Neuroradiological Incidental MRI Findings in Children with Autism Compared to Healthy Typically Developing Controls

Background: Autism is a complex neurodevelopmental disorder that is likely due to many different etiologies.  Comprehensive medical work-ups reveal pathology in only a small fraction of cases (5-10%).  Brain imaging yield is usually reported to be low.  In fact, the Practice Parameter for autism from the American Academy of Neurology does not recommend routine imaging.

Objectives: As part of an ongoing comprehensive phenomenological investigation designed to determine meaningful subtypes of autism, structural brain MRI is performed on all subjects.  We compared results from the autism (AUT) subjects to a typically developing control group (TYP) to determine if there are more neuroradiological findings in autism.

Methods: Autism diagnosis was confirmed with the ADI-R, ADOS and clinical judgment.  Structural brain MRI was performed using a 1.5T scanner in 69 autistic children (12 females, 57 males; mean age 4.5, sd 1.37) and 53 typically developing children (12 females, 41 males; mean age 4.7, sd 1.37).  There were no significant differences between groups for age and gender.  AUT subjects were scanned while sedated and TYP scans were done during natural sleep or while awake.  All studies were read by NIH staff neuroradiologists for clinical purposes.  Clinical reports were then reviewed and categorized as follows: unequivocally normal; normal but with some variant mentioned in the report; possibly abnormal but of unclear clinical significance; or definite pathology with clinical relevance.   

Results: The rate of definite pathology was not significantly different between groups and was low overall: 1/53 (2%) in TYP children (1 with intraparenchymal cystic lesion) and 2/69 (3%) AUT children (1 with an intraparenchymal cystic lesion and 1 with cortical dysplasia).  Rates of unequivocally normal and normal variants were also not significantly different between groups: 67% and 12% respectively in the AUT group and 77% and 17% in the TYP group.  Normal variants including mild cerebellar ectopia, prominent perivascular spaces, or mega cisterna magna were seen in both groups.  However, the rate of possible abnormalities was significantly higher in the autistic children than the controls (Fisher’s Exact, p=.01) with 13/69 (19%) AUT showing a questionable abnormality (5 with some white matter findings, 3 with evidence of old insults, 3 with developmental changes, 1 with large arachnoid cyst, and 1 with a circulatory abnormality) compared to 2/53 (4%) TYP (1 old insult and 1 developmental change).

Conclusions: Compared with typical controls, autistic children had similar rates of clinically relevant structural abnormalities as well as normal scans and non-clinically relevant normal variants.  However, the rate of possible abnormalities was higher in the group with autism.  The significance of these findings is unknown.  One possible limitation of the data is that the neuroradiologists were not blind to diagnosis.  A second blinded reading is being done to ensure that there were no biases in the interpretation of the AUT group.  However, if these represent true abnormalities in this subset of patients, further investigation into their meaning may provide clues to the clinical presentation and/or etiopathogenesis of autism.