Abstracts

RATIONALE: Patients with medically refractory temporal lobe epilepsy can be successfully treated with surgery. However upto 30% of even the most suitable candidates, with concordant electroclinical data and apparently isolated hippocampal sclerosis, are not rendered seizure free following surgery despite complete resection of the visualised lesion. Preoperative identification of these patients would avoid exposing them to the risks of surgery and allow for more rational use of resources. qMRI can reveal subtle structural abnormalities not seen on visual inspection of optimised MRI, which may reflect underlying epileptogenic tissue, and be useful as a prognostic indicator in these patients.
Objective-To understand the role of MRI post processing techniques in the presurgical evaluation of patients with hippocampal sclerosis
METHODS: We analysed T1 volumetric images of 74 consecutive patients from our epilepsy surgery program and 100 age/sex matched controls. All patients had isolated unilateral hippocampal volume loss on preoperative MRI and pathologically proven hippocampal sclerosis. Fully automated post processing of MR data were performed to obtain measures of regional grey/white matter distribution The data was segmented using SPM99, a probabilistic anatomical atlas registered to each subject data set and the volumes extracted. Regional grey and white matter volume was corrected for brain size and the normal range defined from the control data. Processing time for each patient was less than 1hour.
Surgical outcome data was collected blind to MRI and pathology data. In 38 subjects quantitative pathological measures of cortical and white matter gliosis were obtained blind to MRI and clinical data. The presence of extralesional qMRI abnormalities was correlated with surgical outcome.
RESULTS: The mean post surgical follow up period was 36 months (SD [plusminus]11 months). 44 patients had quantitative abnormalities of grey and white matter distribution, in 23 patients there were additional extralesional abnormalities. Only 8 of these 23 patients were seizure free (Engel class Ia) at follow up. 28 seizure free patients had no extralesional abnormalities. The sensitivity of qMRI was 39% and specificity 78% (positive predictive value 0.65). Extending the seizure free group to include Engel class I and II, those with worthwhile improvement, the sensitivity increased to 62%, with a specificity of 79%.
There was no significant correlation of abnormal qMRI measures or post operative outcome with quantitative pathological measures.
CONCLUSIONS: Abnormal qMRI can guide post surgical prognosis. Only 8/23 with abnormal qMRI were seizure free post operatively, whereas 40/51 patients with a normal qMRI achieved a Engel class I outcome.
We did not find a correlation between qMRI and quantitative pathological measures. This suggests that the volume differences detected are not merely reflective of a gliotic process secondary to seizures.
Our automated method provides a fast, reliable, objective method that could be a valuable prognostic tool in the presurgical evaluation of patients with hippocampal sclerosis. New methods have been developed to detect occult epileptogenic foci, but none has been proven to provide prognostic information.
[Supported by: MRC (UK)]