Brain Magnetic Resonance Findings in 117 Children with Autism Spectrum Disorder under 5 Years Old

We examined the potential benefits of neuroimaging measurements across the first 5 years of life in detecting early comorbid or etiological signs of autism spectrum disorder (ASD). In particular, we analyzed the prevalence of neuroradiologic findings in routine magnetic resonance imaging (MRI) scans of a group of 117 ASD children younger than 5 years old. These data were compared to those reported in typically developing (TD) children. MRI findings in children with ASD were analyzed in relation to their cognitive level, severity of autistic symptoms, and the presence of electroencephalogram (EEG) abnormalities. The MRI was rated abnormal in 55% of children with ASD with a significant prevalence in the high-functioning subgroup compared to TD children. We report significant incidental findings of mega cisterna magna, ventricular anomalies and abnormal white matter signal intensity in ASD without significant associations between these MRI findings and EEG features. Based on these results we discuss the role that brain MRI may play in the diagnostic procedure of ASD.


Supplementary Methods
To measure the developmental age, 14 children were administered the Griffiths Mental Development Scales [1] while the Psychoeducational Profile-revised [2] was used for 15 children. The Stanford-Binet Scale [3] was used for one patient. Nonverbal IQ was assessed using the Leiter International Performance Scale-Revised [4] in 38 children and the Wechsler Scales for 4 children: Preschool and Primary Scale of Intelligence, third edition [5] for one patient and Wechsler Intelligence Scales for Children [6] for 3 patients. Forty-five children were not able to complete any of the aforementioned tests due to lack of cooperation or anxiety. In those cases, clinical judgment of level of functioning was made by the child neuropsychiatrists.
Because of a lack of compliance of some of our young patients, neurocognitive testing was done on average 16 months after the MRI assessment (between 31 months before and 106 months after the MRI examination). The reported data on cognitive level (Table 1) were therefore obtained using different tests which allowed us to assess the presence either of a developmental delay (DD) in children up to 4.11 years old or of an intellectual disability (ID) in children of 5.0 years old or older.
The cut-off for DD/ID was set at a non-verbal IQ/GD score =/< 70 (i.e., 2 SD or more below population mean) which constitutes the most commonly used cut-off in both clinical practice and research (see for example [7][8][9][10][11]). For the sake of simplicity, we decided to label all children with ASD presenting with a non-verbal IQ/GD </= 70 as ASD-DD.
A proportion of 16.24% (19/117) of the children were younger than 31 months at the time of scanning (a final diagnosis of ASD was confirmed when they reached at least 31 months), 8 of them were assigned to a HF category, while the remaining 11 children have been considered as ASD-DD. The remaining 83.76% (98/117) were above 31 months at the time of the MRI assessment (65 ASD-DD and 33 ASD-HF).
One might question whether the temporal misalignment between neurocognitive and MRI assessments could have affected the attribution of the children to one or another ASD category (namely, ASD-HF or ASD-DD). While for the children that underwent neurocognitive testing prior to MRI, it was not always possible to access successive neurocognitive scores, if present, in order to confirm our ASD-HF/DD classification, strong evidence has shown that childhood IQ appears to remain stable over time [12]. In all cases, for which a follow up of Developmental level/IQ was available, Child Neuropsychiatrists found a stability over time.