All Roads Once Led to Rome… But Do All Roads Lead to Autism??


Even though, at face value, cancer and autism may seem like worlds apart, if you take a look at their respective genetics you may find more similarities than you bargained for. And because in many ways the field of autism research trails behind that of cancer research, we may be able to learn quite a lot from the other.

In a recent article by Lawrence et al. (2013), the team studied lung squamous cell carcinoma, reporting that 450 allelic variants occurred at a significantly higher frequency in the tumor samples as compared to normal controls. Of those genes, the researchers noted that:


The research team went on to study other cancer types, each returning very similar results as the lung carcinoma samples above.

The point of the research wasn’t to study mutations in lung cancer however. Though the results of course included genes which have known relationship with tumor development, the researchers expressed concern over a significant portion of the results, declaring that,

“…many of the genes seem highly suspicious on the basis of their biological function or genomic properties”.

Thus, the entire point of the study was to draw attention to the fact that genetics studies in the field of cancer research have reported many strong-association genes with dubious ties to cancer development. The olfactory receptors, for instance, play a doubtful role in the progression of lung cancer considering their general function as chemoreceptors for olfaction. In addition, even though OR genes tend to be quite small in length, they frequently house repetitive sequences which may simultaneously explain their higher rates of mutation as well as their genetic diversity (–there’s a HUGE number of OR’s in mammalian genomes) [1].

Larger genes, including those with large coding sequences or introns, are often characterized by higher mutation rates (hence the larger size due to more expansions and mobile element insertions) and are frequently located at or near common fragile sites [2]. (Just as an interesting aside, approximately 40-50% of genes lie very near fragile sites, indicating that gene location is not randomized and probably shares its origins with the biologic function of fragile sites.)

So what does this mean? As Lawrence et al. suggest, OR and large genes may have higher rates of mutation in general. Therefore, specific precautions need to be taken in scientific analysis to account for the varying rates of mutations across genes in order to avoid false positives in cancer research. They state quite clearly that they believe a large number of genes which have reportedly been associated with various types of cancers are in fact not causally related to the conditions but are, in short, coincidental. Especially in the case of a disease like cancer which is characterized by a progressively disturbed and destabilized genome.

Autism. What does autism have to do with cancer? At the moment, my ideas are still somewhat vague. However, I suspect it isn’t coincidence that those findings which predict false positives in the Lawrence et al. study are also extremely common in autism. For instance,


Starting to sound familiar? Many of the same genes which are turning up as false positives in cancer studies are also turning up in autism research. The difference is that we in the autism research community haven’t yet addressed whether these genes are false positives or not.

It truly begs the questions:

  • How many false positives might we be receiving in our research, especially those mutations which are reproducible across studies?
  • Are these highly mutable genes like CNTNAP2 causally associated with the condition, or are they merely coincidental because they’re more mutable?
  • And if they are coincident, why do these mutations occur more frequently in autism versus controls? It’s not as if autism is a cancer that exhibits progressive genetic instability.

What with our lack of an autism biomarker, interpreting these results are especially confusing because genes relating to the nervous system are often large, unstable, and located near fragile sites, all characteristics which should make one cautious about overstating a relationship to autism or any other neurodevelopmental condition in which they may inevitably arise more often [2]. And because we don’t have a proper definition for autism beyond a behavioral syndrome– in other words, we don’t really understand what’s occurring in the brain to create the behaviors– then any mutations that are brain-related are inevitably thrown onto the list of “possibles”. And it is a very, very long list indeed. It includes numerous proliferative and motility factors, adhesion molecules, transcription factors and regulators, synaptic molecules, just about any kind of neurotransmitter you can think of (glutamate, GABA, serotonin, dopamine, acetylcholine, etc.), redox agents, you name it.

As the title of the post says, all roads once led to Rome, but do we honestly believe that all roads lead to autism? I have my doubts. But until we have a better definition of what’s occurring in the developing autistic brain, I guess we can’t really know for certain.

33 responses to “All Roads Once Led to Rome… But Do All Roads Lead to Autism??

  1. Emily, my bet is that the answer to the last questions, and partially to the first two, will turn out to be ‘Yes, due to on-going activity of (endogenous, but who knows!) retroviral elements’ in autism.

    (Yes I know mice are not humans ☺ and before you say that ERVs are ‘hardly ever sources of new mutations in ourselves’, do we really know all there is to know? And how much IS there to know about ERVs and/or their exogenous cousins and our woes?

    Leaving us with the question of what is provoking that activity to start with but if my horse wins we’ll have a smoking gun and some novel clues for autism pathology and treatment directions.

    • Knowing me, I am fascinated by mobile elements, extinct and active. However, higher levels of ERV transcripts in children with autism may not be indicative of what was occurring during spermatogenesis, oogenesis, or embryogenesis. It could be. But it’s a big jump nevertheless. Embryonic studies are vital, and probably also studies of parental sperm and egg, to determine what the state of mobile element activity is and whether it lends a hand in autism development. Keep in mind that only a small minority of cases of idiopathic autism actually have an mutations significantly associated with the condition. The majority of idiopathic cases do not appear to harbor genetic anomalies. Instead, increases in postnatal ERV transcript numbers may instead be indicative of differences in gene expression which may allow transcription to take place more frequently. Remember that there wasn’t an overall increased burden of ERVs but only of certain types. In any case, it’s fascinating stuff and a wonderful line of research to pursue. I definitely agree. 🙂

  2. Hi Emily- interesting post. Don’t you mean “450 ALLELIC variants occurred at a significantly higher frequency”? Secondly, has the Lawrence work been independently confirmed? I understand that often these gene scanning data are hard to duplicate. All in all quite an interesting connection.

    • Hey, John. Good catch. Total brain fart on my part.

      Regarding independent confirmation of the Lawrence work, it’s extremely new, still only published under a DOI I believe (last I checked). Although the fact they replicated their findings, at least in so far as categories of gene types affected, in other tumor types is reassuring. But as I understand from the article itself, these gene categories have been replicated in previous studies, which was one of the reasons it seems the researchers performed the investigation to begin with: because association of OR genes with cancer, for instance, were highly suspicious. Presumably the program which they’ve designed to address differences in basal mutation rates across different genes has been available online through several version updates. We’re personally hoping to learn from the design of the program and exapt some of the concepts for our own work, currently underway.

    • Here I show my ignorance, as I’m not familiar with the proposed mechanisms of action of ERVs with conditions like MS. Can you give me a brief summary of the theory behind the success of the drug trial?

  3. A new study has reported that FOXP1 variants are associated with autism risk. All common variants are pleiotropic with variable genotype-phenotype expressions. FOXP1 variants have been seen in cancer. leukmia, breat cancer as well as increased rates and risk strongly associated with autoimmune and inflammatory conditions (e.g., systemic lupus erythematosus, rheumatoid arthritis, ulcerative colitis, Crohn’s disease, diabetes type 1, multiple sclerosis, Graves’ disease, celiac disease, nodular sclerosis as well as Hodgkins lymphoma

    • Very interesting. I had a suspicion that false positives (if that’s indeed what they are) may be an issue for many genetics studies of different conditions and that it wasn’t relegated to just cancer and autism. But thank you for the list. FOXP1 is an excellent example in which we need to start asking ourselves, “Is this feasibly related to the phenotype we’re observing, or is it associated with the condition for some other reason we’re not yet seeing?”

  4. Hi Emily, this study I believe provided the basis for the first trial

    This paper provides an excellent overview of possible mechanisms “Evolutionary Aspects of Human Endogenous Retroviral Sequences (HERVs) and Disease”

    Back to autism and its genetic risk any thoughts on this one
    (brings to mind potential interactions between ERVs and ubiquitous exogenous viruses as discussed in Blomberg paper above …)

  5. Thanks Emily 🙂 Some more to consider … esp this one on gene-envir interactions in MS (HLA alleles in autism??)

    And this recent one: “signals from bacteria in the mouse gut could lead to increased EVR expression, but that antibodies usually block those microbial signals. Thus, Kassiotis’s team hypothesized that in the absence of such antibodies in their immunodeficient mice, invading microbes could activate ERV expression. Indeed, when placed in germ-free environments, or fed acidic water that would limit their gut microbes, the immune-deficient mice experienced less viral reactivation…”

    • Am still reading atm, but I will be curious if those with ASC who present with increased subtype ERV loads also tend to exhibit more immune-related symptoms, e.g., autoimmune, allergies, etc.

  6. It would be great to know, yes. In the meantime Persico group (Italians again) is actually looking into possible viral presence in ASD parental sperm and egg!! Although it is not clear from this article if this will include ERV activity … actually not much is very clear from the article, esp not from its google translation, but here it is (the original page in Italian

  7. Emily:
    Th Lawrence study discussed the CNTNAP2 gene (the box in your article). CNTNAP2 variants may be another ‘false’ positive finding as far as autism candidate genes are concerned. What is seldom mentioned is that CNTNAP2 gene variations are present in over 60% of the general population.CNTNAP2 variations are associated with an array of genotype -phenotype corelations including general poppulation risk for Obsessive Compulsive Disorder, developmental language disorder, diabetic nephropathy, auto immune synaptic disorders among others.

    • Somehow that doesn’t surprise me. CNTNAP2 is a HUGE gene, one of the largest, has thousands of mobile element insertions, and is associated with a common fragile site, all which are suggestive of considerable instability. It’s size alone is an impressive indicator! Thanks for the references though, those will be helpful.

  8. Pingback: All Roads Once Led to Rome… But Do All Roads Lead to Autism?? | Cortical Chauvinism·

    • Very interesting. I’m curious where specifically the neurons were located in the brains of the women autopsied that showed Y chromosomes. (I wish articles like this would do a better job of citing the original material.) Anyways, I’d be curious to find out where these neurons were located, since it’s believed that new neurons only arise from the dentate gyrus of the hippocampus and the subventricular zone thereby supplying new neurons for the olfactory bulb. The mention of male chimerism with breast tissue is also very interesting. Thanks for the link. 🙂

    • Thanks, yes I had been aware of the study, which certainly holds relevance to what I’ve been doing. In my next study, utilizing bioinformatics, I hope to add additional evidence to the King et al. work. 🙂

    • That’s very interesting, Natasa, thanks. In fact, I’m working on something related to this topic. Although there’s evidence that breakpoints many CNVs in the human genome also coincide with poly-A tails of Alu elements. So it’s probable that repetitive sequences in general underlie aspects of CNV vulnerability. But I’ll definitely be making use of the article, thanks! –Oh, and thank you for the congratulations! 🙂

Leave a Reply to jaksichja Cancel reply

Please log in using one of these methods to post your comment: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s