Abstract | Summary | Original Article

Outcome Classification of Preschool Children with Autism Spectrum Disorders Using MRI Brain Measures

Akshoomoff N, Lord C, Lincoln AJ, Courchesne RY, Carper RA, Towsend J, Courchesne E

Journal of the American Academy of Child and Adolescent Psychiatry 2004; 45(3): 349-357

Question: Can MRI brain measures predict if young children suspected of having autism will later be diagnosed with an autism spectrum disorder (ASD)?

Background: There is no specific biological diagnostic test that can identify autism. The behavioural tests currently used must be done by psychologists or psychiatrists with specific training in diagnosis, and in some areas there are shortages of these specialists. In addition, the test results are not always valid until after the age of three, thereby delaying diagnosis and crucial early intervention. It would, therefore, be very useful if a diagnostic test based on a biological marker was available.

For some time it has been known that there are brain differences in children with autism compared to typically developing children. Young infants with autism, on average, have bigger brains than typically developing children. This is a study of the use of an MRI test that measured brain matter in children who presented with symptoms of autism and in children who were typically developing. The children were followed up later when they were old enough to have received a reliable diagnosis ASD based on proper assessment. The goal of the study was to see if the findings on the MRI scans could predict the diagnosis and the severity of the condition.

Design: Prospective study.

Method: Children with a provisional diagnosis of ASD and typically developing children were given an MRI scan that looked at both white matter and grey matter in the brain. The radiologists reading the scans did not know which children were which.

Participants: The participants were 52 boys, aged 1.9 to 4.8 years, who had symptoms of autism or PDD-NOS identified by an experienced clinician. The controls were 15 typically developing and otherwise healthy boys who were aged 1.7 to 5.2 years at the beginning of the study. Children who had known medical or genetic conditions in addition to their symptoms were excluded from the study.

Description of test and diagnostic standard: Depending upon their age and their ability to speak, the provisional ASD group was given one of the standard psychological and language tests such as the ADI-R, the CARS, or the ADOS. Each of the ASD children was then given a comprehensive physical and neurological examination. The MRI scans were done when the children with a provisional diagnosis of autism were 1.9 to 5.2 years of age. The typically developing children were evaluated using standard tests of intelligence. They were given scans when they were 1.7 to 5.2 years of age.

Main outcome measures: A final diagnosis of ASD, the severity of symptoms of ASD, and general intelligence were correlated to the white and grey brain matter volumes in the frontal area (the cerebrum, the area in which mental processing takes place) and in the back portion of the brain (the cerebellum, the area that coordinates motor control, but also influences activity within the brain). These are the two main areas of the brain in which excesses of brain matter had been previously found.

Main results: Of the original 52 children with symptoms of autism, 81% (42) received a final diagnosis of autism and 19% (10) received a diagnosis of PDD-NOS.

• The low functioning autism (LFA) group had a significantly larger whole brain size than the control group. The high functioning autism (HFA) group did not have bigger brains.
• The greatest difference between the HFA group and the controls was in the overall volume of the cerebellum.
• The volume of the front of the brain (the cerebrum) was significantly larger in the LFA group compared to the control group.

The MRI scans were able to differentiate the groups by final diagnosis.

• All of the LFA children were classified as having autism.
• The scans were highly accurate in identifying the HFA children as having an ASD but less so in specifying if they had either HFA or PDD-NOS.
• The MRI scans were much less accurate in identifying children with a diagnosis of PDD-NOS.
• All but one of the control children was classified as being a control.

Conclusion: In this study, the use of MRI scan to measure brain volumes successfully discriminated children with autism from controls and was able to successfully categorize children with HFA and LFA. It was less successful for children with PDD-NOS. Larger numbers of children should be studied with this technique to see if the findings hold up, and further attempts must be made to standardize the measures so that the MRI scanning use is consistent from centre to centre.