One of the overarching questions about autistic spectrum disorders is how much of the cause is genetic and how much is environmental. Today, findings from a study of Swedish families released in JAMA shed some light on this issue.
Researchers from the Karolinska Institute in Stockholm conducted a nationwide analysis of all children born in Sweden between 1982 and 2006 who had at least 1 sibling. They compared the risk of autism and autistic spectrum disorder among identical twins, fraternal twins, full siblings, half-siblings, and first cousins.
As expected, they found that the probability of developing autism spectrum disorder by age 20 years was higher in children who had a relative with the disorder than in those without an affected relative: 59% for an identical twin, 12.9% for a full sibling or fraternal twin, 8.6% for a maternal half-sibling, 6.8% for a paternal half-sibling, and 2.6% for a first cousin. For the control group (children with no relatives with autism spectrum disorder), the probably of developing the disorder was 1.2%.
The investigators also used various models to quantitatively estimate the strength of the genetic vs environmental factors contributing to the development of autistic spectrum disorder, based on the differing levels of genetic relatedness and shared environments. They found that overall, genetics and environment each contribute half, and that nonshared environmental factors (such as birth complications) play a more important role than shared environmental factors (such as diet).
Lead author Sven Sandin, MSc, from the Institute of Psychiatry at King’s College London and the Karolinska Institute discussed these findings with news@JAMA.
news@JAMA: Why did you conduct this study?
Sven Sandin: In autism etiology there is an important knowledge gap. There are early twin studies that show that as much as 90% of the probability of developing autism is due to genetic factors, and if this is true, it leaves little room for environmental factors. Therefore, most autism research has focused on genetic factors, or “heritability.” However, later studies have questioned this extremely high genetic importance. For example, a big twin study from California 3 years ago estimated a heritability of 38%.
The second major reason is that the concept of heritability only takes into account risks on the population level, not the individual risk. For instance, you can have a high heritability, but still have a low risk for disease on the individual level, and vice versa. There is a lack of information on risk for individuals and for families. A common question among parents is, “If my son has autism, what is the risk that my daughter will also have autism?” And earlier studies done to answer this question have not yielded reliable results.
news@JAMA: Let’s discuss these 2 parts separately. For heritability, what did you find?
Sven Sandin: We found the heritability of autism spectrum disorder to be 50%. In other words, genetic factors account for about half of the risk for autism. This is considerably lower than what earlier twin studies have suggested and closer to the 38% reported from the California twin study.
news@JAMA: What about the other 50%? Because one’s genetics cannot be changed, it seems that this other half would be more important and compelling to study.
Sven Sandin: The nongenetic, or environmental, factors are divided into 2 parts: the shared environment between family members, which includes things like location of housing, diet, or other habits common to the family, and the nonshared environment, which is unique to each child, such as birth complications or maternal infections during pregnancy. Although we didn’t look at individual environmental risk factors, our study suggested that it was the nonshared environment that represented the major environmental source of risk. Shared environment does not seem to be important in explaining the risk of autism spectrum disorder.
news@JAMA: But the results from the California twin study seem to suggest the opposite—that shared environmental factors had a substantial influence on autism risk. Why do you think your results were different, and why would they be more reliable than the previous studies?
Sven Sandin: There may be several reasons for this. With data from over 2 million people, our study was larger than previous studies. Our study also used data from an entire population with equal access to health care and equal clinical diagnostic criteria for autistic spectrum disorder, which should minimize biases due to sample selection, case ascertainment, and differences in follow-up. We also adjusted for the increasing prevalence of autism over the years. In the 1980s, the prevalence of autistic spectrum disorder was approximately 0.5%, whereas today it is greater than 1%. This increase in prevalence can introduce a higher degree of similarity between sibling pairs (both twins and nontwins), which has the potential of inflating the shared environmental contribution. We adjusted for these differences in prevalence, but it is not clear how the California study considered these effects. As a result, we believe that our study achieves a higher degree of accuracy and generalizability.
It’s also important to note that the while the California study suggested a relative contribution of the shared environment of 55%, the confidence intervals were wide, ranging from 9% to 81%. So it’s possible that as little as 9% of the environmental contribution was coming from the shared environment, which would make those results not so different from ours. Finally, we should not rule out that there may be true differences between countries and geographic regions.
news@JAMA: For the second part, how would the findings from your study inform the question, “If my son has autism, what is the risk that my daughter will develop it?”
Sven Sandin: We found that the “relative risk of recurrence” for autistic spectrum disorder is higher for individuals who are more closely related. For example, compared with children who have no autistic relatives, there is an approximately 10 times higher risk of autistic spectrum disorder in children with a full sibling with the disorder, 3 times higher for those with a half-sibling, and about 2 times higher for those with a cousin. Looking at this concept of relative recurrence risk in addition to heritability provides a more complete picture.
news@JAMA: How are these findings useful for the research and clinical communities?
Sven Sandin: With 1 in every 100 children developing autistic spectrum disorder, these numbers can counsel and guide parents and clinicians in understanding the risk for the disorder, which is very important. Also, our findings highlight the importance of the nonshared environment, and the need for more population-based studies looking at these nonshared environmental factors in particular.