Home > News & Opinion > In Brief > 2012 > Genetics: Autism risk higher in full than in half siblings

Genetics: Autism risk higher in full than in half siblings

Popularity tracker
Jessica Wright
23 March 2012

Common fate: A child has about a 5 percent chance of developing autism if their half sibling has the disorder and 10 percent if their full sibling does.

Full siblings are twice as likely as half siblings to share a diagnosis of autism, according to a short report published 28 February in Molecular Psychiatry1. The results suggest that genetic factors play an important role in the risk of developing autism, the researchers say.

Studies generally compare the rate of autism recurrence, or concordance, in identical and fraternal twin pairs to parse out the contribution of genes to the risk of developing the disorder.

Twin studies are limited, however, because they are generally small. They are also complicated by the fact that environmental factors, such as the presence of maternal antibodies that attack fetal brain proteins, would affect both twins during gestation, the researchers say.

To address these limitations, researchers in the new study looked at 5,237 families, of which 619 have maternal half siblings, 55 have paternal half siblings and 4,832 have full siblings. The families are enrolled in the Interactive Autism Network, or IAN, a national, voluntary autism research database, because at least one of their children has autism. The concordance rate is about five percent for half siblings with the same mother, zero percent for half siblings with the same father and about ten percent for full siblings, the study found. 

Taking into account the siblings’ ages, sex and various estimates of the population prevalence, genetic factors account for about 50 to 70 percent of the risk of autism recurrence, the researchers say.

The researchers analyzed another database of families with children who have autism, which contains a high proportion of ethnic minorities, and found similar concordance rates: 11 percent for 132 full siblings and 7 percent for 41 maternal half siblings.

The concordance estimate of 10 percent for full siblings is lower than the 30 percent figure previously shown for fraternal twins, the researchers note. This difference could be due to environmental factors that influence the development of both twins in the womb, such as maternal stress or infection, they say.


1: Constantino J.N. et al. Mol. Psychiatry Epub ahead of print (2012) PubMed


Name: Paul Whiteley
23 March 2012 - 2:48PM

Twin studies are still important to autism research (and many other conditions) although the whole area of genetics and heritability is perhaps undergoing a bit of a shake-up with the rise (and rise) of epigenetics. Slightly outside of autism but this article (full-text) provides fodder for some interesting discussions on how even those sharing the same genome might not necessarily present with the same pattern of genetic expression as a result of various different factors:

Name: MJ
23 March 2012 - 7:50PM

Maybe it it just me, but I have to wonder about the difference in risk between mothers and fathers. Why would the risk for fathers drop back down to close the normal population risk but mothers stay elevated yet less than normal siblings?

If both parents are contributing to the risk then it seems like both parents should have an increased risk. If it was just the mother that was the risk factor then it seems like the maternal half sibling rate should be much closer to the full sibling rate.

The unequal distribution of risk makes me thing that this data is showing that something other than genetics is at work here.

Name: Jessica Wright
23 March 2012 - 8:29PM


Thanks so much for the comment, that is a very good point.

The number of half siblings in the database living together with their father is significantly lower than those living with their mother: 55 vs 619.

A simple answer is that fifty-five families are not large enough for a 5% rate to be statistically valid, as the standard error could easily push it down to 0%.

However, maternal factors, such as the presence of anti-brain antibodies in the mother are believed to increase in the risk of autism, according to some studies. This could also drive up the concordance rate in siblings with the same mother, compared to those with the same father.

However the low number of paternal half-siblings in the study is probably not large enough for the researchers to make any conclusions about the differences between the paternal and the maternal rate.

Thanks for your careful reading of SFARI.org,


Name: MJ
23 March 2012 - 10:21PM

OK, fair enough. I get that the 55 paternal half siblings in the IAN data might be too small to reliably detect the true rate.

But the same pattern also appears in the second data set. In that data set there are only 41 maternal half siblings and 31 paternal half siblings yet the maternal set still manages to show a higher risk (3 of 41 vs 1 of 31).

Perhaps the second set isn't really large enough to be significant but it is presented as a replication set in the paper. So it seems a little odd to consider the replication of the maternal finding to be of some significance but not the paternal one.

Name: M
4 April 2012 - 11:52PM

It is easy to get bogged down with stats. There is another science at play in the numbers we call stats. Anyone who has been tested and measured will understand that converting real analogue into digital is not an 'exact science' .So the stats expert makes a set of extra rules for interpreting the stats. Even with this expertize, the understanding of stats allows scientific comments such as ' There is no evidence' to become a legal commodity.There are always contradictory stats- I suggest that the evidence speaks for itself. The 'Genetic' model science is distracted with in these troubled days still holds sway but I am sure common sense is sufficient to establish that genetics is and always will be at the root of every biological issue. That is not to say that a susceptible gene is the main cause of its link to a statistical concurrence. So the science invents the epi-genics as a way to re-understand the model. Understanding 'complexity' is something more difficult. The science of chaos is the real numbers game. Do I have any takers?The politics of science has a rough history.

Name: Steve White
13 May 2013 - 1:50PM

The lead into the article says the concordance rate for full siblings is twice that of half siblings.
I am just a parent/layman, so maybe I don't understand this at all but that finding does not seem to jibe with other things I read.
Seems like this implies a large percentage of autism is caused by dominant genes:
Man and Woman have child. One of them has dominant autism causing gene. Or, it could be more than one gene but if more than one they are on the same chromosome.
Kid is autistic. Subsequent siblings have essentially the same risk.
But, half siblings have only half the chance of getting the same chromosome from the parent, so their risk is half that of full siblings.
Well, I may be misusing the term "dominant" here, I don't mean everyone with the gene absolutely has to be autistic, there might still need to be environmental factors, but they would average out the same for a large sample size. (when you look at hundreds of study subjects, the same percentage of Moms had a fever the same week of pregnancy in the half siblings as full siblings and so forth)

But my point is, that does not seem to agree with so much that I read.

It seems like they can't find the gene or gene combinations that are always part of some autism cases, seems like this does not account for CNVs which I read were thought to cause a large percentage of case, seems like it does not account for maternal antibodies which also are thought to account for a large percentage, and so forth.
Reading the article it seems like the "half as much risk for half siblings" is a bit oversimplified maybe.

Add a Comment

You can add a comment by filling out the form below. Plain text formatting.

Question: What is 10 + 4 ?
Your answer: