Abstracts

Rationale: Temporal lobe epilepsy is the most common type of medically intractable partial epilepsy. The advent of neuroimaging has diminished the use of intracranial recordings, especially in patients who have epileptogenic lesions on MRI. However, intracranial EEG monitoring (ICM) via hippocampal depth and subdural strip electrodes is commonly employed in a subgroup of patients in which non-invasive testing is considered either inconclusive or discordant. Little attention has been paid to predictors of seizure localization and outcome following ICM. The purpose of this study is to examine which non-invasive presurgical studies are most predictive of seizure localization in refractory temporal lobe epilepsy patients who have undergone ICM. This study may identify subgroups of patients in whom ICM could have been avoided. Methods: All patients (N=54) who underwent ICM for presurgical evaluation of refractory temporal lobe epilepsy at Emory University Hospital between 2000 and 2007 were included in the study. Five patients were excluded from analysis due to lack of seizures occurring during ICM. All patients underwent standard presurgical evaluation consisting of long-term video-EEG monitoring (LTVM), FDG positron emission testing (PET), magnetic resonance imaging (MRI), standardized neuropsychological testing and intracarotid amobarbital procedure (WADA). Primary outcome measure was location of seizure onset on ICM. Results of all studies were classified as either right temporal, left temporal, bitemporal or normal (not temporal). Secondary outcome measure was seizure freedom at 6 and 12 months based on modified Engel’s seizure classification. Concordance between non-invasive studies and seizure localization as seen on ICM was assessed. Results: LTVM correlated with ICM EEG findings in 68% of patients and provided falsely localizing information in only 8%. MRI was falsely negative in 53% while concordant in 43% and falsely localizing in only 4%. There was also a high false negative rate with WADA tests (55%). PET (30%) and neuropsychological testing (44%) both had high rates of false localization. 35/49 (71%) underwent temporal resection following ICM; 61% and 52% were seizure free at 6 and 12 months respectively. 17/24 patients who had normal MRI had unilateral findings on ICM and underwent resection, 47% of whom were seizure free at 12 months. Conclusions: Our results suggest that LTVM is the best predictor of ICM findings although no single study could predict outcome with enough certainty to skip invasive EEG monitoring. Both PET and neuropsychological testing are poor predictors of ICM results. Normal MRI does not predict inconclusive or bilateral findings on ICM that would preclude the possibility of resective surgery.