Increased Connectivity in Children with Autism Spectrum Disorders: Evidence Consistent with Poor Network Segregation

Background: The default mode network (DMN) has generated interest in the study of brain function in individuals with autism spectrum disorders (ASD) because of its purported role in self-reflection, an impaired ability in ASD.  Adult and mostly adolescent samples with ASD reportedly have lower activity correlations (i.e., functional connectivity) between the DMN posterior hub, Posterior Cingulate Cortex (PCC), and other regions in the DMN (e.g., ventral medial prefrontal cortex (vMPFC) and angular gyrus).  These studies limited their analyses to DMN regions only; this approach limits the ability to observe cross-network connectivity. These studies have also not examined connectivity in early school-age children. Given that brain development in ASD is characterized by an atypical developmental trajectory, it is crucial to examine functional networks during childhood.

Objectives: To examine whole brain functional connectivity in data from two key nodes in the DMN (PCC) in children with ASD.

Methods: Forty children participated in this study.  The ASD group (n=18) and the typically developing control (TDC) group (n=22) were matched on age (all p-values>0.33).  Children completed a 5 minute resting state sequence on a 3T Siemens Trio with TIM, with the following parameters: TR=2000; TE=31ms;  90° flip angle; FOV=256x256mm; 3mm in-plane resolution with a 2.7mm slice thickness (0.3 mm gap); 43 slices with interleaved acquisition.  The 150 volumes were slicetime corrected, realigned, normalized, and smoothed in SPM8.  The 150 volumes were then bandpass filtered in FSL from 0.1>x>.01, and volumes with excessive motion were eliminated (up to 30) based on recent evidence that excessive head motion creates systematic correlations.  All children had <30 volumes of excessive motion so we removed the initial volumes until all children were left with 120 volumes total.  Finally, the regressor of interest (PCC) and nuisance regressors (White Matter signal, Cerebrospinal Fluid, and all 6 motion parameters) were extracted and included in a fixed effects model for individual participants with the whole brain, except the cerebellum due to incomplete coverage.  Individual t-test maps generated with PCC seed were then submitted into a two-group t-test. 

Results: There were no regions which showed increased connectivity in TDCs, but we found increased connectivity between the ASD group’s PCC seed and the anterior portion of the left inferior frontal gyrus and left insula/superior temporal sulcus at p<0.005, k=137 (alphasim cluster correction for multiple comparisons at p<0.05).  Follow-up analyses using a secondary DMN seed (ventral medial prefrontal cortex) converged with findings from the PCC. 

Conclusions: The present findings suggest that children with ASD may be characterized by patterns of increased connectivity between some networks.  Our whole brain approach afforded the opportunity to observe correlations with non-DMN regions, and this increased connectivity with non-DMN regions suggests poor network segregation in childhood ASD.  This poor segregation at rest may contribute to difficulties with proper coordination of a network during task-based activities, and thus drive the observed lower correlations (i.e., underconnectivity) often reported during task-based functional connectivity.  Future research is needed to determine whether functional connectivity differences at rest are predictive of functional connectivity differences during task.