White Matter Development in Infancy Predicts Repetitive Behavior and Sensory Features

Background: Restricted and repetitive behaviors are early emerging (Wolff et al., 2014) and possibly prognostic features (Lord et al., 2006) of autism spectrum disorder (ASD). Recent changes to diagnostic criteria include for the first time unusual responses to sensory stimuli as part of the restricted and repetitive behavior domain despite a lean body of literature supporting this conceptual grouping. To date, little is known about the neurobiology underlying the development of these features of ASD early in life, with the vast majority of published work focused on older children, adults, or adult non-human animal models (see Langen et al, 2011, for a review).

Objectives: To investigate the development of specific white matter pathways in relation to emerging repetitive behaviors and sensory features in infants who developed ASD. 

Methods: Diffusion tensor imaging data were collected longitudinally from high-risk infant siblings at ages 6, 12, and 24 months during natural sleep as part of an ongoing study. The present sample includes 44 infant siblings meeting clinical best-estimate criteria for ASD at age 2 years. Cognitive and behavioral data, including the Repetitive Behavior Scales, Revised (Bodfish et al., 2001) and the Sensory Experiences Questionnaire, v2.1 (Baranek et al., 2009) were collected at age 2. Five white matter fiber pathways of interest (see Figure 1) were selected on the basis of previous empirical or conceptual work (e.g. Kern 2002; Langen et al 2014; Pierce et al. 2001; Wolff et al. 2015). Pathways were deterministically segmented in common atlas space (see Verde et al. 2014) and microstructure characterized by fractional anisotropy (FA), a measure reflecting magnitude of diffusion based on tensor shape. The relationships between FA development and repetitive behavior/sensory features were examined using generalized estimating equations controlling for sex and age.

Results: Repetitive behaviors and sensory features were strongly correlated (r = .74, p < .001) and this relationship held when controlling for social symptom severity and IQ. Development of the genu from 6 to 24 months of age significantly predicted repetitive behaviors and sensory features. The mid- and superior- cerebellar peduncles also significantly predicted both repetitive behavior and sensory features. Development of striatal pathways (ATR, CST) was not significantly associated with these behaviors (see full results in Table 1). To determine the specificity of our findings, we next examined whether development of targeted pathways predicted social symptom severity as indexed by the ADOS. This analysis yielded no significant results (p ≥ 0.20).

Conclusions: Our findings suggest that 1) restricted and repetitive behaviors and sensory features co-occur in toddlers with ASD; and 2) share a common relationship with underlying neural circuitry. We specifically identified that these core features of autism are predicted by the structural development of callosal and cerebellar white matter pathways over the first two years of life, suggesting a possible shared mechanism unique to infancy. These brain-behavior relationships were strikingly specific given the absence of any association between targeted pathways and social symptom severity.