Medically refractory epilepsy (MRE) has been widely recognized as a medical problem that significantly impairs quality of life³ and significantly decreases life expectancy.⁸ The surgical treatment of epilepsy was developed more than 50 years ago² and has expanded greatly since the early 1980s.⁴ Epilepsy surgery generally involves a resective surgery with the goal of rendering a patient free of disabling seizures.⁴ Surgical candidates are generally patients who have undergone an extensive evaluation and are thought to have seizures originating from a single brain region. In the majority of patients, this “seizure focus” is felt to be confined to one of the temporal lobes. Resective surgery is pursued if it is thought that there will either be no noticeable neurologic deficit or that any deficit will be “acceptable” (i.e., some verbal function and/or verbal memory impairment with dominant temporal resections). Many medical centers⁴ and a multicenter National Institutes of Health (NIH)-sponsored study²² have reported a high percentage of patients to be seizure free following resective surgery. It is presumed that this result indicates that all (i.e., in patients ultimately able to discontinue antiepileptic medication) or an adequate amount (i.e., in patients rendered seizure free with well-tolerated antiepileptic medication) of epileptogenic tissue has been resected. Many patients continue to have rare seizures or a much lower frequency or severity of seizures and have a difficult-to-assess degree of improvement or benefit.⁵ A significant percentage of patients continue to have medically refractory seizures that are frequent enough and/or of a severity that prompts consideration for a reoperation with the goal of accomplishing the original surgical goal. This was reported by Penfield and Jasper in the 1950s¹⁴ and subsequently further explored by Rasmussen, operating on patients from Penfield’s series.¹⁵ A number of more recent reports^{1,6,7,10,12,16,17,18,19,20,23} have provided much more knowledge regarding the subject of reoperation for MRE. A significant percentage of patients were reported to be seizure free following reoperation, often indicating that the full extent of the epileptogenic tissue had not been appreciated at the initial surgery or (in a minority of patients) that the “seizure focus” had been misidentified. Patients without improvement after a reoperation have often been felt to harbor multiple or widespread epileptogenic brain areas and it is often concluded that these patients are not able to be rendered seizure free by resective surgery. The analysis of patients who failed to achieve the desired result with a first, or subsequent, resective surgery may be one of the keys to improving resective surgery for MRE, both by improving our ability to select patients who can be rendered seizure free by resective surgery and by improving our understanding of the tissue that must be resected to achieve this goal. For the purposes of this chapter, reoperation will refer to a resective surgery intended to achieve the elimination of seizures in a patient who has previously undergone a similar surgery. It is often helpful to subdivide patients having resective surgery for MRE into three subgroups: Medial temporal lobe resection, resection of a structural/pathologic lesion with associated epileptogenic cortex, and resection of a neocortical nonlesional epileptogenic region. Reoperations can be similarly categorized. In this chapter, the published experience of a number of epilepsy centers with such reoperations will be considered by, wherever possible, subdividing the reoperations into these same three subgroups. This may be the most useful way to gain from the reported experience and to use this knowledge in approaching patients undergoing either an initial or a repeat resective surgery for MRE. Glial neoplasms are being excluded from the structural lesion subgroup, because these lesions often progress over time and should be considered a distinct entity with regard to the management of associated medically refractory seizures. This subject is not addressed in this chapter. Reoperation sometimes is a hemispherectomy. This surgery does not have the same goal as the initial surgery (i.e., focal resection with preservation of hemispheric function) and falls outside of the scope of reoperation defined above. Callosotomy is a distinct type of epilepsy surgery that typically aims to eliminate a specific seizure type and will not be addressed here. Palliative surgery following resective surgery, having the goal of reducing seizure frequency and/or severity (not seizure elimination), such as vagal nerve stimulation, similarly will not be considered reoperation. Also, such surgery as an initial operation will not be considered a “first” operation for MRE.

Many of the reports of reoperation for epilepsy have included patients whose reoperation involved resection of medial temporal structures, often the posterior hippocampus. The series reported by Rasmussen in 1975¹⁵ included 129 nontumor patients who had a reoperation between 1928 and 1971. Of the 115 patients with at least a 2-year follow-up (2 to 39 years), 29 (25%) were seizure free and 31 (27%) were categorized as having significant improvement of seizure frequency. It is probable that the majority (possibly all) of the benefited patients had an extension of their original temporal resection and that this included an extension of the medial temporal resection, although the publication does not provide enough detail to confirm this. By 1984, Spencer et al.²¹ had reported that depth electrode evaluations indicated that about 20% of hippocampal seizures had a posterior onset and they described a surgical technique to accomplish a complete hippocampal resection. The importance of this recommendation for complete hippocampal resection is supported by subsequent experience. A large percentage of patients who have been reported to be seizure free after reoperation have had a posterior hippocampal resection at the second surgery. The abstract by Olivier et al. in 1988¹³ reported that 20 of 425 anterior temporal resections for epilepsy were
reoperations. Of these, 13 (65%) had not had any hippocampal resection at the initial surgery and seven (35%) had only an anterior/partial resection. The seizure outcome for this group was six (30%) seizure free, three (15%) with a marked reduction, and seven (35%) with worthwhile improvement. Another series reported in 1989²³ included 28 nontumor patients who had a second focal resection for medically refractory epilepsy. Although not specifically stated, it seems that in 16 of these patients the second surgery was an extension of the temporal lobe resection that included additional hippocampus and that 10 of the 16 (62.5%) were reported seizure free at follow-up (duration not reported). It is important to note that one patient had a permanent hemiparesis after reoperation. Three other patients in this series were seizure free after reoperation (frontal lobe structural lesions) and will be discussed later. Subsequently, Wyler et al.²⁴ reported on a study to compare seizure outcome with anterior versus complete hippocampectomy that showed a significantly greater seizure-free outcome with the complete resection.

The Cleveland Clinic in 1991¹ reported on 15 reoperated patients from an overall series of 154 resections for medically refractory seizures. Ten of these patients had their initial operation at their center. The 15 patients were selected from 23 patients who were felt to have “truly intractable seizures” following a resective surgery. In this group, seizure recurrences occurred within 7 months of surgery and reoperation occurred 3 months to 12 years after the first surgery (mean = 38 months). Eleven of the 14 focal resection patients (one patient had a functional hemispherectomy as the second surgery) had invasive electrode investigation before the second resection. Seven of these 14 were seizure free at the last follow-up (range of 8 to 82 months), but four of the 12 had <12 months and only one had >15 months follow-up (82 months). Reoperation in 6 of the 14 involved resection of residual medial temporal structures in patients without structural lesions, and these accounted for three of the seven seizure-free patients and two of the four patients with a >90% seizure frequency reduction. The other eight patients are discussed later. The series had no mortality or serious morbidity.

The Montreal Neurological Institute reported an experience with reoperation on the temporal lobe (for refractory epilepsy) in 1994.⁷ This included 40 of the approximately 260 temporal lobe surgeries performed by the senior author (A. O.) over a 9-year period. The senior author performed the first surgery in 31 of the 40 (nine had been done at other institutions). The original surgery was based primarily on scalp/sphenoidal interictal and ictal evaluation, with 6 of the 40 having depth electrode evaluations due to inadequate lateralization or localization. Recurrent seizures were similar to preoperative seizures and occurred within 6 months in 60% and by 2 years in 90% of the patients. The mean time to reoperation was 5.5 years. Reoperation was based on scalp electroencephalogram (EEG) in 36 patients (31 with temporal and five with mostly temporal, but multifocal, onsets) and depth electrodes in four patients (three with multifocal and one with temporal onsets). At follow-up (range 2 to 11 years; mean of 4.8 years), 21 patients were seizure free and four had rare seizures (one to two per year). All of these patients had resection of residual medial temporal tissue, with some having resection of neocortex as well. Three of the other 15 patients were rendered seizure free by a third operation, one by resection of a missed cavernous malformation, one by resection of additional neocortex, and one by resection of further mesial tissue. Lack of seizure improvement was associated with multifocal EEG abnormalities and seizure recurrence within 1 month of the first surgery. There was no significant permanent neurologic morbidity in the series.

Experience with reoperation at the University of Washington was reported in 1999.¹⁰ This involved 21 patients who had their reoperation between 1991 and 1996 and represented <5% of the epilepsy surgeries performed over that period at that center. At follow-up (1 to 5 years, mean of 3 years), nine (43%) were seizure free and two (9%) had a >95% seizure reduction. There was no surgical morbidity. Only 17 of the patients had focal resections, with two having multiple subpial transections of motor cortex and two (children) having hemispherectomies. The results of the various types of surgery were not reported. Thus, it is not possible to determine how many patients were rendered seizure free by resection of medial temporal structures or residual structural lesions. Eleven of the 17 focal resections were of temporal tissue only, and seven of these included resection of mesial structures. Also, two patients had calcified lesions in the temporal lobe and one had an intraventricular calcified lesion (the pathologies of the series included one ganglioglioma, one cortical dysplasia, and 19 with gliosis). Resections were described as tailored and 11 of the patients had intracranial electrode evaluations. It is possible that all seizure-free outcomes occurred in patients who had either a resection of medial temporal structures or a structural lesion. The study did find two factors that significantly correlated with outcome: (a) that four of the patients had a central nervous system (CNS) infection before the onset of epilepsy and that none of these patients became seizure free, and (b) that the seven patients with a focal magnetic resonance imaging (MRI) abnormality concordant with the ictal EEG onsets before and after the first surgery had a better seizure outcome than the other 14 patients.