Surgical Treatment of Seizures

(c) G. Rees Cosgrove, MD, F.R.C.S.(C) and Andrew J. Cole MD, FRCP(C) Departments of Neurology and Neurosurgery, Massachusetts General Hospital Epilepsy Center Harvard Medical School, Boston, Massachusetts.

Introduction

In the majority of patients with epilepsy, seizures can be well controlled with appropriate medication. However, current estimates indicate that 20 – 30% of patients with epilepsy are refractory to all forms of medical therapy. (1) These medically intractable patients are candidates for surgical treatment in an attempt to achieve better seizure control. Another group of patients who might benefit are those whose seizures may be relatively well controlled but who have certain characteristic presentations or lesions that strongly suggest surgical intervention might be curative. Overall, the single most important determinant of a successful surgical outcome is patient selection. This requires detailed presurgical evaluation to characterize seizure type, frequency, site of onset, psychosocial functioning and degree of disability in order to select the most appropriate treatment from a variety of surgical options. This type of evaluation is best carried out at a multi-disciplinary center experienced in the investigation and treatment of epilepsy.

In this article, we will outline the essential elements of the presurgical evaluation as well as the diagnostic surgical procedures required for invasive EEG monitoring and then describe the variety of therapeutic surgical options including indications, techniques, results and complications of each procedure.

Definitions

There are many types of seizures and different forms of epilepsy. A seizure is defined as a paroxysmal, self-limited change in behavior associated with excessive electrical discharge from the central nervous system. Epilepsy is defined as a condition of recurrent seizures and medical intractability as recurrent seizures despite optimal treatment under the direction of an experienced neurologist over a two to three year period.

In the past, seizures have been classified based upon their clinical manifestations, which had some relevance for patients and physicians but was of limited diagnostic or prognostic value. This classification scheme is based entirely on the distinct behavioral and electrophysiologic features of the seizures themselves and purposely avoids implications as to the specific underlying pathophysiological mechanisms or substrates. (2) According to this classification, an epileptic disorder is defined as either being generalized, partial (focal) or undetermined. Primary generalized seizures start as a disturbance in both hemispheres synchronously without evidence of a localized onset. The manifestations of these seizures tend to be major motor seizures of a tonic, clonic, tonic-clonic, myoclonic or atonic type. They also include minor events of the petit mal (absence) type. Partial forms of epilepsy start in a focal area of the brain and may remain localized without alteration of consciousness. These events are referred to as simple partial seizures and the symptoms vary with the area of the brain affected. If the event spreads and alters consciousness it is referred to as a complex partial seizure. If the event spreads further and leads to a major motor seizure it is referred to as a secondarily generalized seizure and may be quite difficult to distinguish from the primary generalized forms. Partial seizures often arise from the limbic structures of the temporal and frontal lobes but can occur from any cortical region and are often quite refractory to medical therapy alone. In general, patients with partial seizure disorders are the most amenable to surgical intervention.

A second classification divides the clinical epilepsies into idiopathic, symptomatic and syndromic forms based upon their presumed etiologies. (3) Idiopathic forms include some of the generalized seizure types that may have familial patterns but without a prominent genetic component such as simple febrile convulsions of childhood. Symptomatic forms are those caused by a recognized central nervous system (CNS) lesion. Included in this type are cases of known structural pathology, metabolic abnormalities or neurodegenerative processes. Syndromic forms include disorders that may be idiopathic or symptomatic but seem to follow a clear and predictable course. These include childhood and juvenile absence epilepsies, juvenile myoclonic epilepsy and the Lennox-Gastaut Syndrome. The ability to place a patient in one or another of the syndromic categories has the advantage of providing a basis for predicting long-term prognosis.

Both classifications have certain advantages and are not mutually exclusive but can be combined to provide helpful information. From a surgical point of view however, dividing the seizure types into either generalized or focal appears to be the most useful. This is because most surgical decisions are based upon defining those seizures that originate in one focal area of the brain and localizing that area as a prelude to resection.

Presurgical Evaluation

The goal of epilepsy surgery is to identify an abnormal area of cortex from which the seizures originate and remove it without causing any significant functional impairment. The primary components of the presurgical evaluation include a detailed clinical history and physical examination, advanced neuro-imaging, video-EEG monitoring, neuropsychological testing and assessment of psychosocial functioning. (Table 1) The major surgical questions one hopes to answer with this evaluation are: 1) are the seizures focal or generalized? 2) if focal, are they temporal or extratemporal in origin ? 3) is there a lesion associated with the seizures ?; and 4) if surgery is undertaken what functional deficits, if any, might be anticipated?

The presurgical evaluation of a patient with medically intractable epilepsy begins with a complete history and physical examination. One attempts to classify the different kinds of seizures as well as the frequency, severity and duration of each type. The clinical semiology of these events can yield important localizing information to the experienced clinician. It is also important to determine the age of onset, response to treatment and familial tendency to seizures. The pregnancy and delivery history is helpful in assessing congenital or early-acquired abnormalities. Other past medical history of significance would include a history of febrile seizures, head injury or intracranial infection. An assessment of the adequacy of medication trials must also be made to ensure that the patient is truly refractory to medical therapy.
On examination, the clinician looks for obvious asymmetries of development compatible with an early structural CNS lesion and focal neurologic or cognitive abnormalities suggestive of acquired disease. In the great majority of patients, however, the neurological examination is completely normal.
Modern neuroimaging is crucial to surgical decision-making. In the past, skull x-rays, ventriculograms, pneumoencephalography and computerized tomography (CT) scans demonstrated indirect evidence of cerebral pathology in the form of focal or diffuse atrophy or space-occupying lesions. Recently, magnetic resonance imaging (MRI) has replaced CT scanning as the imaging study of choice to evaluate patients with epilepsy. MRI is an extremely sensitive tool that can detect abnormalities of the brain with exceptional anatomical detail. (Figure 1) This has been especially true for detecting focal atrophy ( e.g. hippocampal atrophy), indolent gliomas, cortical dysplasias, cerebral gliosis and small structural lesions of the neocortex. (4) Functional imaging attempts to visualize alterations in cerebral metabolism using Positron Emission Tomography (PET) and Single Photon Emission Computerized Tomography (SPECT). These studies reveal epileptic areas as hypometabolic between seizures and hypermetabolic during seizures. (5,6) (Figure 2) Although they lack the spatial resolution of MRI, PET and SPECT can play an important role in the localization of abnormal cortex. Ictal SPECT studies can be obtained if injection of an appropriate radioisotope is performed within seconds of a seizure onset. The isotope is concentrated in the region of seizure onset and imaging studies can be obtained up to several hours after injection to demonstrate the area of ictal onset. These studies have been useful in many patients with occult epileptic foci. (7)
Electroencephalographic (EEG) investigation remains the most important aspect of the presurgical evaluation. Analysis of unselected EEG activity between events (interictal) or of specific activity during events (ictal) can provide evidence of focal electrical dysfunction. While certain interictal EEG abnormalities (spike and slow wave complexes) can be of localizing value, it is considered extremely important to record the EEG with concomitant videotape during the spontaneous occurrence of the patient’s events. Video/EEG monitoring can continuously record the EEG over a 24-hour period, which allows for careful inspection of the record during any symptomatic event. Sophisticated computer hardware and software also allows for automatic detection of spontaneous interictal epileptiform transients and electrographic seizures that otherwise might have gone unrecognized. (8) It is the EEG activity at the very beginning of the seizure before spread to adjacent areas that is most important in terms of localization and if a specific cortical area is involved consistently at the onset then that area is likely to be the site of seizure origin. Patients are often hospitalized with reduction in anti-seizure medications and may be recorded for up to 7-14 days in order to capture 3-5 of their habitual seizures.
Detailed neuropsychological testing is carried out to reveal specific focal or multifocal cognitive deficits that might be correlated with the neuroimaging and EEG. This testing may help in localizing an abnormal area of the brain but also serves as a comparison for post-surgical evaluation. An intracarotid amobarbital test is generally done as a prelude to surgery in order to lateralize language and memory function and to avoid neurocognitive deficits. (9)

Psychosocial Assessment

Psychosocial evaluation is also extremely important to assess current level of functioning and to ensure realistic goals and attitudes are engendered in both the patient and their family prior to surgery.

When a primary epileptogenic region or seizure focus is suspected but remains obscure despite appropriate neuro-imaging and scalp (non-invasive) video/EEG recordings, some form of implanted (invasive) electrodes may be indicated. Intracranial electrodes can be placed in areas not readily sampled by routine surface electrodes and can give more precise EEG information because of their proximity to discharging areas of the brain and the lack of movement/muscle artifact on the recordings. They have the disadvantage, however, of sampling from a relatively small area of cerebrum surrounding the contact points and the fact that they are accompanied by a surgical risk. They should only be undertaken after appropriate noninvasive monitoring has been completed so that an hypothesis of seizure onset has been formulated and a clear goal of the investigation has been defined. The diagnostic surgical options of implanted electrodes include epidural, subdural and intracerebral or depth electrodes.
Epidural electrodes are used infrequently and generally only for lateralization and approximate localization of seizure onset. (10) These electrodes are placed through tiny openings in the skull with the electrode contact resting on the dura to provide a high amplitude EEG signal without muscle or movement artifact. Because they do not penetrate the dura the risk of infection is minor. These electrodes can only record from the lateral convexity of the cerebral hemispheres and therefore are limited in their spatial resolution.
These electrodes are placed subdurally on the surface of the brain in the form of rectangular grids or linear strips with flat metal contact points mounted in flexible plastic. The grids require a craniotomy for placement and therefore are limited to unilateral application. (Figure 3) The strip electrodes can be placed through burr holes over the lateral convexity or under the frontal or temporal lobes. (11) It is difficult to place them in the interhemispheric fissure to record from parasaggital regions because of technical risks associated with large cortical veins. The major advantage of subdural electrodes is that they do not penetrate cerebral tissue and can record from a relatively wide area of the cortical surface. They can also be used for extraoperative cortical stimulation to map out specific areas of cortical function. Unfortunately, subdural electrodes cannot record directly from the deep cerebral structures (i.e. amygdala, hippocampus and cingulum), which are characteristically involved in many medically refractory partial epilepsies. They also have a small but real risk of intracranial infection and hemorrhage estimated to be approximately 4%. (12)
Intracerebral depth electrodes can be placed stereotactically into deep cerebral structures with the aid of CT, MR and angiography. Most centers employ flexible electrodes with multiple contact points that are placed through small holes in the skull and secured with some form of cranial fixation. (Figure 4) Electrodes are usually targeted towards the amygdala, hippocampus, orbital-frontal and cingulate regions and may be inserted via a lateral or vertex approach. Using a lateral approach, stereotactic cerebral angiography must be utilized to avoid major blood vessels during placement of the depth electrodes. Depth electrodes may be used in combination with scalp or subdural electrodes for more extensive coverage. Depth electrode investigation is generally indicated for patients with bitemporal, bifrontal of frontal temporal seizures and can localize a focal area of seizure onset not possible with scalp recordings. (13) The major complications of depth electrodes include hemorrhage and infection with mortality and morbidity rates between 1 – 4%. (12) It should be noted that the intracranial monitoring incurs greater risk than resective surgery itself and is also considerably more expensive than a noninvasive evaluation and therefore should be used only when necessary. With modern neuro-imaging, the use of invasive intracranial monitoring has declined from about 40-50% of patients in most centers to 10-20%.
If the information obtained during the noninvasive presurgical evaluation consistently points towards a single area of the brain as being the site of seizure onset, then the patient may be taken directly to surgery for resection of that area. If neuro-imaging demonstrates a well-characterized lesion (i.e. unilateral hippocampal atrophy, cavernous angioma, focal cortical dysplasia, etc.) and is supported by the clinical features of the seizures then surgery may be reasonable without the general requirement for ictal EEG data. However, if the data gathered from the clinical examination, imaging studies and noninvasive EEG evaluation are conflicting or disparities arise in the presumed localization of the seizure, then invasive intracranial monitoring is warranted. This is especially true in the extra-temporal epilepsies where EEG localization is notoriously difficult. If a localized area of seizure onset is confirmed then these patients too can undergo resective surgery.

Therapeutic Surgical Options

Epilepsy surgery began as removal of gross structural lesions of the brain. With the addition of EEG data from preoperative and intraoperative recordings, areas of removal expanded to include tissue that was grossly normal in appearance but known to give rise to epileptiform activity. Small areas of resection were soon replaced by partial lobectomies and more extensive cortical resection. While resection techniques (lesionectomy, lobectomy, hemispherectomy, corticectomy) generally yield the best surgical results, disconnection (callosotomy, subpial transection) and augmentation (cerebellar and vagal stimulation) techniques remain worthwhile considerations (Table 2).

The primary objective of most epilepsy surgical procedures is to accurately localize and then completely excise the epileptogenic region without causing cognitive or neurologic deficit. An important determinant of the risk of surgery is the relationship of the lesion to functionally important or “eloquent” brain regions because injury to these “eloquent” areas can cause irreversible neurologic impairment. The location of many functionally important areas can be approximated using anatomic landmarks but individual variations occur and the presence of local pathology can distort landmarks making localization imprecise. Regions responsible for seizure onset must be distinguished from regions of critical cortical function and a variety of strategies have therefore been employed both pre- and intra-operatively to optimize surgical resection while minimizing risk of injury to functional cortex.

Some centers utilize intraoperative cortical recordings to sample EEG activity from the cerebral surface and to allow for cortical mapping. Classical cortical mapping requires a craniotomy under local or light general anaesthesia and direct electrical stimulation of the cortex using a hand-held stimulator. (14) This allows for precise individual localization of sensory, motor and language areas but unfortunately, the information cannot be used preoperatively for risk assessment, therapeutic decision-making and surgical planning. Centers that use subdural grid electrodes may carry out functional mapping extraoperatively, in advance of the cortical excision, by passing small currents between implanted electrodes. (15) This lacks the spatial precision of intraoperative stimulation but can be very useful especially in children or uncooperative adults. Localization of the rolandic sulcus may also be carried out by recording somatosensory evoked potentials and the recognition of their phase reversal over the sulks. (16)
The newest method of localizing cortical function is with functional MRI. This powerful neuroimaging technique can create an anatomical and functional model of an individual patient’s brain. Rapid echoplanar imaging performed while the patient engages in a specific task (i.e. fist clenching, tongue movement, verb generation) detects small changes in signal intensity related to changes in cerebral blood flow. (17) Intensive computerized image processing can then define the areas of cortex activated by the specific task. Concurrent 3-dimensional rendering of cerebral topography, cortical veins and related pathology gives an unprecedented display of critical relational anatomy. By combining detailed anatomical information with precise physiological information, fMRI is capable of creating a structural and functional model of an individual’s brain. (Figure 5) It is likely that fMRI will play an increasing role in the presurgical evaluation of epilepsy patients.

Other non-invasive cerebral mapping techniques that have evolved to localize functionally important cortical areas are magneto-encephalography (MEG) and positron emission tomography (PET). Both can localize certain cortical functions non-invasively but require dedicated units that are not widely available.

After the resection strategy is decided upon, tissue removal is carried out using subpial resection techniques. Cortical gray and white matter is carefully removed by suction or cavitron so that the pia remains intact over the adjacent gyri. This tends to form a nonscarring barrier and preserves blood supply to the remaining cortex as well. Following removal, some centers carry out post-resection cortical EEG recordings and may carry out further removal if considerable epileptic activity remains at the resection margins.

With the advent of MRI, many patients with recurrent seizures are now discovered to have small, previously unrecognized lesions such as cavernous angiomas, low-grade astrocytomas, cortical dysplasias and areas of focal atrophy that are clearly the cause of their seizures. In general, if these are located in extratemporal sites, removal of the lesion and a small rim of surrounding cortex is often successful in controlling seizures. Removal of significant perilesional cortex may be necessary to achieve optimal seizure control in some patients. In many instances, if only a portion of the lesion is removed, the surgical result is suboptimal. If lesions are located within the temporal lobe, lesionectomy along with temporal lobectomy is carried out including the mesial temporal structures in order to yield good results in 80% of cases. (18) Overall, lesionectomy is associated with excellent results with success rates that are generally better than with surgery performed in patients without discrete lesions.
The majority of resections involve the temporal lobe and initially consisted of the classical anterior temporal lobectomy. This was either carried out “en-bloc” under general anesthesia or using a more tailored resection with electrocorticography and cortical mapping under local anesthesia. The majority of temporal lobectomies, whether in the dominant or nondominant hemisphere, can now be safely performed under general anesthesia with or without electrocorticography. In the dominant hemisphere, temporal lobe removals usually extend back 4.5 – 5 cm. behind the temporal tip or to the level of the central sulcus. In the non-dominant hemisphere, temporal lobectomies can extend beyond 7 or 8 cm but will result in a contralateral superior quadrantanopsia because of encroachment upon the optic radiation. It is important that the mesial temporal structures are included in the removal because most neurosurgeons believe that the hippocampus is intimately involved in seizure propagation or amplification. Studies also indicate that recurrent seizures are more likely following temporal lobectomy when the hippocampus is not removed. (19)

Since almost 80% of temporal lobe seizures originate in the mesial structures, several operative approaches have been designed to reduce the amount of temporal neocortex removed but still resect the amygdala and hippocampus. The so-called antero-medial temporal lobectomy with amygdalo-hippocampectomy is a modification of the classical temporal lobectomy by reducing the amount of cortical removal and extending the hippocampal resection. (20) (Figure 6) Selective amgdalo-hippocampectomy removes the mesial temporal structures via either trans-sylvian, transcortical or trans-sulcal microsurgical approach with the goal of sparing temporal neocortex and reducing any possible neuropsychological deficits. (21) (Figure 7) Some cortical injury and white matter disruption does occur with this technique and it is only applicable to patients with clear-cut mesiobasal temporal lobe epilepsy. No matter which procedure is advocated, if patient selection is appropriate, surgery in the temporal lobe offers good to excellent results in 75 – 85% of the cases. With modern imaging techniques, seizure free rates are now approaching 90% with febrile seizures, hippocampal atrophy and mesial temporal sclerosis being positive predictors of a good outcome.
Morbidity and mortality figures for cortical excisions are quite low, less than 0.2% in one large series with over 2000 patients. (22) The incidence of hemiparesis was 0% following temporal lobectomies and 0.5% for hemiparesis and/or dysphasia following frontal lobectomies. (23) In dominant hemisphere removals, however, there is often a temporary speech deficit. Specific cognitive testing may detect permanent subtle deficits consistent with the site of removal but generally these are nonspecific. (24) An upper quadrantanopsia may occur with larger temporal removals in the nondominant hemisphere. This may be acceptable if required for seizure control since it is usually unnoticed by the patient and does not interfere with normal daily living. Memory impairment has occurred with unilateral temporal removals in rare cases but this complication is avoided by preoperative testing of speech and memory function during the intracarotid amytal test. If memory is affected by amytal injection ipsilateral to the proposed side of the temporal removal, temporal excision may be designed to spare the hippocampus and medial structures but this approaach may reduce operative success rates.

Extra-temporal resections are much less commonly performed with the majority being carried out in the frontal lobe. En bloc standardized resections are not generally suitable and most surgeons guide their resections by detailed electrocorticography, both intra- and extra-operatively along with detailed cortical mapping. Frontal resections range from localized topectomies to complete frontal lobectomies and must be carefully individualized. Identification of the primary motor cortex is essential to avoid motor deficits and anterior language cortex to avoid speech difficulties. Parietal and occipital resections are rarely carried out but may be gratifying in patients with clear structural lesions.
The results of cortical excision for extratemporal epilepsy are variable depending upon patient selection and method of presurgical evaluation. Outcome statistics are not as impressive for extra- temporal resections as they are for temporal removals. Nevertheless, extra-temporal resections including the frontoparietal and occipital regions can give excellent results. Patients with epileptic discharge limited to the lobe of resection obviously tend to do better than those with more widespread discharges. In addition, some patients have more wide spread epileptogenic zones that require multilobar resections. In the largest cumulative series 64% of patients were improved, 36% being seizure free. (23) With advances in neuro-imaging and other aspects of the presurgical evaluation, it is hoped that surgical success rates can improve in the future.
Hemispherectomy is another form of cortical excision that is limited to patients with congenital hemiplegia, chronic encephalitis, hemi-megalencephaly or Sturge-Weber syndrome. These patients tend to have severe epilepsy with wide spread independent epileptic discharges that often extend to the contralateral (normal) hemisphere. It is only performed on patients who have a dense hemianopsia and are already hemiplegic with no fine motor activity on the affected side. The acute surgical risk is that some crude movement or sensation on the opposite side of the body would be adversely affected. A chronic complication was recognized to occur approximately 8 to 10 years after gross total hemispherectomy. This condition called superficial cerebral hemosiderosis resulted from chronic leaking of blood into the resection cavity producing recurrent seizures, sensori-neural deafness and hydrocephalus. It occurred in approximately 25% of patients by ten years and mandated a modification of the procedure. (25) This complication is now avoided by performing an anatomically subtotal but functionally complete hemispherectomy in which the frontal and occipital poles are left in place with their blood supply but all neural connections are transected. Residual cerebral tissue either decreases the risk of hemorrhage into the resection cavity or alternatively absorbs any blood that might leak in. Alternatives to anatomical hemispherectomy include hemispherotomy, cerebral hemicorticectomy, dural plication and ventriculoperitoneal shunting. All of these modifications attempt to reduce the risk of superficial cerebral hemosiderosis by minimizing cortical resection while maintaining complete functional disconnection.
Functional hemispherectomy or any of its variants is one of the most successful surgical procedures for epilepsy with over 85% markedly improved and about 60% seizure free. (25) Many patients also demonstrate behavioral improvement probably on the basis of a better attention span and cognitive functioning.
Corpus callosotomy has been offered as an alternative to hemispherectomy in epileptic patients with congenital hemiplegia but the results are not as good as with hemispherectomy. It is indicated when the patient has a severely damaged hemisphere but motor, sensory or visual function that would be valuable to preserve. In general, however, corpus callosotomy ismost useful for those patients with generalized seizure disorders and bilateral independent epileptic areas in the frontal region. The seizures that respond best to callosotomy are sudden falls or “drop attacks” with injury to the patient. Some patients with additional focal seizures may experience an improvement or overall reduction in these partial seizures but about 20% of patients will have an increase in the number of focal seizures. The generalized seizures and drop attacks tend to improve markedly although a complete cure of seizures is extremely rare. (26) Early surgical experience included deaths and severe morbidity but the risks have become extremely low with modern microsurgical techniques. The current practice is to section the anterior 2/3 of the corpus callosum on the first procedure. The posterior 1/3 may be sectioned at a second procedure if the results of anterior section are not satisfactory. Transient abulia is common following anterior callosotomy but other disconnection effects are fortunately mild and uncommon. In patients with complete callosotomy, disconnection symptoms are more frequent. There is often some difficulty in bimanual tasks and apraxia for commands directed to the nondominant extremity. Visual presentation to the hemifield opposite to the dominant hemisphere cannot be comprehended or described by language modalities and there is often significant difficulty writing with a nondominant hand. Fortunately, most of these functional deficits are not noticeable in normal daily living and are balanced by the improved seizure control.
In patients with seizure onset or epileptic zones located in eloquent cortex, multiple vertical subpial transections have been recommended as an alternative to cortical resection. This technique leaves the vertical columnar arrangement of the cortex intact thereby preserving function but prevents spreading of the seizure discharge in the horizontal plane to reduce seizures. Some neurological deficits appear postoperatively but these generally resolve over several weeks with satisfactory improvement in seizure control in 70 % of patients. (27) Experience with this technique is still rather limited but it does provide a surgical option in patients with seizures arising in cortex that has been previously considered inoperable.
Stereotactic lesions of deep cerebral structures have been carried out for a variety of generalized and focal forms of epilepsy in the past. Bilateral cingulotomies, amygdalotomies, lesions in the Field of Forel and thalamic lesions have all been tried. (28) Results are scattered and too few for any conclusions to be made although generally they are unimpressive. While some lesions may have an initial good result, seizures tend to recur in virtually all patients and stereotactic ablations of subcortical structures are no longer in use.
Cerebellar electrical stimulation has been used to treat generalized focal and myoclonic seizures as well as for spasticity of cerebral palsy. (29) Cerebellar stimulation has a theoretical basis from animal studies in which lesion induced cortical discharges were reduced or inhibited by cerebellar electrical stimulation. Initial reports of clinical success could not be reproduced. Reports followed of tissue damage from the cerebellar stimulator and a large number of late failures. Improvement in the technical quality of the electrodes and stimulating devices has led to some renewed interest in this technique but it is not currently a recommended treatment and no definitive evidence supporting its use in controlling epilepsy is available.
More recently, a number of patients with both focal and generalized intractable seizures have undergone implantation of a nerve stimulator around the left vagus nerve. Less than half experienced a 50% reduction in seizure frequency and only the rare patient became seizure free. (30)

Conclusions

The success or failure of the surgical treatment of epilepsy depends in large part on the proper selection and investigation of patients. Recent advances in imaging and long term EEG monitoring have allowed for a greater accuracy in the localization of the seizure focus with overall surgical results better than those of the prior decades. Continued investigation into the basic mechanisms of the epilepsies as well new forms of medical and surgical therapy is necessary in order to help the many patients with severe and disabling intractable seizures.

References

1. Robb, P: Focal epilepsy: the problem, prevalence, and contributing factors. In Advances in Neurology, eds. D. Purpura, J. Penry, R.D. Walter, pp. 11-22. New York: Raven Press, 1975.

2. Commission on Classification and Terminology of the International League Against Epilepsy: Proposal for Revised Clinical and Electro-graphic Classification of Epileptic Seizures. Epilepsia 26:258-268, 1985.

3. Commission on Classification and Terminology of the International League Against Epilepsy: Proposal for Classification of Epilepsies and Epileptic Syndromes. Epilepsia 26:268-278, 1985.

4. Lessor R, Modic M, Weinstein M, et al. MRI in patients with intractable epilepsy, Arch Neurol 43:367,371,1986.
5. Engel, J., Kuhl, N., Phelps, M., Crandall, P. Comparative localization of epileptic foci in partial epilepsy by PET and EEG. Ann Neurol 12:529-37, 1982.

6. Lee, B., Marklan, O., Siddiqui, A., Park, H., Mack, B., et al. Single photon emission computed tomography (SPECT) brain imaging, intractable complex partial seizures. Neurology 36:1471-77, 1986.

7. Marks DA, Katz A Hoffer P, et al. Localization of extratemporal epileptic foci during ictal single photon emission computed tomography. Ann Neurol 31:250-255,1992.

8. Gotman, J., Ives, J., Gloor, P., eds. Long-Term Monitoring in Epilepsy (Suppl. 37 to Electroencephalography and Clinical Neurophysiology). Amsterdam: Elsevier, 1985.

9. Wada J, Rasmussen T: Intracarotid injection of sodium amobarbital for the lateralization of speech dominance; experimental and clinical observations. J Neurosurg 17:226-282.1960.

10. Barnett GH, Burgess RC, Awad IA, et al: Epidural peg electrodes for the presurgical evaluation of intractable epilepsy. Neurosurgery 27:113-115,1990.

11. Wyler, A.R., Ojemann, G.A., Lettich, E., Ward, A.A: Subdural strip electrodes for localizing epileptogenic foci. J. Neurosurg. 60:1195-1200,1984.

12. Van Buren JM: Complications of surgical procedures in the diagnosis and treatment of epilepsy. In: Engel J Jr, (ed), Surgical Treatment of the Epilepsies. New York: Raven Press, 1987:465-475.

13. Spencer, S. Depth electroencephalography in selection of refractory epilepsy for surgery. Ann. Neurol. 9:207-14, 1981.

14. Ojemann, G.A. Surgical therapy for medically intractable epilepsy. J. Neurosurg. 66-489-909, 1987.
15. Lessor RP, Luders H, Klem G, et al: Extraoperative cortical functional localization in patients with epilepsy. J Clin Neurophysiol 4:27-53,1987.

16. Dinner DS, Luders H, Lessor RP, et al: Invasive methods of somatosensory evoked potential monitoring. J Clin Neurophysiol 3:113-130,1986.

17. Cosgrove GR, Buchbinder BR, Jiang H: Functional magnetic resonance imaging for intracranial navigation in Neurosurgical Clinics of North America. Maciunas R (ed), WB Saunders and Co:Philadelphia, 1995.

18. Spencer, D., Spencer, S., Mattson, R., Williamson, P. Intracerebral masses in patients with intractable partial epilepsy. Neurology 34(4):432-36, 1984.

19. Rasmussen, T. Surgical treatment of complex partial seizures: results, lessons and problems. Epilepsia Suppl. 1 24:65S-76S, 1983.

20. Spencer DD, Spencer SS, Mattson RH, et al: Access to the posterior medial temporal lobe structure in surgical treatment of temporal lobe epilepsy. Neurosurgery 15:667-671,1984.

21. Weiser, H.G., Yasargil, M.G. Selective amygdalohippocampectomy as a surgical treatment of mesiobasal limbic epilepsy. Surg. Neurol. 17:455-57, 1982.

22. Rasmussen, T. Cortical resection in the treatment of focal epilepsy. In Neurosurgical Management of the Epilepsies. Adv. in Neurol. 8:139-54, 1975.

23. Rasmussen, T. Surgery of frontal lobe epilepsy. In Neurosurgical Management of the Epilepsies. Adv. in Neurol. 8:197-205, 1975.

24. Lieb, J.P., Rausch, R., Engel, J., Brown, W.J., Crandall, P.H. Changes in intelligence following temporal lobectomy: relationship to EEG activity, seizure relief and pathology. Epilepsia 23:1-13, 1982.

25. Rasmussen, T. Hemispherectomy for seizures revisited. Can. J. Neurol. Sci. 10:71-78, 1983.

26. Gates, J.R., Rosenfeld, W.E., Maxwell, R.E., Lyons, R.E. Response of multiple seizure types to corpus callosum section. Epilepsia 28:28-34, 1975.

27. Morrel F, Whisler WW, Bleck TP: Multiple subpial transections: a new approach to the surgical treatment of epilepsy. J Neurosurg 70:231-239.1989.

28. Ojemann, G.A., Ward, A.A. Stereotaxic and other procedures for epilepsy. In Neurosurgical Management of Epilepsy, Adv. in Neurol. 8:241-65, 1975.

29. Cooper, I.S., Amin, I., Riklan, M., Waltz, J.M., Poor, T.P. Chronic cerebellar stimulation in epilepsy. Arch. Neurol. 33:559-70, 1976.

30. Ramsey RE, Uthman B, Ben-Menachem E, et al: Efficacy of vagal nerve stimulation in partial seizures: double blind comparison of two stimulus paradigms. Epilepsia (suppl) 32:90-91,1991.