Predicting pragmatic language abilities from brain structural MRI in preschool children with ASD by NBS-Predict

Area of Science:

• Neuroscience
• Developmental Psychology
• Medical Imaging

Background:

• Pragmatic language skills are essential for social communication and are often impaired in children with autism spectrum disorder (ASD).
• The relationship between the brain's structural connectome (SC) and pragmatic language abilities in children with ASD is under-researched.
• Understanding these neural underpinnings is crucial for developing targeted interventions for ASD-related communication difficulties.

Purpose of the Study:

• To investigate how the structural connectome (SC) predicts pragmatic language abilities in preschool children with ASD.
• To identify specific white matter structural network (WMSN) subnetworks associated with pragmatic language deficits in ASD.
• To explore the neurobiological basis of pragmatic language challenges in young children with ASD using advanced neuroimaging analysis.

Main Methods:

• Diffusion tensor imaging (DTI) and deterministic tractography were used to construct whole-brain white matter structural networks (WMSNs) in 92 children with ASD and 52 typically developing (TD) preschoolers.
• Network-based statistic (NBS)-Predict, a machine learning (ML) integrated approach, was employed to identify dysconnected subnetworks and predict pragmatic language abilities.
• Classification accuracy and correlation coefficients were used to assess the predictive power of SC features for pragmatic language scores.

Main Results:

• NBS-Predict identified a subnetwork with 42 reduced connections across 37 brain regions (p=0.01) in the ASD group, achieving 79.4% classification accuracy.
• Dysconnected regions were primarily located in frontotemporal and subcortical areas, with the right superior medial frontal gyrus (SFGmed.R) showing the most extensive disconnection.
• A significant association was found between pragmatic language abilities and white matter connections linking the SFGmed.R with the bilateral anterior cingulate gyrus (ACG) (correlation coefficient=0.220).

Conclusions:

• Abnormal white matter subnetworks, concentrated in frontotemporal and subcortical regions, are characteristic of ASD in preschool children.
• Specific white matter pathway abnormalities between the SFGmed.R and ACG are associated with pragmatic language deficits in children with ASD.
• These findings highlight potential neurobiological targets for understanding and addressing pragmatic communication challenges in young children with ASD.