Abstracts

Rationale: Hemispherectomy in children with intractable seizures and unilateral hemispheric abnormalities results in seizure control and functional improvement. The preservation of various cognitive and motor functions following hemispherectomy has been attributed to plasticity of the remaining hemisphere. We investigated white matter changes in the intact hemisphere using diffusion tensor imaging (DTI) combined with an objective voxel based analysis (Tract-Based Spatial Statistics (TBSS), Smith et al., Neuroimage. 2006 Jul 15;31(4):1487-505) by comparing pre- and postsurgical DTI scans. Methods: We studied 10 children with intractable seizures who underwent cerebral hemispherectomy (five left-sided) (age: 4.2±3.2 years) and pre- and postsurgical DTI scans. All 10 children had normal pre-surgical MRI and FDG PET findings in the contralateral hemispheres and became seizure free after surgery. TBSS analysis involved nonlinear registration of individual pre and postsurgical fractional anisotropy (FA) images of the normal hemisphere, followed by skeletonization and voxel-wise paired statistical comparison of the skeletonized pre- and postsurgical FA images. Regions with significant differences were identified using significance threshold: p < 0.05, cluster threshold: 3, and with age at the time of scan as a covariate. The absolute FA values and the differences in FA values between pre and postsurgical FA images of these identified regions were further correlated with the duration between surgery and the second scan with age as a covariate. Results: Our analysis revealed regions with decreased FA after surgery in the contralateral genu (GC) and splenium (SC) of the corpus callosum (CC) and corona radiata (CR) (Figure 1). Whereas FA values from the GC and SC of the postsurgical scans showed a significant negative partial correlation (r = -0.667; p = 0.049 and r = -0.897; p = 0.001, respectively), the difference in FA values between pre- and postsurgical scans for the CR region showed a significant positive partial correlation with duration after surgery (r = 0.778, p= 0.014), indicating that while FA of the CC progressively decreased after surgery, FA in CR showed an initial decrease followed by an increase (Figure 2). Conclusions: The postsurgical decline of FA values in CC suggests degenerative changes, probably associated with Wallerian degeneration. A specific segment of corona radiata, comprising various association and projection pathways (superior longitudinal fasciculus, corticospinal tract, superior thalamic radiation), probably undergoes transient structural changes immediately after the hemispherectomy followed by compensatory changes over and above the normal age-related changes. Our data suggest dynamic postsurgical white matter changes possibly associated with brain plasticity.