Facial phenotypes of autism?

Since a small boy, many moons ago, I have always had a fascination with cars. I'm not talking about the inner workings of cars - I barely know a spark plug from a fan belt - but rather how the front of cars (lights, grill, etc) share more than a passing resemblance to faces and expressions of faces. Odd I know, but trawling the Internet it is a relief to know that others share this... this... 'interesting' hobby like this article in the LA Times on the 10 most sinister car faces. Seems Disney might have been on to something with the Cars franchise?

Putting aside whether you drive a 'sinister' car or perhaps a car with a more happy go-lucky face, the type of face cars have is very often a good indicator of the brand and other cars made by the same manufacturer. Without name-dropping most people would recognise a BMW face as a BMW face and a Ford face as a Ford face. But what about if you moved from the faces of inanimate objects to human faces. How about faces serving as phenotypes or 'markers' of conditions like autism?  

A recent study by Alridge and colleagues* (full-text) cropped up on my research radar. I was initially a little weary about posting on this study for fear of being misconstrued. Differentiating faces based on groups has a rather dark recent history and studies which attempt to classify in this way are always going to have this cloud lurking nearby. In the end I have decided to post and the Aldridge study details some interesting findings.

Because the paper is open-access I won't summarise in too much detail what was done and what was found. Suffice to say that 105 boys (n=64 with autism) aged between 8 and 12 years old were investigated using the 3dMDcranial System to map their faces. After some statistical wizardry, face coordinates and differences between them were used to see if the autism group could be differentiated from controls and whether there were any correlations between autistic faces and clinical and behavioural markers. The results: yes, 39 of 136 total linear distances were statistically significantly different between autism and control groups around the nose, philtrum and the eyes but the results overlapped between the groups. Authors also discussed a couple of sub-groups based on autism severity scores and diagnosis given that might be present.

I was interested to see that the authors suggested that the results seemed to match their clinical opinion of autism as per a quote: "This facial phenotype is similar to one we recognised clinically..". Interesting also on their link to the facial phenotype described and Kanner's original description of a 'beautiful face' being associated with some cases of autism.

This is not the first time this kind of research has been done. Then as now, the suggestion was that brain and other tissues develop in tandem with each other and there is much genetic and signalling overlap including Sonic hedgehog (which was mentioned a few days back in this post). I have often wondered about brain and other tissue such as gut tissue sharing some embryological connections also. Just a thought.

As the authors point out there are some limitations to this study. White males might well make up a large proportion of the autistic population but autism is not exclusive to them. I note that some exclusionary criteria were applied on the experimental group but I can find no mention of any other comorbidities such as epilepsy, other developmental diagnoses or other health complaints as being present or not. Maternal health during pregnancy and little things like any gestational complications/exposures/ factors are also absent from the paper. As has also been described in recent days, accurately gauging IQ in cases of autism might currently be more art than science, which might have quite broad implications for lots of autism research particularly those which use IQ as a factor in cluster analysis.

The final word on loving the face you have goes to Peter Gabriel and the fantastic Sledgehammer video.

* Aldridge K. et al. Facial phenotypes in subgroups of pre-pubertal boys with autism spectrum disorders are correlated with clinical phenotypes. Molecular Autism. October 2011.