Autism Genetics Overview
“Maybe there’s overlap in those pathways among the genetic causes, the epigenetic causes,
the environmental causes. I hope to make some sense of this downstream, across the etiologies.”
Wendy Chung, MD, PhD, professor of Pediatrics at Columbia University Medical Center, researches molecular genetics as it relates to a variety of pathologies, including cognitive disorders, metabolic dysfunction and cancer. This busy researcher also directs a clinical genetics program and directs a research program at the Simons Foundation. We talked to her about the state of autism clinical genetics and the future of autism genetic research.
Interviewed by Jill Escher, April 2014
You work on autism, and many other diseases and disorders, from a research angle as well as in clinical practice. How did you get into these fields?
I went to medical school and graduate school as an MD/PhD, and I thought I would be studying inborn errors of metabolism. But the year I started medical school was the year the human genome project started and while all of my medical school professors were complaining about the spending of millions on the human genome project, I started looking at it more carefully, and I thought by the time I finished my bazillion years of training would about when they finish sequencing the genome, and at that point they’ll need someone who knows what to do with it. So I learned in graduate school and medical school the tools to find genes. It took me eight years to clone my first gene, but eight days to clone my last gene, it’s amazing how rapidly the tools have developed.
But I also learned it was not just about science. The more I took care of patients, the more I wanted to have an impact on them. Making a diagnosis is nice, but we have to get beyond the diagnosis and impact care. My fantasy is to cure a person with a genetic condition.
I recently spoke with a clinical geneticist who said he could find relevant genetic mutations in only about 10% of his autism patients. Has your experience been similar?
It depends on how hard you push. We’ll do a molecular karyotype, a chromosome mircoarray, Fragile X testing, screens for inborn errors of metabolism or metabolic disorders, and the newest big thing that increases yield is doing the clinical exome sequencing, which is a big additional push. With that, the yield goes up somewhere between 20 to 25% and it depends in part on whether this is an isolated case of autism with no family history or whether it’s a familial condition where there’s a higher probability for it being a heritable genetic condition.
(Editor's note: A karyotype is a visual depiction of the number and appearance of the chromosomes. Microarrays are able to detect much smaller genetic changes than routine karyotypes. The exome sequencing refers to detailed analysis of the protein-coding portions of the genome.)
For the families with multiple kids impacted by the same type of severe autism, our yield goes up to 50%.
What do you usually find in the exome sequencing?
We don’t know all the autism risk genes, so you don’t always know for sure when you find something suspicious that it is the cause. Sometimes you’re sure that you found it, and other times you have a hunch that it’s right but you’re not 100% sure until you start identifying other families that have the same type of genetic disorder.
With that limitation, in an exome there are about 10% of individuals with autism in whom you’ll see de novo mutations that appear to be the cause of the autism, a new genetic change in the child but not present in the parents’ blood samples.
Many times when we see a family history of autism, we find an autosomal recessive condition that’s shared between the siblings with autism. We use the power of the family when we can to compare the genetic differences to members of the family who don’t have autism. When we find answer, about 60% of the time we’re finding genes clearly associated with autism but another 40% of the time they are genes that look like they ought to be autism genes but may not have been implicated in autism before. Sometimes they have been implicated in epilepsy or other conditions but not autism per se, sometimes animal models suggest an association, sometimes you have no animal or human data to support your hunch.
You can’t really be confident that a new gene causes autism until you find other families with a similar genetic problem who also have autism. But when you can find those genes that are common across families with autism, you can link families up to learn from each other about what to expect. Now, they can find each other online; it used to be with rare genetic conditions you were isolated.
Have you found anything relating to autism with respect to imprinted genes in particular? (Editor's note: Imprinting is an epigenetic phenomenon by which certain genes can be expressed in a parent-of-origin-specific manner, with one allele of a gene turned on, and the other shut off.)
It’s not just the genes you’re born with, but the effect of the environment on the genes, and the ways some genes are expressed in an allele-specific way. We’ve tried to get at this, though they are very difficult experiments to do, because you need to look at the tissue that is relevant-the brain. The great challenge we have with autism and related neurodevelopmental disorders is of course we don’t have access to the brain.
With imprinting, it’s not necessarily that every tissue in the body is imprinted in the same way. Blood is not always imprinted the same way as the brain. You can use animal models, but there is no substitute for humans.
We know there is no singular “autism.” What can we do to clarify the great heterogeneity within this vast landscape?
I think about this in terms of dimensionality. It’s clearly clustered in terms of core features but along those there are varying dimensions in severity. One that we use that stratifies the population to a certain extent is IQ, it’s certainly not the only one. There are some of the research measures we use that measure severity. We also think about co-morbid conditions, some people with autism have seizures or epilepsy, some have GI symptoms, some have dysmorphic features, physical traits, birth defects, or growth problems. Some individuals will have a family history of neuropsychiatric conditions or autism itself.
In terms of etiology, some of them will have an identifiable genetic etiology, or identifiable features such as prematurity or exposures during the mother’s pregnancy or even the grandmother’s pregnancy. We try to stratify across all of those dimensions. Potentially we could do a cluster analysis where you have all these dimensions and you can see what patients cluster together with similar features. I think in part we hope there might be treatments that can be efficacious in clusters. There may not be treatments where it is one-size-fits-all, but maybe there might be ten different clusters.
How much autism in the population do you think may be attributable to abnormal prenatal events, whether it’s exposures or prematurity or other perinatal complications?
If I could draw a pie chart of how much of the cause of autism we understand, it would be the minority that we understand. We have guesses in individual cases, but what we need are large epidemiological studies to answer these questions. For most of the cohorts we don’t have good data on the moms, much less the grandmothers, during the pregnancy. There are cohorts in Norway and Denmark with well-characterized pregnancy information, where the kids are just getting old enough to be studied.
Of the things we know and understand at this point, less than 10% are caused by prenatal exposures we understand. I think the actual percentage is likely to be higher than that, but we don’t know for sure.
What about post-natal factors?
Our evidence is definitely weaker there. There are circumstances when a child seems to be developing normally and then regresses, and sometimes this coincides with an infection, so people make the cause and effect connection there.
Do you think environmental factors could be involved with some of the de novo mutations seen in autism?
We don’t entirely know what causes those mutations, but I’m sure that as with cancer there could be environmental factors that cause changes in the sperm or the egg. In the occupational health sphere there is concern about radiation and chemicals with DNA changes, but in general we don’t have any hard data to say we know the source of these DNA changes.
There are different mechanisms causing DNA changes so if there were an association you’d expect to see certain types of mutations or changes associated with certain exposures. But right now no one has sufficient information about what agents might be doing to the DNA of sperm or egg, but we’re starting to ask the questions.
Do you have any thoughts about the gender disparity in autism, where the ratio exceeds 4 males to 1 female affected?
Unfortunately, we really don’t know. It’s very clear that there’s a gender difference, the difference becomes more substantial the higher you go up in IQ. If you look at the individuals with IQ of say, over 110, then it ends up being more like 8:1. If you go to the lower end of the range then it’s more like 2:1.
There are theories about nature v. nurture or about the female brain being more resilient, people wonder about genes on the X chromosome, but the real, true, honest answer is we don’t know.
What do you think should be the priorities for further autism genetics research?
We want to understand the biology with an eye toward the treatments. If we know the disrupted genes we can better target interventions. It’s not overnight to go from gene to cure, but hopefully there will be some convergence in the genes we are finding, and hopefully it’s not 500 forms of autism but five or ten forms with common mechanisms. And maybe there’s overlap in those pathways among the genetic causes, the epigenetic causes, the environmental causes. I hope to make some sense of this downstream, across the etiologies.
My fear with autism is that it involves this very early process of neurodevelopment that can’t be reversed in any way. That we should instead focus on prevention.
Yes, prevention is always important, but in addition to prevention, we must think about what can we do to make the lives better for those with autism. Personally, I don’t think you should do just one or the other.
Thank you so much, Dr. Chung, for your insights and your time, I enjoyed speaking with you.
Thank you for the work you are doing on your website to address this scientific niche.
I went to medical school and graduate school as an MD/PhD, and I thought I would be studying inborn errors of metabolism. But the year I started medical school was the year the human genome project started and while all of my medical school professors were complaining about the spending of millions on the human genome project, I started looking at it more carefully, and I thought by the time I finished my bazillion years of training would about when they finish sequencing the genome, and at that point they’ll need someone who knows what to do with it. So I learned in graduate school and medical school the tools to find genes. It took me eight years to clone my first gene, but eight days to clone my last gene, it’s amazing how rapidly the tools have developed.
But I also learned it was not just about science. The more I took care of patients, the more I wanted to have an impact on them. Making a diagnosis is nice, but we have to get beyond the diagnosis and impact care. My fantasy is to cure a person with a genetic condition.
I recently spoke with a clinical geneticist who said he could find relevant genetic mutations in only about 10% of his autism patients. Has your experience been similar?
It depends on how hard you push. We’ll do a molecular karyotype, a chromosome mircoarray, Fragile X testing, screens for inborn errors of metabolism or metabolic disorders, and the newest big thing that increases yield is doing the clinical exome sequencing, which is a big additional push. With that, the yield goes up somewhere between 20 to 25% and it depends in part on whether this is an isolated case of autism with no family history or whether it’s a familial condition where there’s a higher probability for it being a heritable genetic condition.
(Editor's note: A karyotype is a visual depiction of the number and appearance of the chromosomes. Microarrays are able to detect much smaller genetic changes than routine karyotypes. The exome sequencing refers to detailed analysis of the protein-coding portions of the genome.)
For the families with multiple kids impacted by the same type of severe autism, our yield goes up to 50%.
What do you usually find in the exome sequencing?
We don’t know all the autism risk genes, so you don’t always know for sure when you find something suspicious that it is the cause. Sometimes you’re sure that you found it, and other times you have a hunch that it’s right but you’re not 100% sure until you start identifying other families that have the same type of genetic disorder.
With that limitation, in an exome there are about 10% of individuals with autism in whom you’ll see de novo mutations that appear to be the cause of the autism, a new genetic change in the child but not present in the parents’ blood samples.
Many times when we see a family history of autism, we find an autosomal recessive condition that’s shared between the siblings with autism. We use the power of the family when we can to compare the genetic differences to members of the family who don’t have autism. When we find answer, about 60% of the time we’re finding genes clearly associated with autism but another 40% of the time they are genes that look like they ought to be autism genes but may not have been implicated in autism before. Sometimes they have been implicated in epilepsy or other conditions but not autism per se, sometimes animal models suggest an association, sometimes you have no animal or human data to support your hunch.
You can’t really be confident that a new gene causes autism until you find other families with a similar genetic problem who also have autism. But when you can find those genes that are common across families with autism, you can link families up to learn from each other about what to expect. Now, they can find each other online; it used to be with rare genetic conditions you were isolated.
Have you found anything relating to autism with respect to imprinted genes in particular? (Editor's note: Imprinting is an epigenetic phenomenon by which certain genes can be expressed in a parent-of-origin-specific manner, with one allele of a gene turned on, and the other shut off.)
It’s not just the genes you’re born with, but the effect of the environment on the genes, and the ways some genes are expressed in an allele-specific way. We’ve tried to get at this, though they are very difficult experiments to do, because you need to look at the tissue that is relevant-the brain. The great challenge we have with autism and related neurodevelopmental disorders is of course we don’t have access to the brain.
With imprinting, it’s not necessarily that every tissue in the body is imprinted in the same way. Blood is not always imprinted the same way as the brain. You can use animal models, but there is no substitute for humans.
We know there is no singular “autism.” What can we do to clarify the great heterogeneity within this vast landscape?
I think about this in terms of dimensionality. It’s clearly clustered in terms of core features but along those there are varying dimensions in severity. One that we use that stratifies the population to a certain extent is IQ, it’s certainly not the only one. There are some of the research measures we use that measure severity. We also think about co-morbid conditions, some people with autism have seizures or epilepsy, some have GI symptoms, some have dysmorphic features, physical traits, birth defects, or growth problems. Some individuals will have a family history of neuropsychiatric conditions or autism itself.
In terms of etiology, some of them will have an identifiable genetic etiology, or identifiable features such as prematurity or exposures during the mother’s pregnancy or even the grandmother’s pregnancy. We try to stratify across all of those dimensions. Potentially we could do a cluster analysis where you have all these dimensions and you can see what patients cluster together with similar features. I think in part we hope there might be treatments that can be efficacious in clusters. There may not be treatments where it is one-size-fits-all, but maybe there might be ten different clusters.
How much autism in the population do you think may be attributable to abnormal prenatal events, whether it’s exposures or prematurity or other perinatal complications?
If I could draw a pie chart of how much of the cause of autism we understand, it would be the minority that we understand. We have guesses in individual cases, but what we need are large epidemiological studies to answer these questions. For most of the cohorts we don’t have good data on the moms, much less the grandmothers, during the pregnancy. There are cohorts in Norway and Denmark with well-characterized pregnancy information, where the kids are just getting old enough to be studied.
Of the things we know and understand at this point, less than 10% are caused by prenatal exposures we understand. I think the actual percentage is likely to be higher than that, but we don’t know for sure.
What about post-natal factors?
Our evidence is definitely weaker there. There are circumstances when a child seems to be developing normally and then regresses, and sometimes this coincides with an infection, so people make the cause and effect connection there.
Do you think environmental factors could be involved with some of the de novo mutations seen in autism?
We don’t entirely know what causes those mutations, but I’m sure that as with cancer there could be environmental factors that cause changes in the sperm or the egg. In the occupational health sphere there is concern about radiation and chemicals with DNA changes, but in general we don’t have any hard data to say we know the source of these DNA changes.
There are different mechanisms causing DNA changes so if there were an association you’d expect to see certain types of mutations or changes associated with certain exposures. But right now no one has sufficient information about what agents might be doing to the DNA of sperm or egg, but we’re starting to ask the questions.
Do you have any thoughts about the gender disparity in autism, where the ratio exceeds 4 males to 1 female affected?
Unfortunately, we really don’t know. It’s very clear that there’s a gender difference, the difference becomes more substantial the higher you go up in IQ. If you look at the individuals with IQ of say, over 110, then it ends up being more like 8:1. If you go to the lower end of the range then it’s more like 2:1.
There are theories about nature v. nurture or about the female brain being more resilient, people wonder about genes on the X chromosome, but the real, true, honest answer is we don’t know.
What do you think should be the priorities for further autism genetics research?
We want to understand the biology with an eye toward the treatments. If we know the disrupted genes we can better target interventions. It’s not overnight to go from gene to cure, but hopefully there will be some convergence in the genes we are finding, and hopefully it’s not 500 forms of autism but five or ten forms with common mechanisms. And maybe there’s overlap in those pathways among the genetic causes, the epigenetic causes, the environmental causes. I hope to make some sense of this downstream, across the etiologies.
My fear with autism is that it involves this very early process of neurodevelopment that can’t be reversed in any way. That we should instead focus on prevention.
Yes, prevention is always important, but in addition to prevention, we must think about what can we do to make the lives better for those with autism. Personally, I don’t think you should do just one or the other.
Thank you so much, Dr. Chung, for your insights and your time, I enjoyed speaking with you.
Thank you for the work you are doing on your website to address this scientific niche.