Verified by 
Key points
- Michael Garvin and his colleagues at Oakridge National Laboratory have demonstrated a new way to determine a genetic basis for ASD.
- While it's long been accepted that Autism Spectrum Disorder is heritable, no studies have been able to offer a genetic explanation.
- The new study looks at how genetic mutations can offer insight into the heritability of ASD.
I’m always intrigued by the work of writers, filmmakers, and artists whose work, like mine, has been influenced by their siblings with disabilities.
Right now, I’m particularly excited about the work of a sibling in the sciences who also happens to be my brother.
This month, the journal Human Genetics and Genomics Advances published a study by my brother, Michael Garvin, and his colleague, David Kainer—both computational biologists at the Oak Ridge National Laboratory. The paper, “Structural variants identified using non-Mendelian inheritance patterns advance the mechanistic understanding of autism spectrum disorder,” demonstrates a new way to determine the heritability of Autism Spectrum Disorder (ASD)—specifically by using previously unstudied or even ignored rare genetic variants.
“The heritability of autism spectrum disorder, based on family studies, has been estimated to be between 50 percent and 90 percent” they wrote. “A recent study of more than 2 million individuals and 680,000 families from multiple countries provides a best estimated of 80 percent, yet like many complex diseases, very little of this heritability has been explained by significant genome-wide association study loci despite ever-increasing sample sizes.”
Garvin and Kainer looked for genetic markers of autism by considering genomic mutations, also known as structural variants—a new approach to explain ASD heritability. They believe their study shows, “a clearer mechanistic understanding of ASD.” They hope their discovery will lead to better diagnostics and drug therapies for children diagnosed with ASD.
To date, ASD has been diagnosed by developmental observation and screening. According to the Centers for Disease Control and Prevention, children can be diagnosed as young as 18 months of age and reliably by the age of two. The CDC’s Autism and Developmental Disabilities Monitoring Network estimates that ASD affects 1 in 44 eight-year-old American children across all racial, ethnic, and socio-economic groups. Boys are four times as likely to be identified with ASD as girls.
Michael is a computational molecular geneticist in a family of humanities majors. Consequently, he struggles to translate his work into terms we can understand. He’ll often try, wearing a patient look like he’s talking to a fourth grader, to break down some rudimentary concept of DNA that I will never grasp. In the spirit of clarity, I asked him to talk about his latest study and its implications.
Eileen Garvin: Can you explain what you did with this study of the heritability of ASD?
Michael Garvin: The first thing people need to understand is that somewhere between 50 and 80 percent of autism is “heritable,” or due to genetics. To put that in perspective, human height is 80 percent heritable. The second thing is that these estimates are very accurate because you can measure how “genetic” something is without genetics. We’ve been doing this since the 19th century in things like dairy cows and agricultural products because a proxy for genetics is just how related individuals are. Identical twins are more likely to be the same height than siblings and siblings are more likely than cousins, who are more likely than second cousins. So we can simply observe the presence of autism in millions of families and correlate it to how related individuals are.
EG: What does your study suggest regarding diagnosing ASD?
MG: Now that we have established that autism is mainly due to genetic factors, why can’t we find them? There is currently no set of genetic markers that are consistently found in people with autism compared to those without autism that we can use for diagnostics like we do for things like cystic fibrosis or other biomedical conditions. This is known as the “missing heritability problem.” It’s common to many human traits like bipolar disorder, ADHD, and addiction.
The current explanation for this “missing heritability” is that thousands of genetic changes (i.e., mutations) that are common in the general population each contribute a tiny proportion of variation to generate autism and, therefore, our current computational models are inaccurate because that is difficult to capture. What we’re saying with this paper is that there are genetic changes that are frequent in the autism population, but rare in the general population that contribute large effects and could be used for diagnostics. In addition, the functions of those genes give us insight into the biology of autism. The reason this information has not been included in previous studies is because the mutations themselves are complicated and the current sequencing technologies have difficulty identifying them. So the standard practice is to ignore these regions of the genome with the assumption that there is enough information in the easily catalogued variation nearby to explain what is going on. But that hasn’t worked—this is the “missing heritability” I mentioned above. When we do this, we are literally removing genetic information from the study that is being passed on from parent to child, or more specifically, we are generating some of the missing heritability ourselves.
EG: How do you hope your study will affect ASD research?
MG: Our hope is that others who are researching ASD—and many other neuropsychiatric conditions such as ADHD, bipolar disorder, and schizophrenia—will use the approach we have used here to incorporate this genetic variation into their studies. We used previously published data available at the National Institutes of Health to show the proof of concept and identified a gene variant and biological pathway that is associated with non-verbal forms of autism. That was the only clinical information from the study, which is more than a decade old. There are now substantial efforts, like those at the Northwest Autism Center in Spokane, Washington, to more accurately define each case of autism. With that level of detail, we can start to get at this idea of “personalized medicine” in autism. We can catalogue these genomic structural variants at the individual level and then look to see which traits correlate with which genes. That may allow us to identify, for example, the genetic reason some kids have auditory processing disorder.
EG: How do you hope your research will affect families of children with autism?
MG: In the short term, it may provide, for the first time, accurate diagnostics based on genetics, which is less prone to bias and is (theoretically) available in most communities. Assembling a team of specialists to diagnose a child in rural compared to urban America is very different. Once we have identified the genetic changes that are best, a diagnosis becomes a simple cheek swab to get a DNA sample that can be sent anywhere for testing. Accurate diagnoses would open a set of social tools for parents and would do so much earlier than is occurring now. With expanded studies that use this approach, we can link things like visual and auditory functions to the genes that are affected in each child, which can provide useful information for behavioral therapy and learning. Farther down the road, this may even lead to pharmaceutical interventions for some individuals. The biological pathway I mentioned that we found was associated with non-verbal forms of autism in our study has been the focus of drug development for decades, but the enzyme we identified—ACMSD—was not their focus. In the context of neurobiology one can now see why it may be better to target this enzyme because its job is to convert a neurotoxin to something that is neuroprotective.
EG: What are you working on now?
MG: We’re applying this approach that we detail in the publication to many other biomedical conditions and finding individual level solutions. As with the autism study we did, there are previously published data I can use for obsessive-compulsive disorder. We are also using family-level data from direct-to-consumer genomics data like 23andMe as well as biobanks to look at rare diseases. A major take-home message here is that what we showed is likely the case for many biomedical conditions of interest. The autism work was a great place to start because there is so much family level data out there.
EG: We both grew up with a sibling with autism. How has Margaret influenced your work?
MG: To be honest, autism was the last thing in the world I ever wanted to study. As you well know, our house was pretty chaotic growing up. Lots of laughter and some tears, which you really captured in your book. I love Margs and I gained so much from our upbringing, but I was ready to walk away and headed for evolutionary genetics. I’m kind of a genetics nerd and likely on the spectrum myself, so during a post-doc in Israel, when I was working on the evolutionary genetics of fire salamanders, I was also analyzing 23andMe data to sort out my own genetic-based medical condition (hereditary hemochromatosis). Looking at our entire family, I noticed a pattern in Margaret’s genome that is essentially the core of what we did in this paper. Although I didn’t really want to work in this area, I realized that it was my responsibility to share this if it could help. Now, when I work with families with a child with a special needs disorder, I can see our parents’ faces in those parents. I look at it as a way to give back.
Click here to learn more about Dr. Michael Garvin and his work.
