Multidisciplinary prediction of seizure relief from cortical resection surgery

Abstract

Preoperative variables from a full range of medical specialties were used to predict degree of seizure relief from cortical resection surgery as treatment for epilepsy in 100 patients. General, seizure history, electroencephalographic (EEG), radiological, surgical, and psychological/neuropsychological data were considered. The patients were divided into one large predictive group (n = 75) and a smaller independent cross-validation sample (n = 25). Eight predictive variables emerged: single EEG focus; anterior-midtemporal lobe discharges; discharges only from the side of surgery; rate of occurrence of discharges in surgical area; Wechsler Adult Intelligence Scale Digit Symbol subtest; Marching Test, preferred hand, time; Minnesota Multiphasic Personality Inventory (MMPI) Hysteria scale score; and MMPI Paranoia scale score. By use of multivariate procedures, increased predictability of surgical outcome was obtained not only with the predictive group but with the independent cross-validation sample. The results demonstrate that predictions of seizure relief from epilepsy surgery can be made with 80% accuracy using multiple, rather than single, predictors.