Abstracts

Rationale: Stroke is the cause of epilepsy in as much as 45% of individuals above age 60. Although some risk factors for the occurrence of post stroke epilepsy (PSE) have been identified, the mechanisms of developing epilepsy are not known. Here we explore the frequency of PSE in a large imaging stroke research database and assess characteristics of lesions in PSE, compared to the remaining patients with chronic strokes, without epilepsy. Methods: We identified 450 patients with information on the occurrence of seizures from an existing research database (the Predicting Language Outcome and Recovery after Stroke, PLORAS database) that holds high-resolution MRI scans from stroke survivors. MRI data acquisition and analysis: T1-weighted anatomical whole-brain scans were acquired with a 3 T MRI scanner (Sonata, Siemens Medical Systems, Erlangen, Germany) at the UCL Wellcome Trust Centre for Neuroimaging (London, UK) at a 1mm 3 voxel size. Lesions were automatically identified for each patient and thresholded into a binary lesion image. Lesion overlap maps were created to indicate the number of patients who share lesions at a given voxel. Results: 369 patients suffered left hemispheric strokes (LHS), 42 of these (11.4%) had PSE, the remaining 327 did not have PSE. The incidence of PSE was the same in 81 patients who suffered right hemisphere strokes (RHS): 9 (11.1%) had PSE and 72 did not. Those with and without PSE were comparable for gender, handedness and stroke etiology but those with PSE were significantly younger than those without PSE (44+/-13 versus 56+/-13 years, P < 0.0001). Lesion analysis reveals that PSE patients had significantly larger lesions than non-PSE patients (148+/- 101 cm3 vs. 73+/-87, P < 0.0001). Large lesions are more likely to damage deep white matter. Fig 1A illustrates the lesion overlap from all 51 patients with PSE. Figure 1B highlights a contiguous region extracted from this map with damage in 27/42 (64%) of the LHS patients with PSE; 55/327 (17%) LHS patients without PSE also had damage over the same region. Therefore damage to this region was associated with PSE in 27/82 (33%) of patients. In patients with large LHS and PSE, commonly affected areas include the basal ganglia (globus pallidus and caudate nucleus), and most nuclei of the thalamus (anterior and ventral nuclei; posterior regions, including pulvinar). Figure Legend: Figure 1A Lesion overlap map of all patients with PSE. The color scale indicates the number of patients with damage at a particular voxel, superimposed on a template image. L = left hemisphere. Figure 1B: The brain region that was completely damaged in 27/42 patients with LHS and PSE. This was the most consistent lesion site from all those with PSE. Conclusions: We have found a high percentage of patients with PSE (11%) in this cohort with chronic stroke, in line with recent reports. Younger patients appeared at greater risk to develop PSE. The average stroke volume in patients with PSE was roughly double that of non-PSE patients suggesting that the frequency of epilepsy increases with the size of the lesion. Lesion analysis in our cohort identified a region that was damaged in 64% of patients with LHS and PSE. However, the incidence of PSE after damage to this region was only 33%. Funding: This work was supported by Epilepsy Research UK and by the National Institute for Health Research University College London Hospitals Biomedical Research Centre.