Factors Associated With Pre- and Postoperative Seizures in 1033 Patients Undergoing Supratentorial Meningioma Resection

Risk factors for pre- and postoperative seizures in supratentorial meningiomas are understudied compared to other brain tumors.To report seizure frequency and identify factors associated with pre- and postoperative seizures in a large single-center population study of patients undergoing resection of supratentorial meningioma.Retrospective chart review of 1033 subjects undergoing resection of supratentorial meningioma at the author's institution (1991-2014). Multivariate regression was used to identify variables significantly associated with pre- and postoperative seizures.Preoperative seizures occurred in 234 (22.7%) subjects. At 5 years postoperative, probability of seizure freedom was 89.9% among subjects without preoperative seizures and 62.2% with preoperative seizures. Multivariate analysis identified the following predictors of preoperative seizures: presence of  ≥1 cm peritumoral edema (odds ratio [OR]: 4.45, 2.55-8.50), nonskull base tumor location (OR: 2.13, 1.26-3.67), greater age (OR per unit increase: 1.03, 1.01-1.05), while presenting symptom of headache (OR: 0.50, 0.29-0.84) or cranial nerve deficit (OR: 0.36, 0.17-0.71) decreased odds of preoperative seizures. Postoperative seizures after discharge were associated with preoperative seizures (OR: 5.70, 2.57-13.13), in-hospital seizure (OR: 4.31, 1.28-13.67), and among patients without preoperative seizure, occurrence of medical or surgical complications (OR 3.39, 1.09-9.48). Perioperative anti-epileptic drug use was not associated with decreased incidence of postoperative seizures.Nonskull base supratentorial meningiomas with surrounding edema have the highest risk for preoperative seizure. Long-term follow-up showing persistent seizures in meningioma patients with preoperative seizures raises the possibility that these patients may benefit from electrocorticographic mapping of adjacent cortex and resection of noneloquent, epileptically active cortex.