New Findings on ASD Traits in Siblings
Significant new research findings just published in the American Journal of Psychiatry include a higher than expected rate of language delay, sometimes with autistic features, in siblings of children with autism spectrum disorders (ASDs). The study, which was based on information submitted by families participating in IAN Research, also provides evidence that girls with ASD or ASD-like traits are being under-identified. IAN interviewed lead author John Constantino to learn more. (IAN questions appear in boldface.)
John Constantino, M.D., Professor of Psychiatry and Pediatrics at Washington University in St. Louis, was the developer of the Social Responsiveness Scale (SRS) - a key instrument that measures autistic social impairment and one of the online questionnaires families participating in the IAN Research project complete.
Dr. Constantino, congratulations on the publication of your paper, and thank you so much for agreeing to share your findings with our IAN Community readers.
I am happy to do so. It is thanks to the families participating in the IAN Research project that this study was possible. I wanted to let them know how much I appreciate the time they spend answering IAN's online questionnaires, because we really do use this information to advance autism research. The study my team just published was based on the input of 1,235 families who had filled in IAN questionnaires for nearly 3,000 affected children and unaffected siblings.
Your paper, which is called Sibling Recurrence and the Genetic Epidemiology of Autism, * contained several important findings. Before you tell us about those, may I ask what you set out to explore, at the beginning?
We set out to do a definitive study on sibling recurrence in autism using quantitative measures, not just pre-existing diagnoses.
Let me explain. We suspected that there are mild autistic traits that frequently occur in other members of families with a child on the autism spectrum. If that is true, it has implications for all kinds of research, including genetic research. Imagine that a researcher is comparing children with autism spectrum disorders (ASDs) and their siblings, and assuming that the siblings are completely unaffected, with no autistic traits at all. But what if some of those siblings have subtle autistic traits, and we are counting them as "completely unaffected"? Researchers might draw wrong conclusions, or miss important clues.
I know that recurrence, or how often siblings in the same family as a child with ASD also have ASD, has been studied in the past. How was your study different?
Previous studies looking at sibling recurrence depended on actual diagnoses. We call those "categorical" diagnoses because a person has been placed in a category. This one has autism, that one has Asperger's syndrome, and that one has no diagnosis at all. The trouble is: categories like this are "all or nothing." You have an ASD, or you don't.
Now, when you try to discover how many families have more than one child on the spectrum, this is a problem, because what if children haven't been diagnosed, but have some or many autistic features? It would be a mistake to categorize them as "unaffected" when they actually may have some mild autistic traits, or even an ASD.
We wanted to measure ASDs in a way that would be quantitative. In other words, each child would get a number or score for how many autistic symptoms he or she has. The Social Responsiveness Scale, or SRS (a measure of current symptoms), and the Social Communication Questionnaire, or SCQ (a measure of developmental history), allow us to do that. Parents answer a number of questions about their child's development and behavior, and this is translated into what you might think of as an "autism symptoms score."
For example, we know if you have a very low SRS score, you probably don't have any autistic traits, and if you get a very high score, you likely have a diagnosable ASD. But the SRS also lets us see when someone has some autistic features, scoring in between individuals with a full blown ASD and those with no autistic features at all. It gives us the ability to measure "sub-clinical autism" and helps resolve the "all or nothing" constraint.
In essence, we wanted to find out if non-diagnosed siblings of children with ASDs looked just like children in families with no history of ASD whatsoever, or if some were mildly affected.
How did IAN data help you tackle the problem?
To explore this question, we needed a very large sample of children with ASD and their siblings, and SRS scores for all of them. To go out and find these families, to recruit them to participate and administer the SRS and other questionnaires, would have been incredibly expensive and difficult. Thanks to IAN, and the families who had joined and donated their time to answer online questionnaires, including the SRS, the data were already there, at no cost to us. I cannot stress how amazing this is, and how grateful we are to the families, because we made several important findings.
I know one of those findings involved differences in simplex families (where a family has only one child with ASD) and multiplex families (where a family has more than one child with ASD). Can you tell us more?
Yes. We started by comparing the supposedly unaffected siblings in two kinds of families: those with only one diagnosed case of ASD and those with more than one. At this point, you will note, we were using diagnosed cases to decide if a family was simplex or multiplex.
Even doing it this way, we discovered that most of the siblings in simplex families looked just like children in the general population. They had very low SRS scores, just like you'd expect a child in a family with no ASD whatsoever to have.
But siblings in the multiplex families looked different. Some of them scored like the typical children in the general population did, but some scored high enough to warrant a possible ASD diagnosis, while others scored in between because they had some mild autistic traits.
In addition, among siblings throughout the sample, we discovered that mild ASD features occur about 10 times more frequently than they do in the general population. This provides some evidence that some specific autism spectrum traits can be inherited even when the whole syndrome is not.
You also looked at language delays in the siblings. What can you tell us about that?
Yes, that was an additional finding. Looking at the data, we found that 20% of the presumably unaffected siblings had been diagnosed with some type of language disorder. This came as somewhat of a surprise, although it has been reported in a number of smaller sibling studies involving families affected by autism. Out in the general population, the percentage of children that are diagnosed with a language delay is only about 7- 9%.
In addition, we found that half of the children with these language disorders had specifically autistic qualities of speech. This would include things like pronoun reversal (where they say you when they mean I), odd or repetitive phrases, and asking socially-inappropriate questions. Most of the children with these autistic features of speech also scored as having mild autistic social impairments.
In other words, some of the children had a history of language delay with autistic speech, and elevation in levels of autistic social impairment, but were not categorically diagnosed as autistic. This indicates some kind of genetic "loading" for these features, which occurred in up to 25% of families affected by ASD.
It should be noted that, depending on which way we define "having ASD," the number of families with more than one child on the spectrum changes:
- About 11% of families had more than one child with a diagnosed ASD, that is, they were officially multiplex.
- 12.5% of families had a sibling with a standardized score of 75 or more on the SRS, which is generally high enough for boys to be diagnosed with an ASD (but not always for girls).
- Nearly 11% of families had an "unaffected" sibling with a history of speech delay and autistic features of speech.
- All together, fully 26% of families with a child on the spectrum had another child who had a diagnosed ASD, scored in the "diagnosable ASD" range of the SRS, or had a history of speech delay with autistic features of speech.
This changes our past understanding of rates of recurrence, and may shift our notion of which families are simplex and which are multiplex.
I understand you had some especially interesting findings about the girls in your sample.
Yes. When it comes to ASD diagnosis, our results provide the most stringent evidence to date that girls with ASD susceptibilities may be missed when undergoing a typical diagnostic screening or evaluation. There were far more girls than boys in our sample who had very high SRS scores, such that you would expect them to have been diagnosed with an ASD, but who had not received such a diagnosis. In fact, about 40% of the girls who scored very high (in reference to the general population distribution for girls) had not been diagnosed.
This suggests that there must be some bias in how diagnoses are being made. Whether girls with ASD have a different profile that is not reflected in diagnostic criteria, or whether girls who are very socially impaired compared to other girls do not exhibit behaviors as extreme as those of very socially impaired boys, it is clear that girls with ASD-related developmental abnormalities are under-identified. It would be similar to a pediatrician using the "blue" growth chart for boys to measure girls' growth, too. There's a reason we have different growth charts for boys and girls. We're looking at norms based on the whole population, and we know what is typical or extreme for each gender in terms of height or weight at a certain age. A girl, who is very tall for a girl, may not look that tall if measured with the boys' growth chart. In the same way, a girl who scores very high for a girl as far as ASD traits may not look that impaired when compared to a boy with ASD. But she is still scoring extremely high compared to the rest of the girl population, and may benefit from clinical help and support.
This has implications for the gender ratio in autism. What has been said for years is that for every 4 boys diagnosed with ASD there is only 1 girl, and much attention has been paid to finding out why there is such a skewed gender ratio. What we found suggests that the gender ratio in autism is far less pronounced, more like 3 to 2 than 4 to 1.
What are the implications of your study as a whole, especially for the genetics of autism?
The genetic picture is still very complex, and we are just beginning to untangle it. We do know from a variety of earlier studies that truly unaffected parents sometimes have a child with ASD. That's what researchers refer to as simplex or "sporadic" autism -- most often, the siblings in those families are truly unaffected as well. However, it must always be considered that at any given stage a family may simply not be large enough to exhibit recurrence among siblings. We also know that there are families in which many individuals exhibit ASD and/or autistic traits, and that these traits appear to be inherited.
Children with mild traits, like some of the siblings in our study, may hold the key to resolving crucial questions. They may help us better understand how autistic traits are inherited, and how some children with autism susceptibility (i.e. those with milder manifestations of recurrence) are spared the full autistic syndrome. Understanding precisely what protects such children may lead to new and better interventions for those who will otherwise develop more severe impairments as a result of the same susceptibility factors.
Thank you so much for taking the time to share your findings with our readers, Dr. Constantino.
I am happy to do it. And thank you again to all the families participating in IAN Research!