A Graph Theoretic Examination of Social Brain Networks at Rest in ASD

Background: Deficits in social interaction and social communication are core features of autism spectrum disorder (ASD). Dominant theories suggest that these deficits may be due to reduced or atypical social motivation (i.e., the motivation to seek out social interaction) and/or social cognition (i.e., inferring and reasoning about the mental states of others). Further, these social motivation and social cognitive systems may be linked such that reduced social motivation may affect social-cognitive development. While studies have demonstrated atypicalities of both social-cognitive and motivational networks in ASD, these typically are not examined together within the same individual. Further, key nodes within each of these networks such as the amygdala (social motivation) and temporo-parietal junction (TPJ) (social-cognition) may contribute to these atypicalities, as they play an integrative role both within and between functional networks.

Objectives:

We use resting state functional connectivity, which is an ideal method to examine baseline functional network organization, to examine within and between network organization with a focus on key nodes within networks associated with social motivation (amygdala) and social cognition (TPJ) using graph theoretic methods.

Methods:

Using the Autism Brain Imaging Database Exchange, we examined network organization in both ASD (N=79) and TD (N=104) groups. After strict motion control (mean frame displacement < 0.1mm), age, and gender matching, our final sample included N=76 individuals (ASD=30;TD=46; age range: 6.47-19.73, all male). Whole-brain networks were constructed from a freely available functional atlas (Yeo 17-networks), bilateral amygdala, and 12 ventral striatum regions for a total of 128 nodes. In order to examine within and between network organization, we assigned two community affiliations of interest. The social motivation community consisted of the amygdala, ventral striatum, and OFC and, since previous work has implicated the default mode network (DMN) in social-cognitive processing, we assigned the social-cognitive community as all DMN regions. All other regions preserved their community affiliation from the functional atlas. We used graph theory to calculate measures of within- and between-community connectivity for each key node (within-community connections and participation coefficient, respectively). Finally, controlling for head motion, we examined these metrics for group, age, and group*age effects.

Results:

For between-network connectivity, we found significant group (F(1,71)=5.31,p<0.05) and group*age (F(1,71)=10.41,p<0.01) effects in the left TPJ such that this region became less functionally connected with other communities with age in ASD and increased in connectivity in TD. Similar effects were seen in the right TPJ, but only reached trend level significance. No significant between-network effects were seen in the amygdala. For within-network connectivity, no significant effects were seen in either of the key nodes.

Conclusions:

Our results highlight a differential timecourse in the left TPJ’s role in connecting the social-cognitive network with other large-scale networks. While we do not see differences within the amygdala, a key node within the social reward network, group difference may manifest earlier in development and/or be exacerbated in the presence of socially salient stimuli. Our results have implications for understanding functional brain development in ASD at the network level.