Prognostic factors in the surgical treatment of temporal lobe epileptics

Abstract

Factors or combinations of factors that would be significantly different between "surgical successes" and "surgical failures" of unilateral temporal lobectomies carried out for the treatment of temporal lobe seizures were sought and how such factors could be used to arrive at a prognostic evaluation Of a giVen patient before operation. Two diametrically opposite groups of patients were defined and studied; those who either stopped having seizures following temporal lobectomy or had occasional ones during their first postoperative year but none whatsoever thereafter (group 1) and those whose seizure problem was either unchanged or worse following operation (group 2). These criteria formed the fundamental basis for patient selection. Both groups were matched as to sample size (group 1 had 50 and group 2 had 54). Other additional criteria included a minimal interval of 5 years between operation and the beginning of this study. Adequate personal or postal follow-up or both had to be available for this period. Patients in whom the pathological lesion was a neoplasm or a vascular malformation were excluded. A code sheet to analyze 81 variables was used. The data were then transferred to IBM punch cards, and frequency tables for each variable were made with the use of an IBM computer 7044. Chi-square analysis of each variable showed 19 of them to be statistically significant (p less than 0.05). These included the age at operation, presence or absence of seizure-precipitating factors, presence or absence of extratemporal features in the clinical seizure pattern, presence or absence of generalized seizures, clinical localization, presence or absence of radiological abnormalities, localization by psychological test patterns and by eeg, lateralization by psychological studies and by eeg, the postexcision electrocorticogram pattern at operation, the eeg changes during the immediate postoperative period, psychiatric abnormalities, and the histological alterations found in the removed specimen. Results of the analysis are presented, and the significant variables are discussed in greater detail. In particular, an explanation is offered as to why previous reports on this subject have failed to find significance with a number of variables which we have found to be significant in this study. Finally, the limitations of single-variable analysis for prognosis are pointed out as well as possible solutions by using multi-variate analysis.