Early Disruption of Neural Circuits for Social Communication in Tuberous Sclerosis Complex (TSC)

Background: Tuberous Sclerosis Complex (TSC) is a genetic disorder with high prevalence of autism spectrum disorder (ASD). Prior work has identified MRI features associated with ASD in both adult and pediatric TSC patients. Multiple abnormalities exist throughout the brain and alter over time, with varying degrees of severity. Comorbidities including seizures and intellectual disability are also common in TSC and add to the severity of imaging features. There is an urgent unmet need for imaging biomarkers that can be used in the first year of life to identify and direct early intervention towards infants most likely to develop ASD.

Objectives: We sought to determine if early disruption of the microstructural integrity of the neural circuitry of brain regions for social communication is associated with later diagnosis with ASD.

Methods: This research was conducted under a prospective, ongoing, multisite TSC Autism Center for Excellence Network (TACERN) study. We analysed 125 MRI scans of 50 infants between birth and three years of age, with recruitment planned for the first year of life, and annual MRI thereafter until reaching age three. 3T MRI of the brain was acquired using conventional structural MRI and diffusion MRI with multiple non-zero b-values and b=0 images with reversed phase-encoding directions for distortion compensation. Images were aligned to 1 mm³T1-weighted MPRAGE, followed by segmentation into 114 cortical and subcortical regions so that whole brain single-peak diffusion tensor imaging (DTI) and multi-peak diffusion compartment imaging (DCI) tractography could be performed. Neuronal circuitry measured via DTI and DCI fingerprint connectivity of the left and right fusiform gyrus to the remaining 113 compartments was determined and associated with AOSI measures of social communication assessed at 12 months.

Results: The DCI connectivity fingerprint had a direct, in-sample prediction of AOSI with a high correlation. Our quantitative analyses revealed four key findings. First, the fusiform fingerprint-estimated AOSI was significantly correlated with the actual 12-month AOSI at all time points (12mo: p<1e-06 / 24mo: p<1e-06 / 36mo: p<1e-06). The correlation was weak at 12mo (r2=0.49) but strong at 24mo (r2=0.68) and 36mo (r2=0.75). Second, the correlation monotonically increased with age (r2: 0.49 →0.68 →0.75) while the error consistently decreased with age (RMSE: 6.43 →4.71 →3.79). Third, the area under the receiver operating characteristic curve (AUC) when classifying AOSI≥10 and AOSI<10 was high and consistently increased with age (AUC: 87.8% →90.6% →94.1%).

Conclusions: The DCI fingerprint of FUSG characterized the neural circuit underlying the structure of abnormal connectivity in infants at risk of ASD. The improved specificity and sensitivity of the fixel-based DCI analysis, over DTI, was substantial. This is due to the benefit of separately characterizing fascicles and free water. In contrast, the inability of DTI to adequately represent diffusivity properties of individual fascicles lead to a poorer association. Our work demonstrates that changes in the structure of the brain in children at risk of ASD occur early in life and strongly associate with AOSI measures at 12 months.