity of bipolar disorder and ADHD and also having some pretty outspoken views about what ADHD might be.

The paper is full-text so I won't start copying and pasting large parts of it, but will provide a short summary:

• Children/young adults (n=567) diagnosed with ADHD (or hyperkinetic disorder) were included for study. Participants were also screened for an autism spectrum condition and excluded from the dataset if autism was suspected.
• The sample was divided up into those with at least one large rare CNV (n=77) and those without (n=490) and compared on various measures.
• Those ADHD cases with CNV(s) presented more commonly with intellectual disability (ID) (IQ<70); although this finding was not mutually exclusive to the CNV group. No other measure or variable of early developmental history or symptom severity was significantly different between the CNV(s) and non-CNV(s) groups (maternal smoking during pregnancy being slightly higher in the CNV group wasn't that far off at p=0.07).

The authors report that this finding points to no presence of an 'atypical' group of children where ADHD is comorbid to CNVs. They also point to a few caveats based on things like sample sizes and power, whether CNVs were de novo or inherited and the 'scattergun' approach they used (looking at CNVs located at various different parts of the genome with various potential weightings in relation to symptom presentation). Bear also in mind that these participants were all white, British children, predominantantly male, and some carried other behavioural comorbidities, and therefore to what group/population the results are probably going to be most applicable to.

I agree with all this issues and tread cautiously as a result. One thing does however stick out from this work about CNVs and ADHD and the question of whether the presence of CNVs might be more related to intellectual disability (ID) rather than other presentations like ADHD or even autism spectrum conditions? Looking at the degree of significance (table 1), the ID difference was pretty stark between the groups (p=0.0001). I know, I know, I am compartmentalising and that is never a good idea (the whole being greater than the sum of its parts and all that). One possibility to answer this question would be to look at an independent sample of children (white, British) with and without ID, screening negative for ADHD and autism as per the same instruments used in the current study and see what the rates of similar CNVs are. The list of CNVs is available at the foot of this article as a supplementary PDF. Indeed how about a similar study comparing groups of children with autism with and without those CNVs (given the suggested overlap) and see if this illuminates the path any better?

* Langley K. et al. Clinical and cognitive characteristics of children with attention-deficit hyperactivity disorder, with and without copy number variants. Br J Psychiatry. November 2011.

Heroes, mutations and diversity

What is your first thought when you see or hear the word 'mutation'? Does it invoke a positive or negative image? Does it suggest 'difference' or does it suggest 'diversity'?

If I'm honest, the first thing I think about when I see or hear the word 'mutation' is the X-Men. I don't know whether it is because that fabulous Marvel comics franchise has just released another X-Men movie or because it takes me back to my childhood reading comic book days about the exploits of a group of people, who for one reason or another are 'different', championing good over evil (or evil over good if you are that way inclined) with some fantastic super-powers to boot.

Indeed, mutation is one of a number of common threads throughout the Marvel series. Spiderman, the Hulk, the Fantastic Four - all mutated by some force or factor; ordinary people turned into extraordinary heroes. Without trying to philosophise too much, the X-Men brand in particular, brings the issues of difference and diversity to the forefront with its discussions on how mutation leads to difference, and how difference can sometimes bring out both the best and worst of human emotions and behaviour. Comparisons have already been made between the X-Men films and autism.

I digress. The reason for the questions and post is the recent media interest in several studies appearing in the journal Neuron. The studies, whose details can be found here, are all related to our old friends CNVs and in particular how finding lots and lots of CNVs paints a very complex picture of the genetics of autism. To reiterate, CNVs are the losses and gains in genetic material.

I am not going to go through all the studies with my 'questioning answers comb' but will rather reflect on the conclusions cumulatively reached from the papers. The first reflection follows my previous CNVs post in that when looking at the genetics side of things, autism is a complex condition.. nay, a very complex condition.  This research kinda confirms what everyone already knows in inferring that there is probably no one set Mendelian pattern of inheritance when it comes to autism - so unlike hair and eye colour, any potential pattern of transmission (either of 'autism' or the various symptoms) is going to be complex. Need some evidence? Have a look at the work being done on the broader autism phenotype.

The second point to make is the fact that where genetic mutations do occur in autism, there is a sizable possibility than some of them are going to be spontaneous in terms of how they come about. I do find some issue with the word 'spontaneous' in that this kinda suggests there is no order or logic behind where or why they occur. My rough-and-ready translation of 'spontaneous' in this context is that the current state of knowledge does not know why something occurs but might do some day. We perhaps had the same issue with PKU before it was known and labelled as PKU as to why some children just 'spontaneously' developed learning disabilities or even worse died. Using the Marvel model, could environment play a role in these spontaneous mutations? Probably, if not why do various animals, including humans, develop drug resistance?

A third point relates to co-morbidity. From what I gather, one or more of the teams looked at children with autism and their unaffected siblings. They then compared and contrasted and found spontaneous CNVs in 8% of the children with autism, compared with 2% of their siblings. OK the differences are not startling but they are there. I have touched upon this point before however in that when we use the word 'autism' all we are really doing is describing a set of diagnostic characteristics derived from behavioural observation and analysis of developmental history. Given that autism does not just 'magically' exist on its own and people with autism as a group carry just about the same likelihood of various co-morbid conditions as anyone else (iron deficiency, diabetes, bowel problems, etc), what's to say that the various CNVs detected were not related to one or more co-morbidites and some perhaps nothing to do with their autistic symptoms? The paper by Gilman and colleagues gives us a clue to this phenomena in action when they say that the identified genes in their study have also been found in intellectual disability phenotypes. Remember also the early test for learning disability.. sorry autism post?

There are other points to make but I will stop there. Don't get me wrong I am certainly not 'geneticist-bashing' when I highlight such issues. Indeed, there are some very interesting questions arising from these collective papers on things like girls and autism genetics. Levy and colleagues ask a very interesting question about the 'fate' of female carriers of the identified CNVs. Could one fate be that of the link between cognitive styles in autism and eating disorders? The points I do however make are that (i) mutation is an important part of human life, and (ii) whilst genetics are no doubt important to autism, these papers reiterate that the role of genes is complex and perhaps not necessarily in isolation from our environment.

I finish with a song about a very different kind of hero.

CNVs in autism

I have always found it a little hard going coming to grips with the various studies on the genetics of autism. Right from the early studies looking at the rates of autism in twins (whether monozygotic or dizygotic) through to the often complicated world of which genes may show linkage to autism, I would probably give myself a D+ (must try harder) in this area of understanding.

What I believe (I think) from the work so far is that: autism is an extremely heterogeneous condition with variable presentation, numerous co-morbidities and most probably a non-Mendelian form of inheritance. In English this means that people with autism are not the same as one another in how their symptoms present and unlike other conditions such cystic fibrosis, the genetic roots are not so easily traceable or identifiable.

In recent years, genetic research in autism has seemingly undergone something of a transformation in terms of what it could do, paralleling research in other areas. Gone are the days where one, two or three genes are implicated; to be replaced with an altogether complex model of variability, overlap and importantly gene-environment interactions (also others epigenetics, epistatis, pleiotropy). Probably without knowing it, readers of this blog will have already been exposed to gene-environment interaction research through my many and varied posts on coeliac disease.

One particular area of genetic research has come to the forefront in recent times - copy number variations (CNVs). OK you say, what on earth are CNVs? Well, the long and short of it is here, but in as few words as possible CNVs are the gains and losses in DNA sequences. Genetics lesson over (phew!). An article published today has caught the eye of quite a few people on the blogosphere in its conclusions regarding genetics and autism. The paper by Gai and colleagues was published in Molecular Psychiatry and happily is open access - a little bedtime reading anyone? There is quite a readable summary of the study on ScienceDaily.

What did it show? Well first of all is suggests that autism is a complex condition. No really, autism is highly complex when it comes to the identification and effects of genes on the condition. So complex in fact that there were just as many CNVs in their non-autism control group as there were in the autism group. What this suggests is that we all carry various deletions, insertions, etc in our DNA, whether we are autistic or not. Next finding: when, during this study, they attempted to test nearly 400 candidate CNVs initially identified in people with autism against another independent set of participants with autism, the grand total of overlap was 11.1%. In other words roughly 10 percent of CNVs from their first autism cohort were present in their second autism cohort. Man, that is complex.

Don't get me wrong, the overlapping CNVs they reported relating to areas such as synaptic transmission are important areas that require further study and replication. The take-home message has to be however that the genetic basis to autism is complicated and currently there are no consistent candidates which differentiate an autism diagnosis from a non-autism diagnosis (so far).

This is a good study - it had a decent initial sample size, control groups and importantly a second cohort of participants with autism to test their initial findings. Autism research could learn a lot from this kind of methodology. I do wonder whether such work is however fundamentally flawed. Not because of the wide heterogeneity present in their sample groups with autism but because of the logic that they are just comparing autism and non-autism; which they aren't. As we know, autism is a collection of symptoms moulded into a diagnostic label. Underneath autism are the same genetic nuances that we all have; our susceptibility to other diseases and conditions, our individual diversity in terms of physical characteristics (why is my nose this big?) and our intimate relationship with our environment (why did I get ill after that and he didn't?). Until we are able to definitively plot some deciding variable (or variables) which says that 'this is autism' and 'this is not-autism' outside of just presented behavior and developmental history, we are stuck with the influence of all these other variables biasing and modulating our findings.

I do wonder whether we will ever get to the point where we can answer questions on genetics and autism given the complexity of the whole thing; or indeed whether we should just accept that the genetic diversity in autism is the same as the genetic diversity in non-autism and look to more functional objectives.