Atypical Connectivity of Default Mode and Salience Networks and Links with ASD Symptomatology

Background: Recent findings from functional connectivity MRI studies suggest that typical brain functioning can be characterized by differentiation and integration of brain networks.  However, in autism spectrum disorders (ASD), network dynamics appear impaired.  Specifically, connectivity within functional networks appears reduced while connectivity between regions belonging to separate networks appears increased.  Such inefficient network organization may particularly affect connectivity for primary nodes or hubs of major functional networks, such as the default mode network (DMN) and salience network (SN).  Both networks have shown abnormalities in ASD and may be associated with characteristics of the disorder. 

Objectives: To examine the functional connectivity of DMN and SN in relation to symptom severity in children with ASD.

Methods: Resting state fMRI data and behavioral measures (Social Responsiveness Scale [SRS], Sensory Profile [SP], and Repetitive Behavior Scale-Revised; RBS-R) were used to assess functioning in 34 ASD group and 31 matched TD participants (ages 7-17 years).  FMRI data preprocessing was performed using AFNI and FSL. Data were slice-time, motion, and field map corrected. Functional data were co-registered onto the anatomical image and standardized into MNI space.  Spatial smoothing at 6mm FWHM and bandpass filtering (.008 < f < .08 Hz) were applied. Time points with >1mm motion were censored and nuisance regressors removed, included 6 motion, physiological (heart rate and respiration), white matter, and ventricular time series and their derivatives. Regions of interest included two nodes in the SN (anterior insula, dorsal anterior cingulate cortex) and two nodes in the DMN (posterior cingulate, medial prefrontal cortex).  Time series for each nodes as well as for combined network nodes were correlated with time series from every other voxel in the brain.  Additionally, time series for each network were correlated with one another to assess between network connectivity.  Connectivity values (z’) were correlated with behavioral measures within the ASD group.

Results: For the DMN, decreased connectivity in the ASD group was found with other regions implicated in DMN functioning, while increased connectivity was found with non-DMN regions.  The SN showed reduced connectivity in the ASD group with the anterior prefrontal cortex bilaterally, which has been implicated in higher-order salience processing. Within the ASD group, connectivity between DMN and SN correlated negatively with behavioral scores from the SRS-Cognition subscale (r(33)= -.567, p<.001), i.e., symptomatic behaviors decreased with increasing connectivity.  

Conclusions: Consistent with previous findings, we found reduced network integration (reduced within-network connectivity) in ASD for both DMN and SN, as well as reduced network differentiation (atypically increased out-of-network connectivity) for the DMN. However, the correlation between reduced symptom severity and increased cross-network (DMN-SN) connectivity suggests that between network connectivity may also be functional beneficial (or compensatory) for some networks in ASD.