In 1996, the American Epilepsy Society, in collaboration with the Quality Standards Subcommittee of the AAN, appointed a panel to develop practice parameters for surgical treatment of epilepsy. The American Association of Neurological Surgeons subsequently agreed to join this project. The core working group consisted of four neurologists who were directors of epilepsy centers that offered surgical treatment (J.E., R.G., M.S., P.W.), a neurosurgeon who was director of an epilepsy surgery program (D.S.), and a neurologist with particular expertise in outcomes research (S.W.). Additional panelists included two members of the Quality Standards Subcommittee, one of whom was an epileptologist (J.F.) and another who was not (C.Z.); a neurologist who was director of an epilepsy program for a health maintenance organization (B.E.); and a general neurologist (E.W.).
Literature review process
The initial comprehensive literature search was performed by the University of Minnesota, using Medline and Current Contents to identify all relevant papers published between January 1, 1990 and June 1999. Two lists of search terms were used, and at least one term from each list needed to be present for a paper to be identified. The first list included the following terms: seizures, epilepsy, Lennox-Gastaut, West syndrome, infantile spasms, Landau-Kleffner, hypothalamic hamartoma, cortical dysplasia, hemimegencephaly, tuberous sclerosis, Sturge-Weber, Rasmussen's encephalitis, mesial temporal sclerosis, hippocampal sclerosis, and drop attacks. The second list included the following terms: surgery, amygdalohippocampectomy, multiple subpial transection, lobectomy, corticectomy, corpus callosotomy, corpus callosum transection, amygdalotomy, hemispherectomy, and resection. This search yielded 1282 citations. After reviewing the abstracts, 415 were considered to contain potentially usable information for this study and were reproduced in full.
Three panel meetings were held. At the first meeting, in December 1997, it became apparent that only papers that reported series with pure anteromesial temporal lobe resections, pure localized neocortical resections, or both, were sufficiently numerous to analyze. For the purposes of this review, no distinction was made among the various types of anteromesial temporal resections performed, which ranged from selective amygdalohippocampectomy to large tailored anterior temporal lobe excisions. All patients included in this initial review underwent surgery for what was considered to be medically refractory complex partial seizures with or without secondarily generalized seizures.
At the second meeting in December 1998, inclusion criteria for published surgical series were further refined to make sure results would be as generalizable as possible to all temporal lobe and neocortical resections. For anteromesial temporal lobe resections, papers were excluded if the study population was limited to only a subset of a larger population of patients who ordinarily would be considered for this surgical procedure. For instance, papers were excluded when they included only patients with tumors on MRI, only patients with bilateral independent EEG spikes, only patients who had invasive recording, or only children or the elderly. Because exclusively pediatric studies were not included and data were not analyzed by age, this review has limited applicability to children. For localized neocortical resections, exclusion criteria were similar, except that series devoted entirely to frontal lobe or occipital lobe resections were permitted.
Two specific postoperative outcome measures were chosen: frequency of epileptic seizures other than auras (12) (simple partial seizures without motor features), and quantitatively measured health-related quality of life (QOL) (13). The majority of papers used a standardized seizure outcome classification system with minor variations (12), which identified patients who were free of disabling seizures (and therefore permitted persistent auras), improved, and not improved (accepting whatever standard the investigators used to differentiate these latter two groups). The evaluation periods varied and it was not possible to segregate outcome results according to all the various periods of follow-up reported. Data were also used from the few papers that did not use this standardized scale when it was at least possible to segregate patients who were free of disabling seizures from those who were not. Papers that included other outcome measures regarding psychiatric status, work, school, neurocognitive function, driver's licensing, and mortality were also reviewed. Finally, all papers included in the study were evaluated for data that would reveal the incidence and nature of surgical complications.
Papers were ranked according to class of evidence (Table 1). In the initial review there were no Class I reports. One would have met criteria for Class II (14) and the remainder for Class III, except that none had a masked outcome assessment; therefore, all were Class IV. Papers were further evaluated according to a rating scale (see the Appendix on the Neurology Web site; go to www.neurology.org) designed to eliminate papers with less reliable data, and to permit stratification of the remainder at a later date, if desired, according to criteria that might influence the results of the evaluation. The content, validity, and relevance of the rating scale were addressed by scoring a large number of articles and by panel discussion. Based on this rating scale, series were excluded if they contained fewer than 20 patients, if the outcome assessment was unclear, if the surgical intervention was not adequately described, or if any patients in the series underwent surgery before 1974, when modern neuroimaging was not generally available. For all outcome assessments, series were excluded if follow-up for any patients was less than 1 year. Of particular interest for later stratification were series in which all patients underwent surgery after 1985, when MRI was widely available, and series in which all patients had at least 2 years of follow-up.
Table 1. AAN evidence-classification scheme for a therapeutic article
|Class I: Prospective, randomized, controlled clinical trial with masked outcome assessment, in a representative population.|
|The following are required:|
| a) Primary outcome(s) is/are clearly defined.|
| b) Exclusion/inclusion criteria are clearly defined.|
| c) Adequate accounting for dropouts and crossovers with numbers sufficiently low to have minimal potential for bias.|
| d) Relevant baseline characteristics are presented and substantially equivalent among treatment groups or there is appropriate statistical adjustment for differences.|
|Class II: Prospective matched group cohort study in a representative population with masked outcome assessment that meets a–d above, OR a randomized, controlled trial in a representative population that lacks one criterion in a–d.|
|Class III: All other controlled trials (including well-defined natural history controls or patients serving as own controls) in in a representative population, in which outcome assessment is independent of patient treatment.|
|Class IV: Evidence from uncontrolled studies, case series, case reports, or expert opinion.|
At the third meeting in August 1999, the rating scale was used to select those papers that would make up the data set from 171 papers that remained from the University of Minnesota search, and 2 others added as a result of independent searches carried out by the panelists. In order to avoid including overlapping data reported more than once from the same center, when two or more papers from the same center met the inclusion criteria, only the largest or most recent study was used for each specific review objective. If several papers from the same center recorded results of patient populations that were overlapping, one paper might be used for one review objective, while a different paper might be chosen for another review objective.
A final literature search, through September 2001, for new studies meeting Class I criteria yielded one randomized, controlled trial of surgery for temporal lobe epilepsy with a masked outcome assessment, published in August 2001 (15). Results of this study were essentially identical to those obtained from our earlier literature review.