Atypical Functional Connectivity in Children with Autism Spectrum Disorder Using Magnetoencephalography (MEG)

Background: Previous functional imaging studies of resting-state brain activity in autism suggest altered connectivity in the “default mode” network as well as regional oscillatory power differences. To date, there are no published reports of resting state whole brain non-phase lagged alpha-band coherence using magenetoencephalography (MEG) to assess children with autism. MEG is a functional imaging technology that quantifies the magnetic activity generated by electrical currents in the brain. We can estimate the degree to which the neurons in specific brain areas (voxels) are firing synchronously with other voxels in the brain – with millisecond temporal resolution and centimeter spatial accuracy. In our previous work, we report atypical somatosensory cortex activation that is directly correlated with observed tactile sensory behavior. Thus we were motivated to investigate both the differences in whole brain coherence between groups as well as the specific coherence related to tactile processing ability.

Objectives: The aim of this study is to determine if boys with autism show altered resting state connectivity relative to age, gender, and performance IQ matched healthy controls (HC) using MEG functional coherence.  We also sought to explore whether behavioral measures of language and social skills and direct sensory assessment of tactile processing is related to the measured neural cortical connectivity.

Methods: Resting activity (eyes closed) was recorded for the autism cohort who met ADI-R and ADOS ASD criteria (n=20, mean age=9.25y) and the HC group (n=25, mean age=9.64y) using 275-channel whole-head MEG (CTF Inc.). Oscillations in the alpha-band (8-12Hz) across a 60 second window were isolated from 4 minutes of continuous recording.  Neural sources were estimated using an adaptive spatial filtering technique and functional connectivity was computed using global imaginary coherence.  Connectivity volumes were compared between the ASD and HC groups.  Voxelwise correlations between global connectivity and autism-related language and social deficits (SCQ) as well as direct measures of somatosensory processing (graphesthesia) were calculated.

Results: The ASD cohort was found to show significantly higher connectivity in three regions: the dorsal medial superior frontal gyrus (D-mSFG), ventral medial superior frontal gyrus (V-mSFG), and precuneus (FDR corrected <5%). Behavioral-Coherence correlations revealed that increased language impairment is directly correlated with higher connectivity of the implicated D-mSFG region for the ASD but not the HC cohort (ASD: r=0.634, p<0.002; HC: r=-.138, p=0.51). No correlations were found with the social construct of the SCQ.  We found additional correlations between measures of tactile processing and right precentral gyrus connectivity (Brodmann area 4) for both groups.  This suggests that the degree of coherence in this region may predict ability beyond the confines of diagnostic labels even though the ASD group performs significantly worse on graphesthesia accuracy (ASD mean=18.16, HC mean=21.61, p=0.02).

Conclusions: School-aged boys with autism show increased resting coherence in medial anterior and posterior regions within the default mode network suggesting an over engagement in regions serving internal preoccupation/reflection and a dysfunction of connections between critical brain regions. Furthermore, we observe a range of connectivity in the primary somatosensory region that may be used to predict tactile processing ability.