This ESI / New Investigator clinician-scientist brings an expert team to test the altered brain functional connectivity hypothesis of autism during infancy. We will conduct graph-theory-based network analyses on functional connectivity magnetic resonance imaging (fcMRI) data acquired in high-risk infants who are currently being studied in the multi-site, NIH-funded Autism Center of Excellence (ACE) study R01 HD055741: A Longitudinal MRI Study of Infants at Risk for Autism - Joseph Piven: PI. We will prospectively study diverging developmental trajectories for functional networks of brain regions in infants who do and do not develop an Autistic Spectrum Disorder (ASD), during a suspected period of altered brain growth in autism, and prior to symptom expression. All four ACE data collection sites and their existing Data Coordination Center will participate. These efforts will provide longitudinal fcMRI data (scans at 6, 12, and 24 months old) in up to 664 high-risk and control infants (~15% of the 544 high-risk infants will develop an ASD). All four ACE data collection sites are already (as of May, 2010) - in advance of any dedicated funding, to ensure enough fcMRI data - acquiring fcMRI data on their subjects. Interfacing with the high-risk infant ACE structural imaging study and its supplements will allow us to build on existing infrastructure, save effort and cost, standardize collections, and merge databases. It also provides the potential for future genetic-functional imaging associations. Our fcMRI approach considers the problem of autism from the perspective of alterations in the behavior of brain-wide networks and changes in strengths of functional connectivity between and within rigorously defined sub-networks of brain regions. We believe this approach is more fruitful than those which consider smaller numbers of brain regions. We will additionally use a novel cortical functional areal parcellation routine (that Co-I Steve Petersen et al. have developed) to enhance the sensitivity of the infant fcMRI data analyses. Unique insights may come from comparing our network-based fcMRI analyses to the structural MRI, diffusion tensor imaging, and extensive phenotypic data generated by the existing ACE study (fcMRI acquired in the same infants). Once diagnoses are assigned in the ACE study, we will use a machine learning, multivariate pattern classification approach (Support Vector Machine: SVM) to explore the possibility of developing a diagnostic classifier for ASD. The eventual goal would be to train an SVM to predict which infants will and will not develop ASD, based on network properties of their fcMRI data, prior to the expression of symptoms. If successful, our proposed program of research could inform the future development of pre- symptomatic diagnostic tests for ASD, increase our knowledge about the neurobiology of autism, and provide methods for studying functional brain changes in response to interventions. These approaches could be adapted for the study of other neurodevelopmental and early-onset psychiatric disorders.
Interagency Autism Coordinating Committee (IACC)
Autism Research Database (AFD)
