Abstracts

Rationale: Temporal lobe epilepsy (TLE), the most common cause of drug-resistant epilepsy in adults, is associated with marked cell loss and gliosis in the hippocampal formation and adjacent regions. Histopathological studies have revealed that these anomalies extend beyond mesial temporal structures into neocortical regions. Voxel-based morphometry and surface-based cortical thickness analyses have documented neocortical atrophy in vivo. However, the topography of neocortical gliosis has not been systematically evaluated. Here, we mapped whole-brain FLAIR signal intensity as an in vivo marker of gliosis and examined its relationship to hippocampal connectivity and clinical factors. Methods: T1-weighted and FLAIR MRI were obtained in 61 patients with TLE (31/30 LTLE/RTLE) and 38 age- and sex-matched controls. FLAIR images were co-registered to T1 MRI. We generated cortical surface models, measured cortical thickness and sampled intracortical FLAIR intensity. Surface-based t-tests assessed group differences. In clusters of findings, we computed average effect sizes. To evaluate intensity variations above and beyond effects of atrophy, analysis was repeated after regressing out vertex-wise cortical thickness. A structural covariance approach in which mean FLAIR intensity of the hippocampus was correlated with cortical FLAIR intensity at each vertex assessed hippocampal gliotic network pathology. Linear models assessed relationships between FLAIR intensity, hippocampal connectivity and clinical variables. Analyses were corrected for multiple comparisons using random field theory. Results: Patients showed marked bilateral FLAIR increases relative to controls (p < 0.05) with a predominantly limbic distribution (parahippocampal, insular, cingulate and dorsolateral and orbitofrontal cortices; Fig.1A). In clusters of findings, effect size estimates were high (Cohen's d=0.68-0.89). Increased FLAIR intensity was largely non-overlapping with cortical thinning (Fig.1B). Results remained virtually identical after repeating the analysis controlling for thickness (Fig.1C). We found a strong correlation (r=0.57, p=0.001) between neocortical FLAIR hyperintensity and hippocampal intensity covariance (Fig. 2A), suggesting regions within the hippocampal networks are more likely to undergo gliosis. Age at onset, duration of epilepsy, generalized tonic-clonic seizures and surgical outcome did not relate to FLAIR hyperintensity. However, FLAIR hyperintensity was more marked in patients with febrile convulsions (Fig.2B) in ipsilateral parahippocampal and posterior cingulate gyri. Conclusions: Our results promote cortical FLAIR mapping as an anatomically and clinically meaningful marker of glia-related pathology in TLE. We reveal a highly specific paralimbic distribution of gliotic changes that are independent of cortical atrophy. Our results showing a positive association between FLAIR hyperintensity and febrile convulsions further supports a pathophysiological link between astrogliosis and epileptogenesis. Funding: This work was supported by the Canadian Institutes of Health Research. SA received funding from the Rosetrees Trust and a Bogue Research Fellowship from University College London.