18524
Distributed Hypoconnectivity As a Neural Endophenotype of Autism
Endophenotypes are quantifiable, heritable and state-independent markers of psychiatric disorders (Gottesman and Gould, 2003). Endophenotypes of autism should reliably quantify genetic effects of the disorder and may play an important future role in prevention, diagnosis and treatment. Identification of endophenotypes in autism has, however, thus far been limited; studies have either reported behavioral endophenotypes, or described changes in task-evoked activations localized to specific brain regions. Such reports of localized changes have been difficult to replicate, in part due to heterogeneous age and gender samples between studies (Müller et al, 2011).
A promising candidate endophenotype of autism is whole-brain hypoconnectivity, which has been reported to varying extents in autism populations, and may underlie the clinical manifestations of the disorder (Di Martino et al, 2014). Problematically, reports of hypoconnectivity have been variable and inconsistent, which may again reflect group heterogeneity, calling for further investigation with well-defined, well-matched groups.
Objectives:
Here we examined whole-brain functional-MRI connectivity from a large sibling-pair neuroimaging dataset during 3 tasks and 1 no-task condition, to investigate whether hypoconnectivity constitutes an endophenotype for autism. We aimed to explore global differences alongside specific networks consistently abnormal in autism, such as the default mode network (DMN).
Methods:
fMRI was performed for 53 subjects with high-functioning autism, 44 unaffected siblings of people with autism ( “siblings”), and 40 controls during rest, an embedded figures and an emotional face processing task, and the ‘Reading the Mind in the Eyes’ task. Following preprocessing, functional connectivity in each of the 4 conditions was established as the correlation between the average time series of each pair of 264 regions of interest (Power et al, 2011). To maximise the homogeneity of participant groups for our analysis, we selected age- and IQ-matched groups of 14 males with ASD (mean age: 15.05), 14 siblings (mean age: 15.03) and 14 controls (mean age: 15.1) using an unbiased algorithm (van Casteren et al, 2007). We computed the relative DMN connectivity as the strength of connections within vs. outside the DMN, and additionally used the Network-Based Statistic (Zalesky et al, 2010) to look for specific network differences.
Results:
Group differences in the average strength of correlations between regions appeared in all conditions, though these did not reach significance in the ‘Eyes’ task. During both other cognitive task-conditions and during resting state, a significant endophenotype effect occurred in which siblings exhibited stronger average network connectivity than autistic participants and weaker connectivity than controls. This effect was non-focal in the embedded figures task but left-lateralised in the emotional faces task, involving anterior and posterior cingulate regions. Preliminary analysis identified the same endophenotype in the DMN during resting state.
Conclusions:
We found significant reductions in cortical connectivity during rest and two different cognitive tasks. In each condition, adolescents with autism displayed weaker connectivity than did siblings; these, in turn, exhibited weaker connectivity than controls. These findings corroborate earlier reports of hypoconnectivity in autism in a replicable well-matched sample, finding these to be distributed and non-focal, and further our understanding of the broader autistic phenotype.