Disrupted Network Differentiation in Autism: An fMRI Study of Intrinsic Brain Activity in Children

Background: Functional imaging studies of autism spectrum disorders (ASD) have focused on regional abnormalities, identifying hypo-activations in task-related areas and greater activation in task-irrelevant regions. Recently, authors have drawn attention to ASD-related disruptions in functional connectivity (FC) in networks involved in social and executive processes. Building upon the existing literature, we propose the hypothesis that the previously observed findings reflect disruptions in the differentiation of functional networks during development. Resting-state fMRI (R-fMRI) approaches are useful for testing this hypothesis. They provide measures of intrinsic brain activity simultaneously in multiple networks with moderate to high test-retest reliability. R-fMRI has been fruitful in characterizing development and aging of brain functional networks and has been used to detect abnormalities in adults with autism. Here, we present the first examination of R-fMRI in children with ASD.

Objectives: To systematically examine the pattern of FC in multiple nodes of two classically examined networks: the default mode network (DMN) and the task positive network (TPN) in children with ASD compared to typically developing children (TDC).

Methods: Twenty children with DSM-IV ASD (10.4 y ± 1.7; 3 girls) and twenty age-matched TDC (4 girls) completed a 6.5 min R-fMRI scan (field strength= 3 T; TR = 2000 ms; 3x3x3mm voxels). Signed consent and assent were provided by all children and parents, respectively. Preprocessing included slice-time correction, motion correction, bandpass temporal filtering, spatial filtering, and spatial normalization. Spherical regions of interest (seeds) were selected for FC analyses at: medial prefrontal cortex (MPFC), retrosplenial complex, bilateral lateral parietal cortex (LatPar), parahippocampal complex and hippocampal formation, within the DMN; and bilateral intraparietal sulcus (IPS), ventral IPS (vIPS), inferior precentral sulcus, and middle temporal area within the TPN. Multiple regression (using FSL FEAT) FC analyses were carried out including the timeseries of each seed and nine nuisance covariates as predictors (i.e., movement, white matter, global signal, and CSF). Group analyses using random effect models implemented in FLAME were carried out. Gaussian random field theory was used to correct for multiple comparisons at the cluster level (min Z > 2.3; cluster significance: p ≤ 0.05, corrected).

Results: Preliminary analyses show that children with ASD lack functional segregation between nodes of the TPN and DMN, compared to TDC. In the DMN, a pattern of ‘promiscuous’ FC was observed, with areas that are typically negatively correlated or unrelated with a seed of interest. Specifically, in ASD, the MPFC showed significantly increased FC with dorsal paracingulate areas, bilaterally. Similarly, the right LatPar seed was significantly correlated with the right lateral temporal occipital junction. These areas are typically negatively connected with the seeds of interest. In children with ASD, within the task-positive network, the right vIPS and the left IPS showed reduced negative FC with posterior cingulate and precuneus. 

Conclusions:  Our data support the hypothesis that functional abnormalities in ASD reflect abnormal network differentiation as indexed by disrupted patterns of segregation and integration. If confirmed, these findings may also account for the variegated imaging literature which contains both hypo- and hyperactivations in ASD.