Local and long-range functional connectivity abnormalities in ASD: Frequency-specific insights from MEG

Background:   Functional connectivity in the brain is often modulated by cortical oscillations occurring on much faster time scales than can be measured using the BOLD signal. MEG signals are well suited for measuring these fast oscillations, and are also easily corrected for motion artifacts thanks to continuous motion tracking. Thus, MEG offers a complementary approach to fMRI for assessing functional connectivity abnormalities in ASD.

Objectives:   To elucidate the nature of both local and long-range functional connectivity abnormalities in ASD with high spectral specificity, and to evaluate the links between these functional connectivity metrics and behavioral measures.

Methods:   MEG data were recorded during resting state, passive tactile vibrations on fingertips, and face processing paradigms. Group sizes varied from N=15 to N=20 per group per paradigm, with some overlap in cohorts across studies. Long-range interactions were examined between functionally distinct cortical regions. Local interactions were examined within a functionally defined cortical region, i.e. on a spatial scale of the order of about one square centimeter on the inflated cortex.

Results:   We found patterns of both increased and reduced functional connectivity across cortical regions in ASD in different paradigms. During resting state in particular, the properties of the network varied consistently with frequency band. In addition, two of the paradigms, tactile vibrations and face processing, also allowed us to investigate specific aspects of local functional connectivity. In contrast to the long-range functional connectivity results, for both types of stimuli and using different approaches we observed only reduced local functional connectivity in the ASD group. These long-range and local neurophysiological functional connectivity metrics correlated with ASD severity as measured on the ADOS, and in the case of the sensory task, also with behaviorally derived sensory processing scores. Lastly, blind statistical classification using these metrics achieved up to 90% accuracy in identifying ASD participants. 

Conclusions:  Our results support and expand on findings from fcMRI, showing the co-existence of atypically increased and reduced long-range functional connectivity in ASD. Additionally, our studies to date suggest that at least local interactions that are likely mediated via inhibitory connections are uniformly reduced in ASD. The correlations between these spectrally specific functional connectivity metrics and behavioral ASD measures suggest a relevance to the etiology of ASD.