Widespread Associations between Behavioral Metrics and Brain Microstructure in ASD

Abstract

AbstractAutism spectrum disorder (ASD) is a neurodevelopmental disorder characterized by deficits in social communication and repetitive behaviors. A diagnosis of ASD is provided by a clinician following cognitive and behavioral evaluations, but there is currently no biomarker associating these metrics with neurological changes. Our lab has previously found that g-ratio, the proportion of axon width to myelin diameter, and axonal conduction velocity, which is associated with the capacity of an axon to carry information, are both decreased in ASD individuals. By associating these differences with performance on cognitive and behavioral tests, we can evaluate which tests most reveal changes in the brain. Analyzing 273 participants (148 with ASD) ages 8-to-17 (49% female) through an NIH-sponsored Autism Centers of Excellence (ACE) network (Grant#: MH100028), we observe widespread associations between behavioral and cognitive evaluations of autism and between behavioral and microstructural metrics. Analyzing data from all participants, conduction velocity but not g-ratio was significantly associated with many behavioral metrics. However, this pattern was reversed when looking solely at ASD participants. This reversal may suggest that the mechanism underlying differences between autistic and non-autistic individuals may be distinct from the mechanism underlying ASD behavioral severity. Two additional machine learning cluster analyses applied to neuroimaging data reinforce the association between neuroimaging and behavioral metrics and suggest that age-related maturation of brain metrics may drive changes in ASD behavior. By associating neuroimaging metrics with ASD, it may be possible to measure and identify individuals at high risk of ASD before behavioral tests can detect them.Significance StatementThis study establishes numerous relationships between multiple behavioral, language, and social metrics in ASD. Subsequently, this study is the first to then show associations between diffusion microstructure and subscales of behavioral assessments. Limited associations of these behaviors with conduction velocity may indicate that axonal diameter is a predominating factor in characterizing ASD over other metrics, such as myelination, however within ASD subjects the g-ratio is more closely related to behavioral metrics, suggesting a potential role for myelination in ASD severity. These findings suggest that some subscales and metrics more accurately capture behaviors associated neurologically with ASD than others, including composite scores, demonstrating the potential to identify children at high risk for ASD at an earlier age.