Autism: Early Bird Diagnosis

25. April 2017

Recognition of autism spectrum disorder currently occurs at the earliest at two years of age. This could soon change due to MRI-based diagnostics – at least for "high-risk children" from families which already have an autistic sibling. But does such an early diagnosis make sense?

In the case of children from families with a previous occurrence of the condition, magnetic resonance imaging (MRI) can already be used during the first year of life to determine whether autism spectrum disorder will develop later. This has recently been reported by researchers in Nature. 1 The big advantage from the perspective of the scientists being: “Early diagnosis of the condition would also allow early preventive measures when the brain is still most malleable”. Early detection also opens up the possibility of early therapy, which is associated with a particularly high long-term success rate.

Opportunity for early therapeutic measures

Nevertheless the causes of autism spectrum disorders are still not treatable. Early therapeutic measures could, however, help people to improve their social interaction and communication skills, thereby keeping concomitant psychological conditions in check, the study authors argue. However, these therapeutic measures would first have to be developed.

Faster increase of the brain’s surface

The study involved 106 children with at least one autistic sibling and 42 children without autistic siblings. The scientists took MRI images of the brain of the sleeping children from both groups, at 6, 12 and 24 months. The brain volume, surface area of the brain and the thickness of certain areas were measured in the cerebral cortex.

The researchers observed that babies which later developed an autism spectrum disorder showed much more rapid enlargement of the surface of the brain at age 6 to 12 months than did babies which by 24 months of age showed no signs of autism. More specifically, the surface of the occiput gyrus, the right cuneus (a portion of the occipital lobe) and the right lingual gyrus area were enlarged.

However, the brain volume growth rate did not differ at that age between the two groups. In the second year of life, the brain volume of the children who were later afflicted with the condition however increased more than in healthy children.

Numerous studies have reported anomalies in brain volume in connection with various mental disorders. A review article 2 from 2006, however, suggests that there could also be scientific bias in the excessively frequently reported phenomenon of increased brain volume in mental disorders in the scientific literature.

Learning algorithm sheds light on the differences

The observations from current research were not apparent to the naked eye, but only came to light through a computer analysis of the application of a deep learning algorithm (a variant of machine learning). This algorithm mainly uses the data of the brain surface of children aged from 6 to 12 months in order to predict whether an autism spectrum disorder gets diagnosed when the child concerned is aged 24 months.


Stephen Dager from the University of Washington is hoping that the chance of recovering is improved for affected children by way of an early diagnosis. Credit: Marie-Anne Domsalla  

Enlarged brain: a risk factor for autism

Previous research had already identified an enlarged brain as a risk factor for autism. The current study makes it possible to detect this early pathological growth process in the child’s brain even before the size of the brain has measurably changed. The excessive growth of the brain also correlated with the severity of social deficits which were diagnosed at age two.

Looking at children with familial tendency who had an autistic sibling, the brain differences at the ages of 6 and 12 months were sufficient to make a predictively correct diagnosis of autism at 24 months for 30 of the 37 children (sensitivity 88 percent, positive predictive value 81 percent). However, this good predictive result only applies for children with a family history. With 4 out of 127 children the diagnosis that was made after 12 months was however not confirmed at 24 months (negative predictive value 97 percent).

Symptoms first recognisable at two years of age

“Typical behavioural problems that may indicate an autism spectrum disorder usually make themselves noticeable from the age of about two”, Annette Estes, head of the Autism Center at the University of Washington says. At this time though, the brain has already changed substantially. “Using our method autism is able to be diagnosed in high-risk children before their first birthday”, Robert Schultz, director of the Center for Autism Research in Children`s Hospital in Philadelphia, USA, says. Among children with an autistic sibling, the risk of also developing an autism spectrum disorder stands at 20 percent.


Annette Estes playing with a child in the Autism Center at the University of Washington. Credit: Kathryn Sauber

In the rest of the population, one in 68 school-aged children has been diagnosed with “autism spectrum disorder”. Around 10 million people worldwide are affected.

Predictive power of the method must be reviewed

However, before the algorithm being tested here is employed clinically, its predictive power must first be verified in further studies. “If we can reproduce our results, we will be able to predict well prior to the occurrence of behavioural problems that familiarly predisposed children will be afflicted later on with an autism spectrum disorder”, says Schultz.


Robert T. Schultz, head of the Autism Center at the Children’s Hospital of Philadelphia, USA © Children’s Hospital of Philadelphia

Information for future treatment strategies

The study could also provide valuable information for future treatment strategies, according to Schultz. “Thanks to the images of the brain, we were able to determine certain regions whose atypical development is apparently involved in the development of autism”, neuropsychologist Schultz explains. If we understand the neural mechanisms behind it, we can deduce potential earlier treatment options – possibly applied even before the symptoms of autism spectrum disorder occur “.

Diagnosis by tablet from three years of age

From three years of age onward, the diagnosis of autism spectrum disorder could be even simpler: only in the autumn of last year, scientists reported about early diagnosis using a tablet.3 Based on the analysis of two age-based game apps, researchers managed a rapid and accurate diagnosis of children aged three to six years. Software developed specifically for the study examined the fine motor skills of children as a background process. They compared how game playing was executed by 37 autistic and 45 healthy children and they were correct in 93 percent of diagnoses. The autistic children were characterised by frequent, heavier movements, they used the tablet with increased pressure and applied more protrusive force and faster movements than healthy children.

Sedated baby, no treatment option

The great advantage of this method: in comparison to MRI scans, the mobile apps are much cheaper and also less invasive for the children, because in order to get meaningful pictures, babies need to be sedated before getting an MRI diagnosis. This measure alone is likely to prevent many parents with high-risk children from seeking an early diagnosis and who – presently at least – still have no therapeutic options.

At three years of age already too late

The scientists involved in the Nature study are indeed aware of these issues, but see their study as a foundation for the development of early therapy for affected children. They fear that a diagnosis at this age might already arrive too late. “At three years of age the affected children are in relation to their social and language skills often already far behind their healthy peers”, says Estes, adding: “if we miss these important development steps, catching up is a perpetual struggle for everyone involved, and almost impossible for many children”.


The MRI scans of the children were only evaluated by means of an algorithm. Credit: University of Washington

Colleague Stephen Dager adds: “We hope that early intervention – before the child’s second birthday – can bring the clinical development of the children back on track again”.


Original publications:

Early brain development in infants at high risk for autism spectrum disorder
H.C. Hazlett et al.; Nature, doi: 10.1038/nature21369; 2017

Excess significance bias in the literature on brain volume abnormalities
J.P. Ioannidis; Arch Gen Psychiatry, doi: 10.1001/archgenpsychiatry.2011.28; 2011

Toward the Autism Motor Signature: Gesture patterns during smart tablet gameplay identify children with autism
A. Anzulewicz et al.; Sci Rep, doi: 10.1038/srep31107; 2016

3 rating(s) (5 ø)

1 comment:

ANP Jo Nell Costello
ANP Jo Nell Costello

My son was diagnosed with Asperger’s Syndrome at age 15 years. I began to take him to doctors from age two months, as knew something was wrong. He was born in 1985, so not in DSM at that time. He was delayed in expressive and receptive language, and socially delayed. However, he was put in the gifted and talented program in elementary school. He was in several advanced placement classes in high school, and placed out of 12 semester hours in college. It took him a long time to graduate from college, but finally got his master’s degree in accounting. He is now 32 years old, working as an accountant part time, and still lives at home. He has managed this with autism and bipolar disorder. He is amazing. I hope that new testing methods such as this, will lead to earluer intervention to improve the lives of those affected by autism.

#1 |

Copyright © 2020 DocCheck Medical Services GmbH
Follow DocCheck: