Abstracts

Rationale: We used Tract Based Spatial Statistics (TBSS; Smith SM et al., Neuroimage, 2006;31:1487), to objectively investigate the fractional anisotropy (FA) and mean diffusivity (ADC) of cerebral white matter (WM) using MRI with diffusion tensor imaging in children with Sturge-Weber syndrome (SWS) with epilepsy and unilateral hemispheric involvement. This automated and unbiased method allows objective analysis of water diffusion changes in cerebral WM tracts. Methods: Twelve children with SWS (age: 3-11 years) were studied prospectively and compared to 11 healthy controls (age: 6-13 years). Nine of the SWS subjects showed pial angioma on the left side (L-SWS) while three had the angioma on the right. In the group comparisons, the images from these three subjects were right-left flipped. TBSS analysis involved nonlinear registration of individual FA maps to a template image, skeletonization of the FA map, followed by voxel-wise statistical comparison of the skeletonized FA and ADC maps of the two groups (controls and SWS) with age as a covariate. Regions with significant differences were isolated and mean FA values of the skeletons were correlated with clinical and neurocognitive data such as epilepsy duration, age at seizure onset, seizure frequency, IQ scores (IQ), and manual dexterity scores (MDS). These correlation analyses included only the L-SWS subjects with age as a covariate. Results: TBSS showed regions with significantly decreased FA in the left (ipsilateral) frontal lobe WM (L-FWM), left superior longitudinal fasciculus (L-SLF), left corticospinal tract (L-CST) and right (contralateral) frontal lobe WM (R-FWM) in the SWS group compared to controls (Figure 1). The FA of L-SLF showed significant negative partial correlation (r=-0.9 p=0.002) with epilepsy duration (Figure 2). In addition, IQ showed positive partial correlation with the FA of L-FWM (r=0.92 p=0.001), L-SLF (r=0.78 p=0.02), and R-FWM (r=0.74 p=0.03). Furthermore, MDS for left and right hands showed positive partial correlation with the FA of L-CST (r=0.82 p=0.01, r=0.74 p=0.03, respectively), L-FWM (r=0.92 p=0.001, r=0.78 p=0.02, respectively) and L-SLF (r=0.85 p=0.008 right hand only). Seizure frequency and age at seizure onset did not show any correlation with the FA values. TBSS analysis of ADC maps showed no regional differences between controls and SWS patients. Conclusions: Our objective analysis revealed abnormalities of WM regions ipsilateral and, to a lesser degree, contralateral to the vascular brain lesion in children with unilateral SWS. Lower FA values in patients with longer duration of epilepsy suggest progressive loss of WM integrity during the course of the disease. These changes involve specific WM tracts, some of which are located in the frontal lobe, remote from the posterior angioma. The significant relationship between WM diffusion abnormalities and neurocognitive scores provides further evidence for the importance of WM integrity for cognitive development in children with SWS and epilepsy.