Abstracts

Rationale: Resective surgery is curative for the great majority of patients with temporal lobe epilepsy and concordant, unilateral abnormalities on MRI, EEG, and PET evaluation. For patients with bilateral or noncondardant preoperative data, prognostication is less clear. This analysis seeks positive and negative predictors of good postoperative outcome in such patients.Methods: A retrospective case series of over 700 temporal lobectomies was culled for cases with bilateral or nonlateralizable surface epileptiform EEG abnormalities (either interictal or ictal patterns). Demographic, etiologic, seizure type, neurophysiologic, MRI, PET, histopathology, and other disease variables were abstracted from available medical records, coded as either present or absent. Cox proportional hazards analysis was performed first of each individual variable, then in combined analyses to examine for interactions between variables. Primary outcome variable for this analysis was defined as seizure freedom at last followup.Results: 105 patients were identified. 31 underwent bilateral invasive electrode implantation to clarify side of predominant involvement. 41 of 105 patients failed (seizures recurred) over a mean followup of over 3 years, a comparable rate versus recent series reporting long-term followup in patients with predominantly unilateral presurgical findings. The only variables independently predictive of seizure recurrence were bilaterally abnormal PET scan (hazard ratio for recurrence, 3.6, 95% C.I. 1.2-10.8) and presence of family history (H.R. 2.4, C.I. 1.2-4.9); predictive of seizure freedom were presence of auras before seizures (H.R. 0.27, C.I. 0.092-0.79) and history of febrile seizures (H.R. 0.21, C.I. 0.045-0.68). The inclusion of MRI variables, seizure-laterality variables, and other potential predictors did not affect the robustness of the bilateral PET predictor effect. Surprisingly, the presence of contralateral MRI abnormalities and even contralaterally recorded EEG seizures did not predict seizure recurrence, nor did inclusion of those variables in models containing the identifiable predictors suggest a hidden effect with interaction. Conclusions: This analysis suggests that – 1. Patients with apparently bilateral surface EEG abnormalities can in fact be chosen on the basis of MRI, PET and, if necessary, invasive EEG evaluation, and patients thus chosen may have a similar proportion of long-term seizure-free outcome to patients with strictly unilateral abnormalities. 2. Bilateral PET abnormalities should disqualify a patient with any combination of favorable EEG or MRI findings from resection surgery if the goal is seizure freedom. The confidence in these findings naturally must be tempered by their retrospective nature, and the comparatively small numbers of patients in some subgroups.