Intractable disabling epilepsy, a localizable epileptogenic zone (defined as the area of tissue that must be removed to prevent further seizures), and a low risk of postoperative deficits are considered the three basic tenets that must be met before a patient can be considered a candidate for epilepsy surgery.² It is important to recognize that medical intractability makes a patient a candidate for further investigation regarding the appropriateness to undergo surgery for epilepsy, but it does not guarantee that surgical management should ultimately be offered. Although no clear definition of medical intractability exists, recent reports in the literature suggest that successful treatment with anticonvulsant therapy is unlikely once two different medicines fail to control seizures.³ The optimal surgical patient will be highly motivated with a social support system in place, and additional coexisting psychiatric illness will be well controlled. Neurocognitive risks must be assessed and discussed with each patient, and risks should be deemed acceptable from both the perspective of the patient and the treatment team. A high probability of improving a patient’s overall quality of life must be confirmed before undertaking the procedure.⁴

Preoperative Evaluation

The starting point for any patient evaluation should include a detailed history and physical examination, with special focus on seizure semiology. Lateralizing and localizing signs, such as ictal eye movements, motor or sensory features, presence or absence of ictal speech, and the presence or absence of aura, should be noted. Video-electroencephalography (VEEG) data allow correlation of a patient’s behavior during a seizure with the changes recorded on scalp EEG. It allows for confirmation of the epilepsy and may provide helpful lateralization/localization to aid in the surgical hypothesis.

MRI studies for patients with suspected extratemporal lesions should include axial T2 sequences, axial and coronal fluid-attenuated inversion recovery (FLAIR) sequences, a sagittal T1 sequence, and a high-resolution (thin-cut) coronal magnetization prepared rapid gradient echo (MPRAGE) sequence.⁵ Although 1.5-tesla magnet strength is considered acceptable, 3-tesla or higher machines and the use of surface coils over the region of interest may add to the diagnostic yield. Additional studies such as subtraction single-photon emission computed tomography (SPECT), [¹⁸F]-fluorodeoxyglucose (FDG), positron emission tomography (PET), or magnetoencephalography (MEG) provide complementary information and may help to formulate the surgical hypothesis (Fig. 52.2). These studies may also be used as an adjunct in patients with multilobar lesions or discordant data.⁶ Functional MRI (fMRI), MEG, or intracarotid sodium amytal testing may be performed for localization of language and sensorimotor cortex (Fig. 52.3). After review of all available data, a likely cause should be determined for the patient’s seizures; an underlying cause has been correlated with the probability of postoperative seizure freedom across a number of surgical case series.^7,⁸

FIGURE 52.2 Example of left temporoparietal and bitemporal hypometabolism seen on [¹⁸F]-fluorodeoxyglucose (FDG) positron emission tomography (PET) scanning.

FIGURE 52.3 Left hemispheric language dominance documented by word generation, listening, and rhyming tasks during functional magnetic resonance imaging (fMRI).

The goal of the presurgical evaluation is the creation of an individualized surgical hypothesis. This is best done by a multidisciplinary team skilled in the treatment of epilepsy. All the data gathered should be presented to the epileptologist, neurosurgeon, neuropsychologist, psychiatrist, social worker, and neuroradiologist on the team. Initially, each available diagnostic study is analyzed independently of the other investigative modalities. No single study can measure the epileptogenic zone and so all data must be considered together with the video-EEG data, seizure semiology, and MRI, which is perhaps the most important. Adjuvant testing such as PET, SPECT, MEG, neuropsychological testing, and fMRI help to round out the surgical strategy.

The strongest hypothesis is formed when the localization predicted by individual studies is concordant. Any discrepancies must be evaluated and resolved in relation to surgical planning. Discrepancies between semiology, EEG, or MRI are considered the most troubling. If localization differs based on these studies, or if the MRI is nonlesional, then potentially clarifying information should be sought with either FDG-PET or ictal SPECT. When discrepancies cannot be resolved, additional methods of strengthening the working surgical hypothesis should be considered. Chronic extraoperative intracranial recordings may provide critical information in such patients (Fig. 52.4). These may also be used when noninvasive evaluation leads to a hypothesis involving a large cortical area and further delineation of the epileptogenic zone is desired. A working hypothesis of the ictal onset zone based on noninvasive data helps guide the placement of the intracranial electrodes in an effort to confirm it prior to surgical resection.

FIGURE 52.4 A, Example of subdural grid electrode insertion over left hemisphere. B, Sagittal magnetic resonance imaging (MRI) reconstruction with electrode colocalization. C, Anteroposterior (AP) and lateral skull film localization of subdural electrodes.

Invasive intracranial electrodes also have a distinct advantage in mapping functional cortex. Consequently, their placement should be considered when epilepsy is thought to arise from within or near eloquent cortex. They are especially useful for localizing functional cortex in those patients unable to tolerate awake craniotomy (Fig. 52.5).

FIGURE 52.5 Example of cortical stimulation map displaying stimulation identified language and face motor sites.

It is also not uncommon that after a full noninvasive presurgical evaluation, the concordance of data results in a hypothesis that would preclude the patient from undergoing surgical treatment. Even though the patient may still have intractable disabling epilepsy, there is no longer a localizable epileptogenic zone, or the localized epileptogenic zone may confer a high risk of unacceptable postoperative deficit. This possibility should be introduced and discussed with the patient at the onset, not the conclusion, of the surgical evaluation.

Surgical Procedures

There is a broad range of surgical techniques for the treatment of extratemporal epilepsy. These procedures can be loosely divided into resective techniques and palliative techniques.⁹ Resective techniques include lesionectomy, tailored focal resection (a “lesionectomy plus” approach), lobectomy, multilobar resections, and hemispherectomy and its variants. Palliative techniques include corpus callosotomy, multiple subpial transections, and vagal nerve stimulation. Other stimulation techniques currently under investigation include thalamic targeted deep brain stimulation, responsive cortical stimulation, and transcranial stimulation. Intraoperative techniques include stereotactic neuronavigation, ultrasound, intraoperative neuromonitoring with somatosensory evoked potentials (SSEPs) or electrocorticography (ECoG), and intraoperative MRI.

The surgical approach depends not only on surgical and functional anatomy, but also on age of the patient, medical comorbid conditions, and anticipated underlying disease. In particular, certain types of focal cortical dysplasias, neoplasms, and cavernomas can often be resected with pure lesionectomy, typically resulting in good surgical outcome (Fig. 52.6).¹⁰^–¹² The general tenet is focal resections for focal disease, and progressively larger areas of resection for more diffuse or for multifocal disease. The extension of a resection beyond the area of visible lesion may be performed with the assistance of intraoperative ECoG, SSEPs, or extraoperative recording with the use of implanted intracranial electrodes. The use of such techniques may be more important in cases of cortical dysplasia where the epileptogenic zone often extends beyond the margins of the radiographic lesion.

FIGURE 52.6 A, Fluid-attenuated inversion recovery (FLAIR) axial magnetic resonance imaging (MRI) demonstrating balloon cell dysplasia in left perirolandic region. B, Intraoperative photograph demonstrating same balloon cell dysplasia (encircled by suture). Ictal onset localized on grid recordings to the borders of the lesion. C, Photograph taken after lesionectomy plus ictal onset zone in adjacent cortex

Hemispherectomy, and related procedures, tend to be limited primarily to the pediatric population. Anatomical hemispherectomy and disconnective hemispherectomy procedures are the available techniques.^7,¹³ To be considered for hemispherectomy, the ideal patient should have epilepsy arising from a single affected hemisphere. The underlying disease should be unilateral and most patients considered for these procedures will have preexisting neurological deficits, including homonymous hemianopia and contralateral hemiparesis with absence of functional fine motor movements of the hand. If they do not, counseling regarding these anticipated deficits should be extensive, including the likelihood that the patient will not drive due to the visual field defect.

Anatomical hemispherectomy remains in use for selected pathological processes such as hemimegalencephaly and diffuse malformations of cortical development. It is also commonly utilized when patients have failed prior disconnective hemispherectomy techniques.¹⁴ Disconnective hemispherectomy techniques are usually considered in those patients with perinatal stroke, Sturge-Weber syndrome, or Rasmussen’s encephalitis.⁷ The goal of these less extensive procedures is isolation of the hemispheric cortex while reducing the perioperative risks that are more prevalent with the anatomical technique (blood loss and hydrocephalus). From a practical standpoint, this translates to disconnection of the horizontal frontal fibers, mesial temporal structures, corpus callosum, and the internal capsule and corona radiata (see video).¹⁵ The completeness of disconnection, in addition to seizure etiology, relates directly to surgical outcome.¹⁶

Functional hemispherectomy as described by Rasmussen implies removal of the temporal lobe (including mesial structures) followed by a limited central resection of frontoparietal tissue.¹⁷ This limited resection allows access to the lateral ventricle, facilitating the disconnection of the remaining hemisphere through the ventricular space. The majority of the disconnected hemisphere is then left in situ in an attempt to reduce perioperative morbidity (blood loss, hydrocephalus) and the long-term risk of superficial cerebral hemosiderosis.

The Rasmussen technique, initially reported in 1973, has since undergone a number of modifications. The resulting evolution has yielded both the hemispherotomy and hemispherical deafferentation techniques.^18,¹⁹ The main variation among these procedures is the amount of tissue resected as the surgeon enters the ventricular space, with more recent refinements of technique focusing on reducing the amount of tissue resected. The smaller resection window of hemispherotomy and hemispherical deafferentation is ideally suited for those patients with anatomical variations that make the process of disconnection straightforward, such as large ventricles or encephalomalacia.

Each hemispherectomy technique has unique advantages and disadvantages, with selection of technique individualized to the patient’s anatomy and underlying disease. For those patients with malformations of cortical development including hemimegalencephaly, anatomical hemispherectomy is the most direct way to completely remove the malformed tissue. These patients often have small ventricles and no clearly defined gray/white interface, producing a situation that makes complete disconnection via a less invasive hemispherectomy technique almost impossible. The disadvantage to the anatomical technique is a slightly higher risk of perioperative morbidity and a higher incidence of postoperative hydrocephalus. The theoretical risk of superficial hemosiderosis is often quoted when discussing anatomical hemispherectomy, but this complication has not been reported in the neurosurgical literature since the 1960s.

The functional technique is well suited to those patients with Rasmussen’s encephalitis where the normal gray/white anatomy is preserved. Removal of the temporal lobe and peri-insular frontoparietal operculum facilitates removal of the insula and may lead to a better outcome in this unique group of patients. Finally, those patients afflicted with perinatal vascular events are ideal candidates for peri-insular hemispherotomy as the surgeon can utilize the encephalomalacia to access the ventricle and disconnect the hemisphere. Advantages to these less invasive techniques include reduced blood loss and risk of hydrocephalus. The one real disadvantage is the higher reoperation rate when seizures persist postoperatively and there is cortical tissue left in place.

Surgical Technique: Anatomical Hemispherectomy

The patient is positioned either on a rigid fixation device or resting on a head support. The head is turned 90 degrees and elevated above the level of the heart to assist with venous return. An ipsilateral shoulder support is placed. A T incision is then marked such that it consists of one line at least 0.5 cm from midline and another perpendicular line extending from the zygomatic root just anterior to the tragus. The midline incision should extend from hairline to 4 to 5 cm above the inion (Fig. 52.7). The incision is infiltrated with a local anesthetic and the skin is incised, taking care not to injure the sagittal sinus in the area of an open anterior fontanelle. Careful attention is paid to hemostasis throughout the procedure, with all bleeding points controlled with bipolar coagulation and hemostatic agents. The temporalis muscle is incised and reflected inferiorly to allow adequate exposure of the temporal lobe. Bur holes are then placed just above the zygomatic arch, the keyhole, and along the parasagittal area to allow an exposure of the entire hemisphere. High-speed air craniotome is then used to remove the craniotomy flap. Dural tack-up sutures are then placed and the dura is opened in an H-fashion. The surface of the cortex is examined, the sylvian fissure identified, and venous drainage patterns noted (see Fig. 52.7). Major draining veins to the sagittal sinus are carefully protected to avoid potentially devastating hemorrhage. Initial dissection is performed to open the sylvian fissure (see Fig. 52.7). The middle cerebral artery (MCA) is exposed at a point just distal to the lenticulostriate branches. The inferior and superior circular sulci of the insula are then exposed (see Fig. 52.7) and the MCA is ligated with bipolar cautery and surgical clips, if necessary.

FIGURE 52.7 A, The T incision for exposure of the cerebral hemisphere. B, The right cerebral hemisphere exposure needed for anatomical removal. C, The right sylvian fissure is opened and the circular sulci exposed (not shown). D, An example of a “split sylvian fissure” with exposure of the middle cerebral artery (MCA) vessels and circular sulci. E, Surgeon’s line drawing demonstrating ligation of anterior circulation branches above the corpus callosum. F, Postoperative anatomical hemispherectomy cavity.

Infrasylvian dissection and ventricular access is then performed. Disconnection of the temporoparietal white matter proceeds through the inferior circular sulcus, facilitating access to the temporal horn of the lateral ventricle. The ventricle is opened anteriorly to the amygdala and then posteriorly so that the temporal horn is exposed from its anterior aspect to the trigonal region. The lateral ventricular sulcus of the temporal horn is then opened and the collateral sulcus is followed to the tentorial edge and mesial temporal pia. The parahippocampal gyrus is aspirated, exposing the tentorial edge, and the pia is coagulated and divided from the temporal pole to the trigonal region. The posterior cerebral artery branches are encountered and ligated as they course over the tentorial edge on their way to the temporo-occipital cortex.

Attention is then turned to the suprasylvian dissection. The superior circular sulcus is opened to divide the corona radiata and expose the lateral ventricle longitudinally. The ipsilateral foramen of Monro is occluded with a cotton ball to prevent blood entering the dependent hemisphere. Both the choroid and basal ganglia are protected in order to minimize bleeding. Hemostatic agents applied to the surface of the basal ganglia are effective in controlling surface bleeding. The frontoparietal white matter is aspirated to open the entire ventricular system, connecting to the posterior temporal disconnection from the infrasylvian dissection. The middle cerebral artery branches in the posterior aspect of the sylvian fissure are coagulated and divided.

Mesial frontoparietal disconnection and corpus callosotomy are then performed. Identification of the corpus callosum is facilitated by careful aspiration of the roof of the lateral ventricle just above its intersection with the septum pellucidum. The gray matter of the cingulate is easily identified. Aspiration of a portion of the cingulate allows clear visualization of the pericallosal arteries and corpus callosum. During this portion of the procedure the surgeon must be careful to avoid injuring either pericallosal artery, as it is difficult to ascertain which branch supplies the contralateral hemisphere. The cingulate and corpus callosum are then aspirated to achieve complete disconnection, being careful to avoid injury to the contralateral frontal lobe. The ipsilateral fornix is then interrupted by aspiration just anterior to the splenium in the posterior aspect of the lateral ventricle.

The anterior callosal dissection is continued, dividing the pia of the mesial frontal lobe and any branches from the anterior circulation, until the olfactory nerve is reached. The falx and pericallosal arteries provide an excellent anatomical reference to ensure complete frontal disconnection (see Fig. 52.7). The mesial frontoparietal pia and anterior circulation branches should be coagulated and divided above the inferior aspect of the falx to ensure that the contralateral frontal lobe is not injured. Posterior dissection is then performed along the falx until it transitions to the tentorium, effectively connecting to the infrasylvian basal temporal disconnection.

At this point, the only remaining connection is that of the basal orbitofrontal surface. The remaining orbitofrontal tissue, extending from the anterior sylvian fissure to the posterior basal frontal lobe is aspirated and the orbitofrontal pia divided. The gyrus rectus is aspirated and care should be taken to protect the contralateral frontal lobe. The lateral aspect of the pia along the olfactory nerve is divided to avoid injury to the olfactory nerve. The residual basal frontal lobe is removed with a posterior limit of the internal carotid artery.

Once the entire mesial pia has been divided, the draining veins to the sinuses are coagulated and divided sequentially. During this portion of the procedure it is common to have bleeding from the dura around the sagittal sinus. Packing the bleeding areas with a combination of hemostatic agents and cotton balls works well to avoid excessive blood loss. Once the draining veins have been ligated, the entire hemisphere has been disconnected and can be carefully removed. Finally, removal of the temporal mesial structures and insula is performed and the cavity irrigated and hemostasis achieved (see Fig. 52.7). A ventricular drain is placed in the cavity and the dura is closed in a watertight fashion. The bone flap is replaced and the wound is closed in anatomical layers.

Complication Avoidance

Complications of epilepsy surgery can be analyzed from a number of different perspectives: immediate versus delayed, minor versus major, general complications of craniotomy versus those specific to a given procedure, and neurological versus surgical. These divisions can be further broken down into expected versus unexpected and may depend on the location of the resection. If a postoperative deficit is anticipated as an unavoidable consequence of the procedure, and the patient receives counseling in light of this expectation, the resulting deficit is typically not considered a complication.

Tanreverdi and associates, in 2009, published an analysis of postoperative epilepsy surgery morbidity rates in 1905 patients over the course of 2449 epilepsy surgery procedures. Of these 1905 patients, 517 underwent extratemporal procedures consisting either of lobectomy or multilobar resections. Morbidity was subdivided by frontal, central, parietal, occipital, or multilobar resection location, resulting in major neurological morbidity rates of 0.7%, 1.4%, 0%, 5.1%, and 2.6%, and minor morbidity rates of 5.2%, 5.9%, 0%, 2.5%, and 12%, respectively. When surgical morbidity was analyzed across the same subdivisions of frontal, central, parietal, occipital, or multilobar resection, rates of complication were 5.6%, 1.5%, 14%, 0%, and 4%, respectively. Whereas parietal resections had the lowest neurological morbidity rate, they had the highest surgical morbidity rate of any reported group. Infection was the most common surgical complication, occurring in 2.1% of extratemporal procedures, followed by hematoma at 1.5%. Only one patient developed hydrocephalus.²⁰

With extratemporal resections, the most common neurological complications are those secondary to working in or near areas of eloquent cortex. Parieto-occipital resections, for example, are reported to have visual field deficits as their most common postoperative complication. This is followed by new or worsening hemiparesis, hemisensory loss, and aphasia.⁸ Resections in or near the dominant parietal lobe may result in a characteristic pattern of deficit. The most common complication of frontal lobe surgery is hemiparesis; however, aphasia is also a potential complication when operating in or near the dominant frontal operculum.²¹ The use of invasive monitoring to delineate eloquent cortex prior to resection aids in preventing and predicting postoperative neurological deficits. When surgical planning involves resection in or near functional cortex the relationship of the epileptogenic zone to eloquent cortex may limit the potential for seizure-free outcome.

Hemispherectomy and multilobar resection patients are at increased risk of specific complications, including hemorrhage, infarction, hydrocephalus, coagulopathy, anemia, and aseptic meningitis, when compared to those patients who undergo lobar resections.¹⁴ These risks are elevated in both the acute and delayed postoperative setting. In addition, disconnective hemispherectomy can be complicated by postoperative hemispheric cerebral edema or infarct, with resultant intracranial hypertension and death.²²

Complication avoidance is performed through a combination of both surgical and medical management techniques. Placement of intraventricular drains, subgaleal drains, peri- and postoperative dexamethasone, and admission to an intensive care unit are all utilized as part of complication avoidance. Routine monitoring of coagulation panel, platelets, and hematocrit every 6 hours for the first 48 hours postoperatively is performed for hemispherectomy patients, with early and aggressive correction of anemia and coagulopathy. Routine postoperative MRI is performed within the first 24 hours, assessing for extent of resection, presence of hemorrhage, and postoperative ventricular size. All hemispherectomy patients have computed tomographic imaging at a 6-week follow-up.