Abstracts

RATIONALE: Failure of temporal resection to control seizures in patients with mesial temporal sclerosis (MTS) presumably occurs if additional epileptogenic brain is left behind. To better explain and predict surgical failure, the clinical characteristics of MTS patients who failed surgery were examined retrospectively. METHODS: All patients with pathologically verified hippocampal sclerosis from 1992 to 1999 were identified. Those with a preoperative evaluation at the Regional Epilepsy Center, and postoperative followup of at least 1 year, were selected for analysis. Patients were classed as seizure free except for possible auras (SF, N=88), improved (IM, more than 75% seizure reduction, N=38), or not seizure free (NSF, N=12), based on their last year of followup. RESULTS: Right sided resection was the single factor correlating most significantly with higher risk of failure (SF 51%R, 75%L, IM 35%R, 21%L, NSF 14%R, 4%L). A abnormal baseline neurological examination also predicted failure (SF 26.1%; IM 29.5%; NSF 58.3%). There was a trend for positive neurological history of significant head trauma, CNS infection or neoplasm, vascular malformation or possible neurological birth injury to correlate with failure (SF 31.8%, IM 39.5%; NSF 50%). Factors which did not correlate significantly with surgical failure included age at onset or operation, gender, handedness, family history, history of febrile seizures or status epilepticus, psychiatric history, preoperative WAIS Full Scale IQ, and extent of hippocampal resection. In the diagnostic workup, there were trends for bilateral interictal or ictal scalp EEG epileptiform activity (SF 37.5%, IM 50.0%, NSF 50.0%) and need for invasive monitoring (SF 21.6%, IM 63.1%, NSF 25%) to correlate with failure. CONCLUSIONS: Factors associated with more widespread cerebral pathology or epileptogenesis can be expected to predict risk for surgical failure in MTS patients. However, the cause of higher failure rates with right temporal resections is not clear, but could be due to referral or selection bias. Outcome data from centers with different referral patterns are needed to clarify this issue.