Between Group Differences in Correlation of FA with Performance IQ and Age in Autism Spectrum Disorder. Results from EU-AIMS Diffusion Imaging.

Background:

European Autism Interventions is a multicentre initiative aiming to stratify biomarkers for Autism Spectrum Disorder (ASD) through acquisition and integration of large datasets from different imaging modalities and behavioural measures. We are investigating Diffusion MRI data acquired as part of EU-AIMS. Previous studies have implicated Fractional Anisotropy (FA) differences in the corpus callosum¹, the uncinate^1,2and reduced volume in the arcuate³with ASD.

Objectives:

We aimed to investigate voxel-wise differences between a large cohort (N=187) of ASD and Typically Developing (TD) subjects imaged at King’s College London. We also investigated the relationship between FA and Performance IQ (PIQ) behavioural scores and age between our cohorts.

Methods:

Diffusion Imaging data is taken from the King’s College London cohort of EU-AIMS. Following exclusion for anatomical/genetic abnormality, high subject motion⁸and intellectual disability we obtain a cohort of 144 subjects (61 TD 83 ASD; ages 6-30,mean=18.2 STD=5.8). Data acquisition parameters were: voxel size 2x2x2 mm, b-value = 1500 s/mm², 60 diffusion weighted directions and 6 b0s.

Data was de-noised⁴, corrected for Gibb’s ringing⁵, for motion and eddy-current distortions using exploreDTI⁶and outlier replacement using FSL eddy⁷. PIQ for each subject was evaluated using the Wechsler Abbreviated Scale of Intelligence (WASI), including matrix reasoning and block design. Voxel-wise statistical analysis of Fractional Anisotropywas carried out using TBSS (Tract-Based Spatial Statistics⁹) part of FSL¹⁰. In this study we present results for group differences in Fractional Anisotropy (FA) while correcting for sex and age, and correlation of FA with PIQ and Age.

Results:

Group comparison (TD vs ASD): We find collections of significant (p<0.05)voxels across the Corpus Callosum body, Superior Cerebellar peduncles and Uncinate fasciculus (fig .1). These results replicate previously published results from UK-AIMS²(males, age 18-40).

Significant correlations were found with performance IQ in ASD but not in the whole cohort (p<.05) in regions consistent with U-shaped fibres connecting medially to Caudal Inferior Parietal Lobule (fig .2, .3) in left hemisphere.

Finally, correlation of FA with age produces large areas of highly significant (p<.01) voxels, as expected from our large cohort age ranges (6-30) (fig .4). By creating masks of significant voxels for each cohort and subtracting TD from ASD we find areas in the genu and splenium of the Corpus Callosum which correlate strongly with age in ASD only (fig .3).

Conclusions:

Group differences observed in this preliminary analysis of the EU-AIMS cohort and their replication with previous diffusion results from the UK-AIMS cohort identify the Corpus Callosum, Uncinate Fasciculus and the Cerebellar Peduncles as anatomical targets for further analysis. FA in the Genu and Splenium also emerge as showing more correlation with age in ASD than TD, suggesting that brain development in autism may follow a delayed developmental trajectory¹¹within these areas. PIQ measures used feature matrix reasoning and block design tests. Block design tests have been previously implicated in autism in the central coherence account¹². Correlation of performance on these tasks with a parieto-temporal region only in ASD may be of interest to this view of ASD.