Longitudinal Trajectories of Large-Scale Brain Network Architecture in Autism

Saturday, May 14, 2016: 11:30 AM-1:30 PM
Hall A (Baltimore Convention Center)
B. A. Zielinski1, M. D. Prigge2, M. White2, D. C. Dean3, B. G. Travers4, A. L. Alexander3, N. Lange5, E. D. Bigler6 and J. E. Lainhart3, (1)Pediatrics and Neurology, University of Utah, Salt Lake City, UT, (2)Pediatrics, University of Utah, Salt Lake City, UT, (3)Waisman Center, University of Wisconsin-Madison, Madison, WI, (4)Occupational Therapy Program in Kinesiology, University of Wisconsin Madison, Madison, WI, (5)McLean Hospital, Cambridge, MA, (6)Psychology/Neuroscience Center, Brigham Young University, Provo, UT

Background: Autism spectrum disorders (ASD) comprise complex neurological conditions characterized by childhood onset of dysfunction in multiple cognitive domains. Accumulating evidence suggests that autism is a network-based disease, and that abnormalities in brain network structure underlie the abnormal brain function at the core of the disorder1. However, developmental timing and topology of brain network development in autism remains unclear. Here we utilize longitudinal structural MRI data to evaluate developmental trajectories of large-scale brain networks in autism.

Objectives: Using an emerging technique known as structural covariance MRI (scMRI)2, in tandem with correlated cortical thickness change, this study sought to characterize autism-related longitudinal growth trajectories of network-level brain architecture.  Specifically, we sought to determine whether specific developmental abnormalities in large-scale brain network organization are associated with autism, and whether network trajectories and correlated inter-regional change are associated with behavioral measures.

Methods: We used scMRI and cortical thickness correlation (MACACC)3 to determine longitudinal trajectories of large-scale brain networks in autism. Using scMRI, we first identified canonical large-scale ‘structural covariance networks’ (SCNs) strongly implicated in autism, namely the socioemotional salience network (SN), the speech network, and the default mode network (DMN). Three hundred and forty-five MRI scans were examined from 97 males with autism (mean age = 16.8 years; range 3–36 years) and 60 age-matched males with typical development (mean age = 18 years; range 4–39 years), with an average of 2.2 scans and interscan interval of 2.6 years. FreeSurfer was used to parcellate the cortex into 34 regions of interest per hemisphere and to calculate mean cortical thickness for each region using methods reported previously4. Partial correlation (controlling for age at date of scan) was used to characterize region-wise relationships in cortical thickness change across time using scMRI-derived network hubs as reference regions. We then determined patterns of regional change that correlated with a broad array of behavioral scores.

Results: Region-based cortical thickness correlation revealed specific perturbations in longitudinal brain network architecture within distinct SCNs, consistent with phenotypic manifestations of autism. Thickness correlation maps in controls were consistent with canonical SCN topologies2. Extent and topology of the salience network (SN), involved in social- emotional regulation of environmental stimuli, demonstrated weaker average correlation in autism, corroborating earlier findings1. The speech network in ASD lacked right temporal regions that were highly correlated with left pars opercularis in the TDC group. The DMN in ASD showed stronger correlations posteriorly and weaker frontal correlations, which were stable across time. Behavioral data also demonstrated distinct differences between ASD and TDC. For example, thickness of the left pars opercularis was negatively correlated with social responsiveness scale (SRS) score in TDC, whereas right pars opercularis was positively correlated with (SRS) in ASD.

Conclusions: Abnormal longitudinal trajectories of large-scale brain network structure characterize autism, and these abnormalities are consistent with phenotype. Moreover, specific patterns of structural coherence in ASD are correlated with behavioral measures. Our findings are consistent with a network vulnerability model of autism, and provide a plausible approach to MRI-based subphenotyping.