Temporal Lobe over-Connectivity Differentiates Autism Spectrum Disorder from Other Neurodevelopmental Disorders in Toddlers: A Hardi Diffusion MRI Study

Background:  One of the greatest priorities for clinicians dealing with autism spectrum disorder (ASD) is the need for early predictive biomarkers, allowing for a prompt specific intervention apt to improve long-term outcome. Advanced connectivity studies are emerging in the literature reporting a disrupted connectivity in toddlers with ASD, detectable even before the full phenotypic expression of the disease. Most of these studies however, compare ASD and typically developing subjects, thus providing little information on the specificity of the abnormalities detected as to other neurodevelopmental disorders (NDD). To our knowledge, only one recent study explored brain structural differences between toddlers presenting with ASD and NDD, through a voxel-based approach. It mainly showed increased volumes of grey and white matter within the temporal lobes and higher fractional anisotropy in the corpus callosum, posterior cingulate cortex, and limbic lobes. 

Objectives:  We performed a whole-brain tractography study in toddlers from two clinical groups, ASD and other NDDs, matched for age and gender, to identify the networks showing significantly different connectivity.

Methods: As part of an ongoing prospective project, we enrolled all children referred to our tertiary care centre for neurodevelopmental disorder from January 2012. Subjects were included if they i) received a clinical diagnosis of Neurodevelopmental Disorder, ii) were aged under 36 months, iii) had no neurometabolic or genetic syndromes and iv) performed a brain MRI. A total of 59 subjects with NDD fulfilled our inclusion criteria. The cohort was further divided into two groups, based on clinical assessment: ASD (n=39; mean age=27 months) and NO-ASD (n=20; mean age=30 months).

Brain MRI was performed on a 1.5T GE scanner. High angular resolution diffusion imaging (HARDI) scans were acquired using 31 diffusion weighted images (b value = 1000 s x mm^-2). Image post-processing consisted of cortical parcellation of the T1 images into 90 regions excluding the cerebellum. Whole brain probabilistic tractography was done using the Anatomically Constrained Tractography (ACT) framework of MRtrix. Further, SIFT was used to reduce bias over longer streamlines. Network connectivity matrices of size 90x90 were built that encoded the number of streamlines connecting each pair of cortical regions. Network Based Statistics (NBS) was finally applied on the connectivity matrices to evaluate the network differences between the ASD and DD groups. 

Results:   The network differences revealed over-connectivity (higher number of streamlines) in the ASD group with a significance of p<0.05. Two main networks were identified, both centered on the superior temporal gyrus, one in the left hemisphere (5 nodes and 4 edges) and the other in the right hemisphere (3 nodes and 3 edges). No networks showed significant over-connectivity in the DD group. 

Conclusions: This is the first tractography study comparing toddlers with ASD and other NDDs. We report a trend in over-connectivity in toddlers presenting with ASD, particularly in networks involving the temporal lobes, known to be crucial for social-skills development. This extends previous findings from voxel-based analysis showing increased volumes in comparable regions. Further studies are needed to confirm our findings and to clarify their translational significance.