24981
Altered Early Development of Resting-State Network Properties in Infants at High Risk for Autism Spectrum Disorder.

Saturday, May 13, 2017: 2:10 PM
Yerba Buena 9 (Marriott Marquis Hotel)
A. Nair1, T. Tsang2, J. Liu2, L. P. Jackson3, C. Ponting4, H. Bowman5, S. S. Jeste6, S. Y. Bookheimer2 and M. Dapretto2, (1)University of California Los Angeles, Los Angeles, CA, (2)University of California, Los Angeles, Los Angeles, CA, (3)Semel Institute, UCLA, Los Angeles, CA, (4)Clinical Psychology, UCLA, Los Angeles, CA, (5)NPI Psychiatry, UCLA, Los Angeles, CA, (6)UCLA, Los Angeles, CA
Background:  Graph theory is a quantitative technique to measure complex dynamics and small-world topology (i.e., segregation of nearby-networks and integration of more distant networks) between brain networks (Bullmore and Sporns, 2009). Prior studies with healthy neonates have applied graph theoretical analysis to characterize early development of functional networks (Gao et al., 2011; Fransson et al., 2011, De Asis-Cruz, 2015). These studies have shown that functional networks in neonates exhibit similar small-world characteristics as in adults, albeit with denser connections in sensorimotor hubs in infants and in association cortices in adults. However, little is known about early development of small-world topology in infants at elevated risk for autism spectrum disorder (ASD).

Objectives: Given prior studies implicating disrupted cortical connectivity in ASD etiology (Rudie et al., 2013, Kana et al., 2014), it is crucial to understand the evolution of small-world topology of functional networks in infants at high risk for ASD.

Methods: Resting-state fcMRI (rs-fcMRI) data were acquired during natural sleep for 8 minutes on a 3T Siemens scanner for 24 infant siblings (6 weeks post-birth) of children with ASD (i.e., high-risk group; HR) and 28 infants at low risk (LR) for ASD (i.e., no family history of ASD). Data were preprocessed using AFNI and FSL, including motion correction, spatial smoothing, isolation of low frequency fluctuations (.008<f<.08), and normalization to the UNC Infant 0-1-2 neonate atlas (Shi, 2011). Time-series were extracted for all voxels within 45 bilateral cortical and subcortical regions of interest (ROIs) using the same atlas. Correlations were computed between all ROI pairs resulting in a 45x45 functional connectivity matrix. Graph theory metrics were used to examine the density of connectivity between ROI pairs (i.e., the percentage of possible voxel in each ROI pair that are correlated above a specified threshold, r>.25, p<.001). Two sample t-tests were performed on z-transformed density values to identify group differences in connection density for each ROI pair.

Results: Consistent with prior studies, results indicated greater overall connection density for rs-fcMRI networks in sensorimotor and subcortical regions in both groups. Between-group comparisons revealed increased connection density for more ROIs pairs in the HR group as compared to the LR group. Specifically, regions involved in visual and sensorimotor processing (i.e., cuneus, middle occipital gyrus, paracentral lobule) demonstrated greater connection density in the HR group compared to the LR group. In contrast, the LR group showed greater connection density in temporal (especially fusiform gyrus) and prefrontal ROIs compared to the HR group.

Conclusions: Our findings suggest that functional network properties may be disrupted as early as the first weeks of life in infants at high risk of developing ASD. Further longitudinal assessment is required to determine if these early differences in connection density could serve as a biomarker in predicting which infants within the HR group will later meet criteria for diagnosis of ASD.