Nonlesional MRI findings in patients who have focal epilepsy based on semiology and a surface EEG record may indicate a lower chance for seizure-free postoperative
outcome (1). In contrast, patients with focal epilepsy as a consequence of a circumscribed lesion, such as mesial temporal sclerosis, low-grade neoplasm, vascular malformation, or focal malformation of cortical development, appear to have a better chance of seizure-free outcome (2,3). Use of adjunctive tests such as 2-[¹⁸F]fluoro-2-deoxy-D-glucose-positron emission tomography (FDG-PET), and ictal and interictal single-photon-emission computed tomography may also contribute to identification of the ictal onset zone in nonlesional cases, but identification of an MRI lesion is still more desirable for delineating the potential epileptogenic zone and facilitating subsequent surgical planning.

Location of the epileptogenic zone in a region that can be safely removed compared with proximity of the epileptogenic zone to eloquent cortical areas has obvious implications in surgical planning. If surgical resection is limited as a result of an unacceptable trade-off in acquiring new neurologic deficit, a less optimistic seizure-free outcome is to be expected. Extraoperative invasive recording with eloquent cortical mapping as well as intraoperative electrocorticography may be used to maximize resection without compromising neurologic function, but ultimately success will depend on the completeness of resection of the epileptogenic zone. Perhaps at least partially because of these issues, patients with mesial temporal epilepsy have a better chance for a seizure-free surgical outcome than those with neocortical temporal epilepsy, and patients with temporal lobe epilepsy have a better surgical outcome than those with extratemporal epilepsy (3, 4, 5). Significant limitations on the size of resection arise in perirolandic epilepsy and neocortical frontal or temporal lobe epilepsy in the language-dominant hemisphere.

Of the different etiologies, results of epilepsy surgery in patients with hippocampal sclerosis are most gratifying, with centers consistently reporting more than 80% seizure-free outcomes. Early experience suggests that both selective amygdalohippocampectomies and less-selective temporal resections (including the mesial structures) are equally likely to produce a good surgical outcome (6,7). Memory findings on preoperative neuropsychological testing may ascribe function to the hippocampus on the side of resection, and in such instances a tailored resection that spares the hippocampus (corticoamygdalectomy) has been performed. However, the seizure-free rates drop to 51% after a hippocampus-sparing resection (8), and whether there is a definite benefit in memory outcome if a somewhat “deafferentated” hippocampus is left behind is debatable. More data are required to realize the merits of such a selective surgical procedure.

Surgery in patients with epilepsy due to low-grade neoplasm is also viewed optimistically, and seizure freedom may be realized in a majority of them. In a pediatric series, 86% of patients in whom gross total resection of the neoplasm was accomplished were seizure free at follow-up (9). In an adult series from the same institution involving patients with ganglioglioma and intractable epilepsy, 63% were seizure free at follow-up overall, and a good outcome was likely with a lower age at operation and a shorter duration of epilepsy (10).

Cortical dysplasia results in seizure-free outcome rates ranging from 40% to 60% (5,11). This may be, in part, because MRI can underestimate the extent of a histologic abnormality and, in part, because of the inclusion of patients with diffuse multilobar or hemispheric malformations together with discrete focal cortical dysplasias in surgical series (12,13). Patients with focal cortical dysplasia are more amenable to a complete resection and as a subgroup of all dysplasia patients tend to have a seizure-free outcome comparable to that of patients with hippocampal sclerosis or low-grade neoplasm (2,14,15). In a study including 22 patients with focal cortical dysplasia, resection of the epileptogenic zone guided by histopathologically proven clear margins was significantly associated with an improved seizure outcome (92% seizure free versus 43% when resection was histopathologically proved to be subtotal) (2).

Multilobar or hemispheric malformations of cortical development form the other end of the spectrum of cortical malformations. Surgical treatment is very effective, and the method of choice is usually an extensive multilobar resection or hemispherectomy. In 12 children who underwent functional hemispherectomy for diffuse cortical malformation involving most or all of one hemisphere, seizure freedom or greater than 90% reduction in seizure burden was achieved in more than 80% of patients; 50% of patients were seizure free (13). In this series, patients with hemimegalencephaly were less likely to achieve seizure freedom (40% seizure free) compared with patients with hemispheric atrophy or relative sparing of one or part of a lobe as evident on MRI (83% seizure free). Possible explanations for this observation include likelihood of epileptogenic potential from abnormal subcortical structures of the removed hemisphere or the presence of contralateral epileptogenicity caused by microscopic abnormalities. Despite these challenges, surgery significantly lessened the burden of disease and must therefore be considered in patients with refractory epilepsy.

Approximately 15% of patients with hippocampal sclerosis evident on MRI also have evidence of cortical dysgenesis. Such “dual pathology” is also well recognized in neuronal migration disorders, low-grade neoplasms such as
gliomas or dysembryoplastic neuroepithelial tumors, porencephalic cysts, vascular malformations, and reactive gliosis (16). A less favorable postoperative seizure outcome is more likely with coexisting dual pathology when only one abnormality is removed; therefore, surgical planning should be appropriately modified to include both lesions when possible (17,18). Removal of mesial temporal structures in addition to a temporal neoplasm was noted to result in a better outcome (92.8% seizure free) compared with that seen in patients who underwent resection of the tumor only (18.8% seizure free) (19). As with localization-related epilepsies in general, large resections are likely to be associated with a better outcome. Equally important, reoperation to extend the surgical resection should be considered if the initial resection is ineffective (18,20). In one series, more than 60% of patients with initial surgical failures (after lesionectomy alone) became seizure free after subsequent reoperation to remove the remaining epileptogenic tissue (19).

Medically Refractory Status Epilepticus (RSE) of focal origin, a medical emergency that may be encountered when seizures cannot be controlled despite high-dose suppressive therapy (HDST) to pharmacologic coma (21,22), is associated with significant morbidity and mortality. Studies report universal morbidity in survivors (23) and mortality ranging between 16% and 43.5% (24, 25, 26). If status epilepticus is deemed to be of focal origin, resective surgery may be an option. Scattered case reports have suggested that multiple subpial transections, cortical resection, corpus callosotomy, or implantation of a vagus nerve stimulator may be effective for controlling seizures in RSE (27, 28, 29, 30). The concern of compounding risks of neurosurgery together with those of status epilepticus is a genuine one and may be one reason for underuse of a surgical approach in RSE of focal origin. However, uncontrolled status epilepticus and prolonged maintenance of pharmacologic coma are themselves not without risks, and these may even outweigh the risks of neurosurgical intervention (23, 24, 25, 26). At tertiary epilepsy surgery centers with sufficient surgical experience and specially trained personnel, the surgical risk may be even lower and must be balanced against the risks of prolonged HDST. In a surgical series of 10 pediatric patients with RSE of focal origin in whom at least 2 weeks of HDST had failed, resective surgery was able to stop seizures acutely in the entire group (31). These patients had experienced morbidity during the prolonged phase of medical management in the intensive care unit. In contrast, no mortality and no substantial morbidity as a consequence of surgery occurred. After a median follow-up of 7 months, 7 of the 10 patients were seizure free, and 3 had a significant improvement in their epilepsy. These initial data highlight a potential role for resective surgery in a clinical situation of very limited therapeutic options and substantial risks for prolonged RSE and pharmacologic coma (31).

Diffuse EEG abnormalities in the presence of a focal MRI lesion constitute another special situation when epilepsy surgery may occasionally be an appropriate option. The presence of generalized interictal and ictal abnormalities on surface electroencephalography, with only a few exceptions, is considered an unfavorable sign for surgical candidacy. However, in the early 1990s, it was first recognized that selected infants with diffuse hypsarrhythmia and nonlocalizable scalp ictal EEG patterns during spasms became seizure free after resection of congenital lesions seen on PET and MRI scans (32). The immaturity of networks is felt to be the basis of such a diffuse EEG phenotype in the presence of a focal lesion during infancy. The interaction between focal epileptogenesis and a developing brain beyond the first year of life is not well known. It is widely appreciated that the developing brain undergoes dynamic changes and that there is an overabundance of synapses, which peak by about 2 years of age and remain high for most of the first 10 years of life, before decreasing to adult levels between 15 and 20 years of age (33,34). It is therefore conceivable that owing to the presence of a focal insult during early life, a diffuse EEG phenotype may be encountered beyond infancy.