Structural variations in corticospinal tract fiber associated with autism spectrum disorder polygenic score in term born infants: cross-sectional study of 221 infants in the developing human connectome project

Increasing lines of evidence suggest abnormalities in motor related brain regions and motor function may be pervasive in autism spectrum disorder (ASD) as early as infancy, but the effect of ASD common variants on these phenotypes remains poorly understood. Recent studies in general paediatric populations found higher ASD polygenic score was associated with lower early motor developmental outcome in infants (Takahashi et al., 2020; Serdarevic et al., 2020), but increased connectivity in motor areas in 3-14-year-old children (Khundrakpam et al., 2020). Concurrently, increased radiate white matter volume within primary motor region was also related to poorer motor function in 8–12-year-old ASD children (Mostofsky et al., 2007). Corticospinal tract (CST), originating chiefly from the primary motor cortex, is the principal motor pathway for voluntary movements; abnormal white matter radial and mean diffusivity of which has been associated with ASD in children and adolescent (Carper et al., 2015). Therefore, the current work aimed to examine the effect of ASD common variants on CST morphology in infants at birth using fixel-based analysis framework. Diffusion imaging and genetic data were analysed for 221 term-born infants (at least 37 weeks of completed gestation) of European ancestry in the developing Human Connectome Project (dHCP) (Edwards et al., 2022). Optimised neonatal multi-shell high angular resolution diffusion imaging acquisition protocol and the following denoising, Gibbs ringing suppression, motion and image distortion correction process were employed according to the dHCP pipeline. Individual fixel (individual fiber population within a voxel) orientation distribution maps generated using MRtrix3 were normalized and registered to a common 40-week anatomical template. Tractography of the CST was performed on the common template using probabilistic algorithm with 10 million seeds. CST fixel-wise measure of fiber density (FD), fiber cross-section (FC) and fiber density cross-section (FDC) were calculated for each individual. Here, FD measures the local differences in intra-axonal volume, FC is the difference in intraxonal volume manifested in the number of voxels the fiber bundle occupies, and FDC is the multiplication of the two values. Polygenic scores (PS), defined as the sum of the number of the risk alleles weighted by the strength of their association with ASD derived from the summary statistics provided by the latest genome-wide association study (Grove et al. 2019), were estimated at 10 different p-value thresholds using PRSice-2 (Choi and O'Reilly 2019). Fixel-wise linear regression was then fitted to estimate the association between each PS and fixel-based fiber measures, adjusting for ages at birth and at scans, first 3 ancestry principal components (PCs), sex and total brain volume (only for FC and FDC). Using spectrum decomposition method to account for correlated PS (Li and Ji, 2005), a fixel was considered significant if the family-wise error pFWE -value was < 0.008. Log(FC) and FDC, but not FD of fibers within the left CST were statistically associated with ASD PS PT=0.01 (pFWE < 0.008), where higher PS was associated with increased intra-axonal volume at macrostructural, but not microstructural level. This is the first study examining CST in relation to ASD PS in a general population at birth. This preliminary result warrants further examination of clinical utility of motor function as early phenotype of ASD.