Exploring structural connectomes in children with unilateral cerebral palsy using graph theory

Abstract

AbstractBackground and objectivesBrain damage during early development impacts the brain structural network and its coinciding functions. Here, we explored structural brain connectomes in children with unilateral cerebral palsy and its relation to sensory-motor function using a novel semi-automated graph theory analysis, investigating both hemispheres.MethodsIn 46 children with spastic unilateral cerebral palsy (mean age 10y7m±2y9m, 27 boys, 25 right-sided hemiplegia; Manual Ability Classification System I=15, II=16, III=15) we assessed upper limb somatosensory (two-point discrimination and stereognosis) and motor function (grip force, Assisting Hand Assessment and Jebsen-Taylor Hand Function Test). We collected multi-shell diffusion-weighted, T1-weighted and T2-FLAIR MRI and performed transcranial magnetic stimulation to identify the corticospinal tract (CST) wiring pattern. Structural connectomes were constructed using Desikan-Killiany parcellations with Virtual Brain Grafting and FreeSurfer, and Multi-Shell Multi-Tissue Constrained Spherical Deconvolution diffusion modelling with anatomically constrained tractography. Graph metrics (characteristic path length, global/local efficiency and clustering coefficient) were calculated for the whole brain, the ipsilesional/contralesional hemisphere, and the full/ipsilesional/contralesional sensory-motor network, and were compared between lesion types (periventricular white matter (PWM)=28, cortical and deep grey matter (CDGM)=18) and CST-wiring patterns (ipsilateral=14, bilateral=14, contralateral=12, unknown=6) using ANCOVA with age as covariate. We used elastic-net regularized regression to investigate how graph metrics, lesion volume, lesion type, CST-wiring pattern and age predicted sensory-motor function.ResultsIn both the whole brain and subnetworks, we observed a hyperconnectivity pattern in children with CDGM-lesions compared to PWM-lesions, with higher clustering coefficient (p=[<0.001-0.047], ), characteristic path length (p=0.003, ) and local efficiency (p=[0.001-0.02], ), as well as a faster decrease in global efficiency with age (p=[0.01-0.04], ). No differences were found between CST-wiring groups. In general, good predictions of sensory-motor function were obtained with elastic-net regression (R2=0.40-0.87). For motor function, the CST-wiring pattern was identified as the strongest predictor. For somatosensory function all independent variables contributed equally to the model.DiscussionWith this study, we demonstrated the feasibility and potential of structural connectomes using graph theory analysis in understanding disease severity and brain development in children with unilateral cerebral palsy. Hence, this exploratory study could support and direct future research.