Autism spectrum disorder (ASD) is a developmental disorder characterized by impairment of social interaction and communication, as well as repetitive behaviors, with severity ranging from mild to signi?cantly disabling. The prevalence in the United States is rising (currently about 1 in 68 children) and the associated costs are great. In our most recent previous efforts on this project, we have initiated development of a graph-based, Bayesian neuroimage analysis framework and have used it to characterize brain pathology in ASD and identify abnormal functional subnetworks from groupwise data. Key results demonstrated clear differences in task- based functional brain networks between ASD and typically developing control (TDC) groups associated with the perception of biological motion. While our efforts (and those of others) are important for characterizing ASD, advances in the characterization of response to therapy with imaging are crucial for improved understanding, and ultimately personalization, of these therapies. Thus, we propose to put forth a bold new direction: to further develop our analysis methodology and study these task-based subnetworks, now with the goal of characterizing individuals in terms of their predicted response to treatment. We will focus on Pivotal Response Treatment (PRT), an intensive behavioral therapy for children with ASD that improves social communication skills. We ?rst propose to fully develop our uni?ed Bayesian framework to detect both hyper- and hypo-synchronous functional subnetworks within whole-brain, groupwise, task-based fMRI data on a large training dataset of ASD and TDC subjects. We will identify dense subgraphs (communities) that exhibit group differences in functional synchrony between ASD and TDC groups. The groupwise subnetworks will then be mapped to single subject, task-based fMRI data acquired from a cohort of ASD subjects treated with PRT. For each subject, imaging biomarkers based on activation signal strength and functional connectivity will be derived for regions within each hyper- and hypo- synchronous subnetwork at both baseline and after 16 weeks of therapy. Using a random forest regression strategy, we will use a combination of biomarkers from the baseline data to predict response to PRT (using change in Social Responsiveness Scale, 2nd Edition as the primary clinical outcome measure). In addition, we will use a combination of biomarkers from baseline and 16 weeks to predict treatment persistence at 32 weeks. We will compare the prediction capability of our new approach using task-based fMRI to a set of biomarkers with regions identi?ed from groupwise analysis of resting state fMRI (rsfMRI) networks found from the same training subjects noted above using an alternative state-of-the-art method. We will also develop methods to examine potential metabolic alterations in networks using magnetic resonance spectroscopy (MRS) of GABA and glutamate (the major excitatory and inhibitory neurotransmitters) to explore possible biochemical differences associated with ASD, changes in response to PRT and the use of this information as additional imaging biomarkers.
Autism spectrum disorder (ASD) is a dif?cult and growing problem. The goal of this project is to develop sensi- tive and quantitative imaging and image analysis methodology for the study of ASD aimed at measuring brain pathology and using it to characterize and predict response to therapy. This work will advance our understanding of this disorder and its treatment and ultimately inform and guide personalized therapies.
Showing the most recent 10 out of 36 publications
