Clinical aspects

Description

Presurgical evaluation. To obtain seizure freedom and improved quality-of-life as outcomes of epilepsy surgery, careful presurgical evaluation and appropriate operative approach are necessary. Presurgical evaluation should delineate the epileptogenic zone, defined as “the area of cortex indispensable to the generation of epileptic seizures,” and its relationship to eloquent cortex (112). At present, the epileptogenic zone cannot be localized directly by any single diagnostic test. Structural abnormalities and areas of functional deficit must be identified as part of the process (42; 112).

Multiple complementary diagnostic tests are used for evaluation prior to epilepsy surgery, with different techniques favored at different epilepsy centers. Preoperative studies include EEG, MRI, radioisotope imaging (PET and SPECT), neuropsychological testing, the intracarotid amobarbital (Wada) test, and more recently developed techniques such as MEG and fMRI. When test results are concordant for the epileptogenic zone, and its spatial relationship to eloquent cortex clarified, the operative approach can be planned (42; 112; 90).

Tests of epileptic excitability. All centers use noninvasive EEG, starting with routine outpatient interictal studies. To localize areas of seizure onset and verify the epileptic nature of clinical events, patients are also admitted to undergo noninvasive video-EEG monitoring designed for simultaneous recording of electrographic ictal events and clinical seizures (42; 35; 74). Other tests used to localize seizure onset include ictal SPECT, subtraction ictal SPECT coregistered to MRI (SISCOM), and ictal MEG (42; 39; 90; 145). The integrated technique of simultaneous electroencephalography and functional magnetic resonance imaging (SEM) also may demonstrate areas of perfusion change temporally related to interictal epileptiform discharges or the ictal onset, and thereby provide another image representation of the irritative and ictal onsets zones (137).

MRI and structural abnormalities. To identify structural brain abnormalities, high-resolution MRI with field-strength of 1.5 Tesla or greater is used with phase-arrayed surface coils that improve the signal-to-noise ratio (63; 66). These high-resolution MR scans should be obtained with a dedicated epilepsy-protocol adequate for the detection of hippocampal sclerosis and cortical dysplasias as well as lesions such as tumors and vascular malformations. T2-weighted and FLAIR sequences are sensitive to hippocampal sclerosis, whereas 3D T1-weighted gradient-echo and inversion recovery sequences aid in the detection of cortical dysplasias. If a mesial temporal etiology of the patient’s epilepsy is suspected, slices are angulated along the long axis of the hippocampus. If primary motor cortex, primary sensory cortex, or the dorsal frontal lobe is suspected, angulation is along the anterior commissure-posterior commissure line. When the acquired images raise suspicion for potentially enhancing lesions, such as tumors or vascular malformations, additional axial and coronal T1-weighted spin-echo sequences are obtained before and after gadolinium injection (112; 66; 144).

According to retrospective studies of MR imaging of epilepsy patients, both sensitivity for focal abnormalities and specificity for correct neuropathologic diagnosis are significantly higher for dedicated epilepsy-MR-protocols than for standard MR protocols, and highest when epilepsy-protocoled MR scans are analyzed by neuroradiologists with epilepsy expertise (88; 147). With respect to sensitivity, in one retrospective study of MR brain scans of 123 epilepsy patients, standard MR interpreted by nonspecialist radiologists or non-epilepsy-trained neuroradiologists revealed focal lesions in only 39% of the cases, whereas, for the same patients, dedicated epilepsy-protocoled MR scans analyzed by epilepsy-specialist neuroradiologists showed focal abnormalities in 91%. With respect to specificity, prediction of neuropathologic diagnosis was correct in only 22% of cases imaged via standard MR and analyzed by nonepilepsy-specialist radiologists, as opposed to 89% when imaged via epilepsy-protocoled MR and interpreted by epilepsy-specialist neuroradiologists (147). Based on this evidence, MR brain scans of epilepsy patients should be performed with a dedicated epilepsy-protocol specifying particular sequences, slice orientation, and slice thickness, and images should be interpreted by neuroradiologists with specific epilepsy expertise (03; 112; 147).

Tests of functional deficit. To delineate areas of functional deficit, all centers use batteries of standardized neuropsychological tests. Most centers also routinely use interictal 18F-FDG-PET, which reveals localized areas of cerebral hypometabolism and is highly sensitive (90% sensitivity) for mesial temporal lobe epilepsy (126).

Tests of normal cortical function. To investigate functional (eloquent) areas of cortex that cannot be resected without causing new neurologic deficits, normal cortical function can be studied via the intracarotid amobarbital (Wada) test, electrical stimulation cortical mapping, and fMRI. In the intracarotid amobarbital test, pharmacologic inactivation of cortex supplied by the anterior and middle cerebral arteries in each hemisphere is achieved via left and right intracarotid injection of sodium amobarbital. As a method to lateralize language and test memory function, the intracarotid amobarbital test is an integral part of presurgical evaluation for the epileptic patient whose planned resection involves cortex potentially serving language and memory functions (112; 01; 92).

Invasive studies. If the data from diagnostic studies are concordant with respect to the epileptogenic zone, there is no need for invasive testing. Noninvasive studies are typically sufficient as presurgical evaluation for the majority of epilepsy patients; for example, 70% to 80% of mesial temporal lobe epilepsy patients are successfully localized noninvasively (42; 112; 127). If the test data are discordant, a hypothesis concerning the location of ictal onset is formulated. On the basis of this hypothesis, patients may undergo invasive tests such as ictal EEG recordings via intracranial depth or subdural strip and grid electrodes. Stereotactic implantation of intracerebral depth electrodes guided by adequate preoperative imaging studies has been shown to be safe in a retrospective analysis of 100 implanted patients (143). In regard to efficacy, invasive electrodes are a demonstrably reliable method of ictal onset localization, with the caveat that inadequate spatial sampling can be a limitation (30; 33; 143; 89). In one retrospective study, sampling error was the cause of failure to localize ictal onset via intracranial EEG recording in 12% (13/110) of patients with partial epilepsy who underwent this invasive testing (119; 112).

Surgical procedures.

After presurgical evaluation has identified the epileptogenic zone, the operative procedure most likely to benefit the patient is planned. Epilepsy surgery includes both resective and disconnective procedures. Resective epilepsy surgery is comprised primarily of (a) anteromesial temporal resections and (b) neocortical resections, including both lesionectomy and resection of electrographically abnormal regions that have no structural correlate visible on MRI. Disconnective procedures include (c) multiple subpial transection, (d) corpus callosotomy, and (e) functional hemispherectomy (The latter two procedures are performed primarily in children and will not be discussed in this article).

The majority of resective procedures performed for epilepsy are variants of anteromesial temporal resections because mesial temporal lobe epilepsy is both the most common and the most medically refractory focal epilepsy (42; 116). In addition, a disconnective procedure for temporal lobe epilepsy referred to as temporal lobotomy has been reported in a small case series with preliminary outcome data (Smith et al 2004).

Anteromesial temporal resections. Anterior temporal lobectomies can be divided into (1) standardized anatomic resections of the anterior temporal lobe and (2) tailored resections. At most epilepsy surgery centers, standardized anterior temporal lobectomy is performed when concordant data gathered during presurgical evaluation supports the diagnosis of mesial temporal lobe epilepsy associated with hippocampal sclerosis (57). Tailored resections are typically performed in those temporal lobe epilepsy cases that do not fit the stringent criteria for mesial temporal lobe epilepsy associated with hippocampal sclerosis and in those cases in which the epileptogenic zone is in dangerous proximity to the eloquent cortex (100).

Standardized anterior temporal lobectomy. Standardized anatomic resections for mesial temporal lobe epilepsy are anatomically uniform operations. Standardization of the anterior temporal lobectomy is based on two principles. The first principle is that a known pathology is consistently limited to the same specific structures in a well-defined subset of mesial temporal lobe epilepsy cases. The second principle is that the location of functional cortical areas is grossly uniform in most mesial temporal lobe epilepsy patients who have undergone adequate presurgical evaluation, so that the safety of standardized resection can be predicted (100; 100).

Several types of standard anatomic resections have been developed for mesial temporal lobe epilepsy, and use varies at different institutions. Their primary difference is the variable volume of mesial temporal structures and lateral temporal structures that are resected (57). The 3 most well-known types of anteromesial temporal resection for mesial temporal lobe epilepsy are: (a) the en bloc anterior temporal lobectomy, (b) the anteromedial temporal lobectomy, and (c) selective amygdalohippocampectomy.

(a) En bloc anterior temporal lobectomy. The en bloc anterior temporal lobectomy, conceived by Falconer and further developed by Crandall, consists of the one-stage removal of temporal structures in one block, including the anterior 3 cm of the hippocampus proper, the uncus, and the dorsolateral amygdala (52; 32; 57). Maintenance of the anatomic relationship among these structures is the advantage of this approach, given that it permits en bloc histologic analysis of the hippocampal formation.

(b) Anteromedial temporal lobectomy. The anteromedial temporal lobectomy developed by D Spencer at Yale minimizes resection of lateral temporal cortex and allows safe removal of mesial temporal structures by means of a 2-part procedure. The first step consists of resection of a small block of lateral temporal cortex, including the anterior middle temporal gyrus and the inferior temporal gyrus, sparing the superior temporal gyrus. This lateral temporal cortical resection is taken to a depth of 2.5 to 3 cm; with respect to length, the resection extends approximately 3.5 cm from the temporal tip in the nondominant hemisphere and 3 cm in the dominant hemisphere. In the second step, the microscope is used to resect mesial temporal structures, including part of the amygdala and the uncus as well as the hippocampus proper and part of the adjacent parahippocampal gyrus (125; 57; 57).

(c) Selective amygdalohippocampectomy. The selective amygdalohippocampectomy performed via the transsylvian approach was developed by Yasargil to obtain access to mesial temporal structures while avoiding both excessive retraction of the brain and injury to the lateral temporal cortex. In this maneuver Yasargil adapted his pterional “keyhole technique”, opening the sylvian fissure to approach the mesiobasal region. Specifically the selective amygdalohippocampectomy consists of (i) microsurgical resection of the amygdala, the uncus, and parts of the hippocampal formation--the anterior hippocampus proper and the anterior parahippocampal gyrus and (ii) interruption of certain fiber tracts--eg, the anterior commissure and the uncinate fasciculus (156; 163; 154).

Different standardized anterior temporal resections are favored by different epilepsy surgery groups. Careful presurgical evaluation to guide appropriate patient selection is crucial to the success of any of the standardized anterior temporal lobe resections, regardless which approach is selected.

Tailored anterior temporal lobectomy. When standardized anatomic resection for temporal lobe epilepsy is not indicated due to discordant preoperative testing, anterior temporal resection can be tailored according to intracranial electrophysiologic data acquired intraoperatively or extraoperatively. From the 1930s through the 1950s, a period of rapid progress in clinical neurophysiology, epilepsy surgery techniques for tailoring resections via intraoperative electrocorticography and cortical electrical stimulation were developed. These operative advances were accomplished by Penfield and Jasper at the Montreal Neurologic Institute in awake patients and by Bailey and Gibbs in Chicago in thiopental-anesthetized patients (104; 07; 105).

Over the past decade, improved MRI techniques have produced a high degree of sensitivity and specificity in the preoperative identification of mesial temporal lobe epilepsy patients who--if they undergo surgical treatment--will have a surgical specimen with pathologic diagnosis of hippocampal sclerosis (147). For this subset of patients with unilateral well-defined mesial temporal lobe epilepsy, tailoring anterior temporal resection with intraoperative electrocorticography is not indicated because anterior temporal margins based on intraoperatively recorded interictal epileptiform discharges do not correlate with postoperative seizure outcome (142; 115). However, the results of one prospective study of temporal lobe epilepsy patients undergoing surgery do support the use of intraoperative electrocorticography confined to the hippocampus proper to decide the extent of hippocampal resection (91).

For cases of temporal lobe epilepsy that do not fit the diagnostic criteria for mesial temporal lobe epilepsy, and for extratemporal neocortical partial epilepsies, intraoperative electrocorticography is at present routinely used to tailor resection (99), though the published data regarding efficacy are conflicting (138). For such patients, extraoperative intracranial EEG recorded via depth or subdural grid and strip electrodes may be used as an alternative or adjunct to intraoperative electrocorticography.

Neocortical resections. Neocortical resections include removal of epileptogenic areas from the lateral temporal neocortex and from extratemporal neocortex. With respect to temporal versus extratemporal resective epilepsy procedures, the lobar distribution in surgical series from different institutions consistently shows many fewer extratemporal procedures performed. One series of 360 resections for epilepsy (in adults and children) consisted of 78% temporal procedures, 16% frontal, 3% parietal, 3% occipital, and less than 1% multilobar (12). Another surgical series of 79 resective epilepsy procedures (limited to adults) included 76% temporal procedures, 15% frontal, 1% parietal, 1% occipital, and 4% multilobar (51). Rasmussen’s personal series from the Montreal Neurologic Institute showed a similarly higher percentage of temporal cases (74%) compared to extratemporal resections for epilepsy (14%) (65).

A primary reason that many fewer extratemporal resections are performed for epilepsy is the heterogeneity of the extratemporal epilepsies (90; 83). Mesial temporal lobe epilepsy is a surgically remediable syndrome with distinct clinical characteristics and uniform underlying pathology, which can be treated with a standardized surgical procedure. In contrast, extratemporal epilepsies have a multitude of different etiologies and variable proximity to functionally critical cortical areas.

With respect to their multiple different causes, the extratemporal epilepsies can be categorized as (1) lesional or (2) nonlesional. (1) In lesional extratemporal epilepsies, the epileptogenic zone corresponds to a radiographic lesion. (2) In nonlesional cases, the epileptogenic zone is a radiographically-invisible region that is electrographically abnormal. These cases include a subset in which no histologic abnormality is identified postoperatively (83).

Lesional neocortical cases. Approximately 30% of extratemporal resections performed for epilepsy are lesionectomies. For example, in one survey of 100 epilepsy surgery centers over the period 1986 to 1990, reported at the 1992 Palm Desert Conference on Epilepsy, 27% of the neocortical extratemporal resections for epilepsy were lesionectomies (298/1098) (49).

In evaluating lesionectomy as a treatment for epilepsy, the focal abnormality on MRI is a positive prognostic factor for good seizure control outcome, especially for tumors and vascular malformations (123; 06; 55; 96). However, neuroimaging alone is not sufficient for presurgical evaluation in lesional epilepsy cases. If the primary indication for the resection of a radiographic lesion is a patient’s apparent seizure disorder, EEG studies--including video scalp EEG--are crucial, given that structural abnormalities are not always epileptogenic. Some small radiographic lesions, such as venous angiomas, do not cause seizures, and some patients’ transient neurologic symptoms may actually be nonepileptic events (71). For this reason, presurgical evaluation for patients whose epilepsy is attributed to a radiographic lesion should include complementary electrophysiologic, anatomic, and functional studies, just as in nonlesional epilepsy cases.

In addition to noninvasive EEG, intraoperative electrocorticography has a vital role in guiding surgical decisions in lesional epilepsy cases. For both lesional frontal lobe epilepsy and for focal cortical dysplasias, the importance of intraoperative electrocorticography for postoperative seizure control outcome has been well-documented (101; 150; 149). For this reason, when the term “lesionectomy” is used in this article, resection of the lesion alone is not necessarily meant. For many lesional epilepsy cases, electrophysiologic techniques should be used to guide resection of epileptogenic tissue surrounding the lesion (168).

Radiographic lesions that are epileptogenic include tumors (both primary brain tumors and metastases), malformations of cortical development, and vascular malformations. Approximately 3% to 15% of the entire epilepsy population has brain tumors. Primary brain tumors with epileptogenic propensity are typically low-grade and can be of glial or glial and neuronal cell origin. These include gangliogliomas, dysembryonic neuroepithelial tumors (DNET), astrocytomas, oligodendrogliomas, and pleomorphic xanthoastrocytomas (168; 151; 165). With regard to malformations of cortical development, focal cortical dysplasias are the most commonly reported in surgical series of lesional epilepsy and may be the underlying cause of seizures in 20% to 30% of all patients with focal epilepsies (29; 53); many cases of focal epilepsy initially described as nonlesional on MRI have been shown after resection and analysis of the pathologic surgical specimen to be caused by cortical dysplasias (89). Vascular malformations are the cause of epilepsy in approximately 5% of epileptic patients, with arteriovenous malformations and cavernous malformations commonly presenting with seizures (151).

Cavernous malformations as epileptogenic lesions. With regard to cavernous malformations, the iron deposited as hemosiderin in perilesional brain tissue due to these lesions’ chronic microhemorrhages is thought to be epileptogenic (158; 135; 157; 05; 11). Seizures are in fact the most common initial presentation of cavernous malformations when located supratentorially and often develop after diagnosis of cavernous malformations in initially seizure-free patients (111; 166; 73; 107; Moriarity et al 1999; 17). In one prospective study of 68 patients with cerebral cavernous malformations, approximately 50% of the patients (33/68) presented with seizures, and the risk of new-onset seizures for those who were seizure-free at initial presentation was 2.4% per patient-year (Moriarity et al 1999). The percentage of initially seizure-free patients with cerebral cavernous malformations who later had seizures ranged from 4.3% to 11% in 3 prospective studies (166; 73; Moriarity et al 1999).

Patients with seizures due to cavernous malformations often go on to develop epilepsy. In a large retrospective literature review of 296 cases of supratentorial cavernous malformations, 65% of patients had epilepsy (93). With regard to medically intractable epilepsy in patients with symptomatic cavernous malformations, reports vary widely depending on patient selection for particular studies: from 10% of patients in a prospective study of 68 patients with cerebral cavernous malformations (both supra- and infratentorial) (Moriarity et al 1999), to 44.7% of patients in a retrospective surgical series of 47 patients with epileptogenic supratentorial cavernous malformations (22). In a recent large multicenter study of 168 epileptogenic supratentorial cavernous malformations, 58% of patients had pharmacologically refractory epilepsy compared to 28% with epilepsy that improved with antiepileptic drug therapy, ie, that was not refractory to medication (10). Postoperative seizure control outcomes of operative treatment for epileptogenic cavernous malformations are addressed below in the Outcome section.

Nonlesional neocortical cases. Seventy-three percent of extratemporal resections for epilepsy were nonlesional cases (805/1098) in the survey of 100 epilepsy surgery centers for the period 1986 to 1990, obtained for the 1992 Palm Desert Conference on Epilepsy (49). In nonlesional cases, defined as cases in which no structural abnormality can be visualized radiographically, identification of the epileptogenic zone can be extremely difficult. Due to rapid spread of epileptic activity, both seizure semiology and scalp EEG often are not helpful in localizing ictal onset and may even be misleading. For this reason, multiple diagnostic modalities--especially intracranial EEG recording--must be used to mark the boundaries of the epileptogenic zone (90; 83).

Multiple subpial transection. Disconnective procedures in adult epilepsy surgery include multiple subpial transection, less commonly corpus callosotomy, and extremely rarely in adults, hemispheric disconnection (functional hemispherectomy). The latter 2 procedures are performed most often in pediatric patients and will not be discussed in this article.

Multiple subpial transection is performed for patients with medically intractable focal epilepsy in whom the epileptogenic zone is unresectable due to its congruence with eloquent cortex, for example, primary sensorimotor cortex or cortex supporting language, memory, or visual function. The objective of the procedure is to avoid causing an unacceptable neurologic deficit when operating to eliminate a patient’s seizures (95).

Morrell and his group developed the technique of multiple subpial transection for the purpose of interrupting horizontal connections between neurons while preserving both vertically-oriented cortical-subcortical projection fibers and vertically-oriented blood vessels (95; Wyler et al 1995). They conceived this technique by synthesizing 3 findings from experimental neuroscience: the discoveries that (a) that the functional organization of cerebral cortex is based on vertically-oriented cortical columns, (b) that epileptic discharge is based on synchrony of neuronal firing, which in turn requires “horizontal interaction of cortical neurons”, and (c) that the normal physiologic function of cerebral cortex remains intact after interruption of horizontal intracortical fibers (95; 122). Prior to operating on patients, Morrell’s and other groups demonstrated the safety and efficacy of multiple subpial transection in experimental animals (95).

Morrell then performed this technique in human epilepsy patients and showed both good seizure control outcomes and preservation of function (95). Morrell’s original description of his technique guides neurosurgeons today. To perform multiple subpial transection, intraoperative electrocorticography is used to map the area to be transected, and multiple parallel cuts are then made subpially across the width of the gyri (95; Wyler et al 1995). Given experimental evidence that synchronous epileptic discharge of neurons requires at least 5 mm of contiguous tissue in intact brain, Morrell specified a transection interval of 5 mm (95). Morrell initially used a steel wire with a 4 mm tip angled upward at 90 degrees into a hook; Wyler modified the transecting instrument by angling its tip downward (95; 161). Wyler has advocated performing multiple subpial transection only under the operating microscope for better visualization than is possible with loupes (161). Intraoperative electrocorticography is often repeated at the end of the operation to confirm completion (97).

The goal of epilepsy surgery is the elimination of disabling seizures without causing new or unacceptable neurologic deficits (42; 112). In other words, its purpose is to “cure” the patient’s seizure disorder. Though some specialists prefer that the term “cure” be limited to patients who are both seizure-free and off all antiepileptic drugs (18), most outcome studies classify seizure-free patients on antiepileptic drugs as having an excellent surgical outcome (152). At present, most epilepsy surgeons and epileptologists consider postoperative seizure-free patients who are maintained on their (presurgical) antiepileptic drug regimen to be cured. The majority of postoperative patients are in fact routinely maintained on antiepileptic drugs for a minimum of 2 years after resection and often over their lifetimes.

The primary measured endpoints of epilepsy surgery are (1) seizure control and (2) quality of life (When evaluated, cognitive outcomes are also typically assessed quantitatively).

Seizure control outcome measurement.

Engel classification. The standard classification of outcome with respect to postoperative seizure control developed by Engel is widely used in the epilepsy community. Patients with no postoperative seizures, no disabling seizures, or only isolated auras are designated as Engel Class I. Patients who experience only rare seizures, eg, 2 or fewer per year if disruptive, 2 or 3 per year if minimally symptomatic, or nondisabling nocturnal seizures alone, are categorized as Engel Class II, and are also considered to have a good outcome. Engel Class III refers to patients who have a “worthwhile” reduction in seizure frequency, and Engel Class IV is reserved for patients with “no worthwhile improvement” postoperatively (41; 47). Different epilepsy surgery centers interpret Engel Classes III and IV variably; for some, “worthwhile” means a greater than 75% reduction in seizure frequency, while others define it as greater than 50% or 90% (27).

ILAE classification. To quantify seizure control outcome after epilepsy surgery more precisely, Wieser and colleagues proposed a new classification based on postoperative “seizure days” (ie, the number of days when seizures occur), rather than on the absolute number of postoperative seizures (153). This 6-class proposal was published by the International League Against Epilepsy as a formal commission report and has been used in assessments of postoperative seizure outcome at specific institutions (153; 155).

Quality of life outcome measurement.

To measure epilepsy patients’ quality of life, Devinsky and colleagues developed the Quality of Life in Epilepsy (QOLIE-89) inventory, an 89-item health-related quality of life (HRQOL) instrument specifically targeted to epilepsy. In a systematic study of 304 adult epilepsy patients, they demonstrated the reliability and validity of the QOLIE-89 as a HRQOL measurement for epilepsy (34). The QOLIE-31, a shorter form of the 89-item measure, was subsequently developed (31). Both the QOLIE-89 and the QOLIE-31 measures are responsive to changes in clinical status over time and, thus, have longitudinal validity (19). The QOLIE-89 is commonly used to quantify postoperative quality of life for epilepsy patients after resective epilepsy surgery (152; 128).

Health-related quality of life (HRQOL) is often assessed in cost-utility analyses. For this purpose, standardized, generic, preference-based instruments – rather than disease-specific instruments such as the QOLIE-89 - are recommended. To identify the best preference-based instrument for cost-utility analysis of epilepsy treatment, multiple preference-based HRQOL instruments have been evaluated for validity and responsiveness in assessing HRQOL in chronic epilepsy. The SF6D was identified as the most valid and responsive preference-based instrument for chronic epilepsy, though for obvious reasons it was less valid and responsive than the epilepsy-specific QOLIE-89 instrument (81). The SF6D may be viewed at this time as the most appropriate generic preference-based instrument for assessment of HRQOL in cost-utility analyses of epilepsy surgery.

Indications

Medically intractable seizures. Patients with frequent and disabling seizures that are medically intractable are candidates for epilepsy surgery and should be referred to epilepsy specialty centers for evaluation. Typically, seizures that impair consciousness are disabling, whereas simple partial seizures are not. However, there are no formal criteria for “disabling” or intolerable frequency; these are subjective categories best decided by the patient in collaboration with an epilepsy specialist (48; 132; 127).

There has also been no formal definition of medical intractability for epilepsy (42). However, observational studies of newly diagnosed epilepsy patients have provided evidence that a large number of medically refractory cases manifest their intractability early in the course of disease (75; 76; 78; 84). Kwan and Brodie have shown that for patients with newly diagnosed epilepsy, failure to respond to the first antiepileptic drug is a powerful prognostic factor for the development of medically refractory epilepsy. Their observational study included 470 previously untreated epilepsy patients, with median five-year follow-up, and found that only 11% (12/113 patients) of the patients for whom the first antiepileptic drug lacked efficacy later became seizure-free with medical treatment (75).

After failure of the first antiepileptic drug due to lack of efficacy, seizure-free rates with subsequent antiepileptic drugs were low for both duotherapy (add-on) or alternative monotherapy (substitution). These 2 treatment options--antiepileptic drug add-on versus substitution--showed no significant difference in their achievement of seizure control: add-on, 26% seizure-free rate (11/42 patients), substitution, 17% seizure-free rate (6/35 patients); p=0.25 (76). Further analysis found that patients unresponsive to 2 antiepileptic drugs are also unlikely to respond to a third. Specifically, none of the 11 patients whose seizures were not controlled by 2 tolerated serial monotherapies and who then received an add-on antiepileptic drug became seizure-free (76). In addition, only 1% (6/301) of the newly diagnosed epilepsy patients who became seizure-free with medical treatment achieved this during monotherapy with a third antiepileptic drug (75).

The outcome data provide strong evidence that after 2 tolerated antiepileptic drugs have proved ineffective for an epilepsy patient, further antiepileptic drug manipulations are unlikely to achieve seizure control (75; 76; 78). Though not all cases of refractory epilepsy manifest intractability early (14), many cases are evident soon after diagnosis. In such cases, referral for epilepsy surgery evaluation should not be delayed in order to attempt multiple antiepileptic drug regimens (76; 78). Subsequent antiepileptic drugs are unlikely to control seizures, and delay may be injurious to the patient (75; 78; 84).

Escalating antiepileptic drug dosage in an effort to obtain seizure control also has a low likelihood of success and should not be a cause for delaying surgical evaluation. Of the above discussed cohort of 470 newly diagnosed epilepsy patients, more than 90% of those for whom the first prescribed antiepileptic drug was effective became seizure-free at moderate antiepileptic drug doses. Given this data, increasing antiepileptic drug dosage to near-toxic levels in an effort to achieve seizure control is not indicated (77). High antiepileptic drug doses do not increase rates of seizure control; they are likely only to increase adverse reactions and undesirable side-effects such as sedation. In addition, there is no need to delay surgical referral in order to try newer antiepileptic drugs. Multiple studies have shown that for epilepsy patients who manifest medical intractability early, recently approved antiepileptic drugs had no greater efficacy than older antiepileptic drugs such as phenytoin (75; 15).

Surgically remediable syndromes. Generally, resective epilepsy surgery is indicated for patients who have medically intractable partial epilepsies with a localized epileptogenic zone that does not overlap with functionally critical (eloquent) cortex. Both mesial temporal lobe epilepsy and discrete neocortical lesions are considered surgically remediable focal epilepsies, with resection performed for a curative goal. Secondarily generalized epileptic syndromes such as Lennox-Gastaut syndrome in children and adults are considered surgically remediable from a palliative viewpoint because corpus callosotomy can stop drop attacks but in general does not ameliorate the patient’s other types of seizures (42; 90).

Mesial temporal lobe epilepsy. Epilepsy patients diagnosed with surgically remediable syndromes such as mesial temporal lobe epilepsy should be referred early for surgical evaluation. Mesial temporal lobe epilepsy is a recognizable syndrome with a typical history, an unremarkable neurologic examination except for short-term memory deficits in some patients, and complex partial seizures. Specific diagnostic criteria for mesial temporal lobe epilepsy include: (a) anterior temporal site of interictal epileptiform discharges and ictal onsets on scalp EEG; (b) hippocampal sclerosis visualized on MRI as high signal on T2-weighted images and as decreased size of hippocampal formation best seen on T1-weighted images; (c) ipsilateral temporal lobe hypometabolism on interictal FDG-PET; and (d) cognitive dysfunction lateralized to the involved temporal lobe on neuropsychological testing. The pathologic substrate of mesial temporal lobe epilepsy is hippocampal sclerosis, a distinct histologic pattern of neuronal cell loss and gliosis in the hippocampus. The neuronal cell loss of hippocampal sclerosis described by Sommer in 1890 is currently defined quantitatively as a loss of greater than 30% of neurons compared to control values. The loss of neurons found in hippocampal sclerosis affects all areas of the hippocampus, though CA2 is typically less affected than other regions (146; 154; 42; 127).

Of the focal epilepsies, the surgically remediable syndrome of mesial temporal lobe epilepsy has both the highest rate of cure with resective epilepsy surgery and the highest rate of medical intractability. In studies comparing patients with focal epilepsies of different etiologies, 58% to 89% of patients with mesial temporal lobe epilepsy did not achieve seizure control with antiepileptic drugs (116; 136). Given the high probability of medical refractoriness, a patient whose epileptic disorder meets diagnostic criteria for mesial temporal lobe epilepsy and who has failed to respond to 2 antiepileptic drugs should undergo evaluation for epilepsy surgery as soon as possible (43; 127; 44; 44).

Contraindications

General and particular contraindications. Whereas progressive neurologic disease is a general contraindication to epilepsy surgery, there are some contraindications to epilepsy surgery that are particular to specific resective procedures. For example, poor memory function contralateral to the epileptogenic zone is a contraindication specific to anterior temporal lobectomy for mesial temporal lobe epilepsy, given that mesial temporal structures supporting memory are resected. In the intracarotid amobarbital (Wada) test, memory performance after injection ipsilateral to the epileptogenic zone in the dominant (left) hemisphere is significantly predictive of the verbal memory decline that will occur after left anterior temporal lobectomy (72; 23). When such functional studies show that the patient’s memory is not sufficiently supported by the mesial temporal structures contralateral to the epileptogenic zone, the patient is at risk for disabling postresection memory deficits and should not undergo resection (85; 127; 01). In addition, patients whose Wada results indicate high memory capacity of the epileptogenic mesial temporal lobe, especially in the dominant hemisphere, (ie, good material-specific memory score after contralateral injection), are also at risk of significant memory decline after anterior temporal lobectomy (02).

Older age not a contraindication. Older age, low IQ, and medical and psychiatric diseases were thought in the past to be contraindications to resective epilepsy surgery, but with appropriate patient selection none is now so considered (127). For older adults (age 50 years or greater) with refractory temporal lobe epilepsy, anterior temporal lobectomy was found to be both safe and beneficial in 2 separate retrospective analyses of older versus younger patients, though the probability of achieving seizure freedom was not as high in older patients as in younger patients (121; 20). One of the studies found that similar percentages of older and younger patients who gained postoperative seizure-freedom started to drive, an important psychosocial marker of independence (121).

A more recent retrospective study comparing the outcomes of 2 different age cohorts – older than 50 versus younger than 50 - after resective surgery for medically intractable temporal lobe epilepsy yielded different results. The older patients were just as likely to obtain postoperative seizure freedom as the younger patients; however, perioperative complications occurred significantly more frequently in the older group (64). It can be concluded from these 3 studies that regardless of age, patients suffering from medically intractable epilepsy may attain excellent seizure control after resective surgery. Careful patient selection with preoperative medical risk management may decrease the likelihood of surgical morbidity.

Relative contraindications. Likewise, low IQ in itself is not a contraindication to resective procedures for epileptic patients with a well-defined resectable epileptogenic zone. Postoperative seizure control outcomes in this population are similar to those in epileptic patients with normal IQ (62). Also, medical and psychiatric diseases are not necessarily contraindications to epilepsy surgery. They are contraindications only if they increase the risk of any required invasive presurgical test or of operation itself to a level higher than the possible benefit from surgery (127).

Contraindications to resection that are indications for different procedures. Contraindications differ for resective versus disconnective epilepsy surgery. Idiopathic generalized epilepsy is a contraindication to resective epilepsy surgery, given the absence of focal functional and structural abnormalities in a discrete resectable area of cortex. In focal epilepsies, eloquent cortex in the epileptogenic zone is a contraindication to resection because its removal may result in devastating neurologic deficit. In these cases, disconnective surgical procedures may instead be performed, for example, multiple subpial transections of the epileptogenic zone overlapping eloquent cortex, or corpus callosotomy for some secondarily generalized epilepsy syndromes (48; 42; 112; 130).

Outcomes

Outcome measures of epilepsy surgery include (1) seizure control and (2) quality of life. Postoperative outcomes differ according to the etiology of the patient’s epilepsy and the type of operation performed.

Seizure control.

Seizure control after anterior temporal lobectomy. Only for anterior temporal lobectomy as treatment for temporal lobe epilepsy is the highest class of evidence regarding outcomes available. This Class I evidence consists of data from a prospective randomized controlled clinical trial with masked outcome assessment designed by Wiebe and colleagues that compares surgical and medical therapy for temporal lobe epilepsy due to any cause. Wiebe and his group found anterior temporal lobectomy significantly superior to antiepileptic drugs for temporal lobe epilepsy patients with respect to both seizure control and quality of life at one year follow-up. According to this study’s intention-to-treat analysis of seizure control outcomes, 58% (23/40) of the temporal lobe epilepsy patients randomized to the surgical group had no seizures-impairing-awareness at one year, compared to 8% (3/40) of patients randomized to the medical group (p< 0.001). When analysis of outcome was restricted to patients who actually underwent anterior temporal lobectomy rather than including all patients randomized to the surgical group, 64% (23/36) of anterior temporal lobectomy patients were found to be free of seizures-impairing-awareness at one year. As is standard of care, all patients undergoing surgery were continued on their antiepileptic drug regimen postoperatively (152).

Using the results of this randomized controlled clinical trial as their primary evidence, the American Academy of Neurology, the American Epilepsy Society, and the American Association of Neurological Surgeons published a practice parameter recommending that patients with medically intractable, disabling complex partial seizures who are appropriate candidates for anterior temporal lobectomy should be offered surgery. Their literature review also included less rigorous data: pooled results of anterior temporal lobectomy case series from 24 epilepsy centers, primarily retrospective studies with minimum one-year follow-up. The seizure control outcome for anterior temporal lobectomy derived from these combined results was similar to that of Wiebe’s randomized controlled clinical trial: 67% (1285/1952) of anterior temporal lobectomy patients attained freedom from disabling seizures postoperatively (44).

Seizure control after neocortical resections. For localized neocortical resections, the probability of excellent postoperative seizure control is not as high as after anterior temporal lobectomy. The Multicenter Study of Epilepsy Surgery, a prospective multicenter study assessing 396 patients undergoing resective epilepsy surgery found that among the 355 patients followed for at least one year, 56% of neocortical resection patients achieved a one year, seizure-free period postoperatively compared to 77% of anterior temporal lobectomy patients (p=0.01) (128). Similarly, in an analysis primarily of retrospective case series pooled from 8 epilepsy centers with minimum one-year follow-up, 50% (148/298) of patients who had undergone localized neocortical resections were free of disabling seizures. This pooled sample combined patients with epileptogenic zones in extratemporal and in lateral temporal neocortex and included both nonlesional and lesional epilepsy cases (44).

Seizure control after resection for lesional epilepsy. Resections for lesional epilepsy yield better seizure control outcomes than resections for nonlesional epilepsy (167; 123; 26; 55; 96; 44). Analysis of lesional cases alone revealed a higher postoperative seizure control rate, with 63% (83/131) of patients free of disabling seizures, based on results pooled from the case series of 5 epilepsy centers (44). With respect to specific types of lesional epilepsy, resections for epilepsy due to tumors and vascular malformations appear to yield better postoperative seizure control than resections for epilepsy due to trauma or to malformations of cortical development (06).

Seizure control after lesionectomy for epileptogenic cavernous malformations. Most studies of surgery for supratentorial cavernous malformations causing seizures have been retrospective analyses examining the seizure control outcomes of lesionectomy alone (28; 169; 21) or studies that group together lesionectomy and temporal lobectomy variants concomitant with lesionectomy of temporal cavernous malformations. For example, in their large, multicenter 2007 study, Baumann and colleagues did not differentiate between these 2 different operative strategies. In their retrospective analysis of 168 patients with a single epileptogenic supratentorial cavernous malformation, Baumann’s group assessed postoperative seizure outcomes after any type of resection. They found that 70% of resected patients were seizure-free (Engel Class I) at 1 year after operation and 68% were seizure-free at 2 years after operation (10). In a smaller series in which length of follow-up was not reported, Stefan’s group classified 53% (17/31) of resected patients as Engel Class I postoperatively (134). There are currently very few reports in the literature that address the question of dual pathology and therefore the possible requirement for concomitant or staged resection of mesiotemporal structures ipsilateral to temporal epileptogenic cavernous malformations (93; 118; 102; Upchurch et al unpublished). The published data at this time address the question of optimal management of epileptogenic cavernous malformations in general without specifying their anatomic location. The current evidence shows that surgery provides better seizure control than medical treatment for appropriately selected patients with epileptogenic cavernous malformations.

Specifically with regard to lesionectomy for epileptogenic cavernous malformations, "extended" lesionectomy has been shown in multiple studies to yield better postoperative seizure control than "restricted" lesionectomy (109; 11; 67). Extended lesionectomy consists of microsurgical resection of the cavernous malformation and any perilesional gliotic or hemosiderin-stained tissue that has an abnormal appearance on direct intraoperative visualization, whereas restricted lesionectomy is resection of the cavernous malformation alone. Comparing restricted lesionectomy to extended lesionectomy, Baumann and colleagues in 2006 investigated the effect of removal of the hemosiderin-stained brain tissue surrounding epileptogenic supratentorial cavernous malformations on postoperative seizure control outcomes in a well-designed retrospective study of 31 patients. In contrast to prior studies (169), they found postoperative seizure control better 3 years after lesionectomy procedures that included resection of surrounding hemosiderin tissue (Their study included 20 temporal cavernous malformations, of which 8 were mesiotemporal cavernous malformations for which resection of both the cavernous malformation and amygdalohippocampectomy were performed. However, the decision-making for the temporal cavernous malformation operations was not discussed because that was not the purpose of their study.). Other studies have yielded similar results regarding improved postoperative seizure control after extended lesionectomy of cavernous malformations (109; 67). The seizure control benefit of extended lesionectomy compared to restricted lesionectomy likely lies in the removal of the epileptogenic hemosiderin-laden tissue surrounding the lesion in the extended procedure.

Seizure control after multiple subpial transection. To evaluate the efficacy of multiple subpial transection both alone and with concomitant cortical resection, a metaanalysis combining multiple small case series assessed postoperative seizure control according to seizure type. Of 54 patients undergoing multiple subpial transection alone, 63% (12/19 patients) had a greater than 95% postoperative reduction in simple partial seizure frequency, 62% (13/21 patients) had such a reduction in complex partial seizure frequency, and 71% (10/14) in generalized seizure frequency (130).

Quality of life.

Quality of life after resective epilepsy surgery in general. Quality of life improves substantially for patients who are completely seizure-free after resective epilepsy surgery compared to those who still have seizures postoperatively. In the above-discussed prospective multicenter study of 396 patients undergoing resective epilepsy surgery, the Multicenter Study of Epilepsy Surgery, seizure-free patients had higher postoperative quality of life scores than patients who still had seizures on 3 measurements made over a 2-year follow-up period: the overall QOLIE-89 (p=0.01), the physical health domain (p=0.002), and the epilepsy-targeted domain (p=0.001) (128). An extension of this prospective study that examined the relationship between health-related quality of life (HRQOL) and postoperative seizure outcome longitudinally over time found that the quantified HRQOL increased initially for all patients within 6 months after surgery, regardless of seizure control outcome, and then subsequently increased as a function of square root of time seizure-free, generally reaching a plateau 2 years after operation if seizure control remained stable. Only patients who became seizure-free postoperatively sustained their improvements in HRQOL (129).

Patients’ perspectives on their epilepsy surgery were evaluated as an important aspect of postoperative quality of life in this same large prospective multicenter study of operative treatment (88% temporal and 12% extratemporal resections). At 2 years after operation, 79.1% of the study participants reported a positive overall impact of the surgery on their lives, and 75.5% stated that they would have the surgery again if they were given that decision to make over. Not unexpectedly, patients with postoperative seizure-freedom expressed more positive perceptions of their operative treatment than patients who still had seizures. In this study 47% of the patients had complete remission from seizures at 2 years after operation. Right-sided surgery (compared to left) was associated with more positive perceptions regarding life impact of resective surgery at 12 and 24 months after operation; as a corollary, analysis of nondominant resection showed a similar association with positive perceptions, but only at 24 months, not at 12 (24).

This study also highlighted employment gain as a crucial factor in patients’ positive perceptions of resective epilepsy surgery (24). Many patients cite the goal of improved employment as motivation for undergoing resective epilepsy surgery (160; 141). However, only modest gains in employment after resective epilepsy surgery have been demonstrated in retrospective studies and in a recent prospective study (04; 133; 25). In a prospective analysis from the Multicenter Study for Epilepsy Surgery group focused specifically on employment, one fourth of those unemployed before surgery were employed at 2-year follow-up (28/112, 25%) (25). The reality of these limited employment gains is an important consideration for the preoperative counseling of patients and consideration of epilepsy surgery earlier in the course of medication refractory epilepsy.

Quality of life after anterior temporal lobectomy. In Wiebe’s randomized controlled clinical trial comparing anterior temporal lobectomy versus antiepileptic drugs as treatments for temporal lobe epilepsy, the quality of life at one year for the surgical group was significantly better than that of the medical group. This assessment of the superiority of surgical treatment in improving patients’ quality of life was based on the QOLIE-89 mean global score of 73.8 for the surgical group versus 64.3 for the medical group (p< 0.001) (152). In the prospective Multicenter Study of Epilepsy Surgery, HRQOL assessment via the QOLIE-89 after temporal lobe resection for intractable temporal lobe epilepsy revealed that patients with good seizure control (Engel classes Ia and Ib) at 2 and 5 years after operation reported improved HRQOL even if they experienced postoperative memory decline (82).

Health care costs and evaluation of quality epilepsy care.

Health care costs. Outcome assessment has been addressed not only from the point of view of individual patients, but also from a public health perspective in analyses of the cost-effectiveness of resective epilepsy surgery. Retrospective studies focusing on anterior temporal lobectomy have found this procedure to be a cost-effective treatment for intractable temporal lobe epilepsy (70; 79). In a recent observational study, health care costs decreased substantially by 2 years after successful anterior temporal lobectomy, with success defined as postoperative freedom from seizures (80).

Evaluation of quality epilepsy care. Quality indicators for epilepsy care have been proposed, however, these proposed indicators have not yet included assessment of the selection of epilepsy patients for epilepsy surgery evaluation (108).

Conclusions regarding outcome data. Resective and disconnective epilepsy surgery is successful in improving seizure control and quality of life, particularly in the case of anterior temporal lobectomy for temporal lobe epilepsy. There is as yet no evidence from randomized controlled trials on which to base recommendations regarding neocortical resection or multiple subpial transection. However, the data available suggest that surgical treatment is beneficial for appropriately selected patients, particularly in comparison to the success rates of pharmacologic therapy (152; 44).

Adverse effects

Mortality. For resective epilepsy surgery, mortality rates have declined over time due to technologic progress in neurosurgery and neuroanesthesia (83; 106). In a large case series from a single epilepsy surgery center, 708 invasive procedures (429 therapeutic and 279 diagnostic procedures) performed in 429 consecutive patients resulted in no deaths (12). In pooled case series of 556 patients undergoing resective epilepsy surgery, there was no operative mortality; 2 deaths occurred within 1 month of the surgery but both were nonoperative (44).

Morbidity. The morbidity of resective epilepsy surgery is specific to the location of the resected epileptogenic zone, ie, to the anatomic structures removed. Among the most disabling of possible postoperative deficits are neuropsychologic morbidities, especially of language and memory. Such deficits have been most extensively studied in the surgical epilepsy population in which they are most common--temporal lobe epilepsy patients undergoing different types of anterior temporal resections (68; 110; 106).

Verbal memory deficits. With respect to material-specific memory deficits, verbal memory dysfunction has long been recognized as a complication of dominant hemisphere anterior temporal lobe resection for epilepsy (98). Declines in verbal memory performance were reported in 25% to 50% of 53 temporal lobe epilepsy patients after left anterior temporal lobectomy in one study (87). Similarly, 24% to 51% of 66 mesial temporal lobe epilepsy patients were found to have verbal memory deficits after left-sided selective amygdalohippocampectomy at 3-month follow-up (60), with persistence of deficits demonstrated at 1-year follow-up (61). Controversy remains regarding whether limited mesial temporal resection in the dominant hemisphere--eg, selective amygdalohippocampectomy rather than standard anterior temporal lobectomy--results in a higher probability of preserved verbal memory function, with some studies supporting the value of limited resection (27).

Counseling about operative risk. From the practical standpoint of counseling temporal lobe epilepsy patients about operative risks, those at high risk for postoperative verbal memory deficits include patients who undergo dominant hemisphere temporal resection, those with high preoperative verbal memory scores on neuropsychologic testing, and those whose Wada test shows better preoperative verbal memory function in the hemisphere ipsilateral to the epileptogenic zone compared to the contralateral hemisphere (114; 110; 139). In accord with Sabsevitz’s findings and supportive of the functional adequacy hippocampal model, Andelman and colleagues also found in their recent retrospective analysis a significant relationship between high preoperative Wada memory scores for the ipsilateral hemisphere and postsurgical deficits in verbal learning and memory (02).

Estimation of recurrence rate of seizures after resective epilepsy surgery varies depending on the type of epilepsy, the operative procedure, and the length of follow-up (164). In selected patients, reoperation may be the appropriate treatment for recurrent seizures (117).

Scientific basis

Epilepsy is a common chronic neurologic disorder consisting of recurrent, unprovoked seizures. It comprises a multitude of diseases and syndromes with heterogeneous causes. The recognition that certain epileptic syndromes are “surgically remediable” has improved patient selection for epilepsy surgery. Surgically remediable epileptic syndromes are conceptualized as “conditions with a known pathophysiology and natural history that have a poor prognosis with purely medical treatment, but that respond well to surgical treatment” (42).

Mesial temporal lobe epilepsy. Mesial temporal lobe epilepsy associated with hippocampal sclerosis is the most common and the most drug-resistant of the partial epilepsies (116; 136) and has been called the “prototype” surgically remediable syndrome (154; 42). The majority of resective epilepsy surgery procedures are performed for mesial temporal lobe epilepsy, and the scientific basis of curative epilepsy surgery has been most extensively investigated for this syndrome.

The epileptogenic zone. A primary goal of presurgical evaluation for resective epilepsy surgery has been the identification of “the epileptogenic zone”--defined as “the volume of brain tissue necessary and sufficient for the generation of seizures” (42; 112). In mesial temporal lobe epilepsy, this epileptogenic zone has been variably identified at different historical periods. One or other of the mesial temporal structures underlying mesial temporal lobe epilepsy--the dentate gyrus, hippocampus proper, entorhinal cortex, and the amygdala--has in turn been deemed the culpable epileptogenic structure at different institutions at different times (54). The purpose of the resective procedure has been conceptualized as removal of the epileptogenic anatomic structure, whether identified preoperatively or intraoperatively or by structural and functional abnormalities.

The epileptic network. The concept of the epileptogenic zone is crucial for the operative planning of epilepsy surgery. However, clinical and experimental studies of mesial temporal lobe epilepsy have generated substantial evidence to support the theory of an “epileptic network” underlying mesial temporal lobe epilepsy and other epilepsy syndromes (126; 18). For example, data from intracranial recordings show that the electrical onset of seizures may vary among the mesial temporal lobe structures from patient to patient and may differ in the same patient at different times (126; 148). The evidence for human epileptic networks from PET and SPECT clinical imaging studies is also extensive, particularly in revealing thalamic involvement in limbic epilepsy (126).

Interruption of the epileptic network. Changing the purpose of presurgical planning from identification of the epileptogenic zone to delineation of an epileptic network alters the conceptualization of surgical treatment for epilepsy syndromes. Instead of endeavoring to resect an ambiguous, ever-changing epileptogenic zone, the surgical goal is to interrupt the epileptogenic circuit or network. Different types of procedures--including resection, disconnection, and electrical stimulation--can disrupt the network (Electrical stimulation, such as vagus nerve stimulation, will not be discussed in this article). Resective procedures primarily consist of anterior temporal lobectomy variants and neocortical resections; whereas disconnective procedures include multiple subpial transection and corpus callosotomy.