Temporal Lobe Resection

The role of tailored resection is least agreed on for epilepsy involving the temporal lobe. There are standardized approaches to temporal lobe epilepsy surgery that vary slightly according to side with respect to language dominance and inclusion of mesial structures, and reasonably good seizure outcomes overall have long been reported with these standardized approaches. The 1991 survey of epilepsy surgery centers compiled by Engel for the Second Palm Desert International Conference on the Surgical Treatment of the Epilepsies, held in Indian Wells, California, in February 1992, found seizure-free outcomes in 67.9% of patients undergoing anterior temporal resection.⁴ Recognition of variability among patients with respect to both the epileptogenic zone and language cortex, however, has led a number of surgical epilepsy teams to advocate tailoring the resection.

Distinguishing between mesial temporal sclerosis and temporal neocortical epilepsy represents one of the first rationales for implementation of this strategy. It was the ability to diagnose seizure onset localized to the hippocampus or amygdala that made selective amygdalohippocampectomy a viable treatment in the first place. Intracranial recording, usually with depth electrodes, made possible the successful selective procedures reported from Zurich^5,6; alternatively, neuroimaging showing localized lesions limited to medial structures similarly sufficed. Conversely, neurophysiologic or radiographic data indicating a pathologic process confined to neocortical tissue justified a surgical procedure that spared the amygdala and hippocampus. These strategies are widely implemented today with the use of advanced magnetic resonance imaging (MRI) data or invasive monitoring, or both. Surgical planning and decision making in these instances generally occur outside the operating room, before the actual resective procedure.

Intraoperative “tailoring” of the resection procedure with the assistance of ECoG and functional mapping data obtained at the time of surgery is widely performed today. Intraoperative recordings, with rare exception, necessarily rely on interictal epileptiform discharges, or spikes, rather than ictal onset, which can be diagnosed only with extraoperative studies. The relationship of such interictal spikes to the epileptogenic zone or the area whose resection will result in a seizure-free outcome is not as clear as that for ictal-onset data. Animal models using either penicillin or alumina foci have demonstrated tight correlations between spikes and the epileptogenic zone.^7,8 However, human clinical investigations have shown varying degrees of correlation. Using preresection and postresection intraoperative ECoG in a series of 29 patients undergoing standard temporal resection for mesial temporal sclerosis, Schwartz and colleagues investigated the predictive value of corticography for seizure outcome and found no correlation.⁹ Sugano and associates, in contrast, reported an analysis of 35 temporal lobe epilepsy patients who had undergone either lesionectomy or resection of a lesion plus tissue with residual spike activity (most often the hippocampus) and noted 3-year seizure-free rates of 76.9% and 90.9%, respectively.¹⁰ This relationship plus the enhanced ability of ECoG recording to accurately define an epileptogenic zone whose resection will lead to cessation of seizures is one of the two major rationales for tailoring resections.

The other rationale for individualization of temporal lobe seizure surgery relates to the known variability in the spatial representation of normal temporal lobe functions, particularly those related to language. Conventional teaching for untailored temporal lobe resection surgery advocates removing no more than the most anterior 4.5 cm of the middle temporal gyrus in the language-dominant hemisphere and preferably sparing the superior temporal gyrus entirely. With extensive experience in mapping language during epilepsy resections, however, Ojemann and coworkers have clearly documented the high variability of cortical language representation, with some critical language sites observed within anterior temporal lobe regions that would be removed during a standardized resection procedure.^11–13 On this basis, cortical mapping has been strongly advocated.¹⁴ Those who argue against this position would note that others have reported the absence of new language deficits after standard resection.¹⁵

Extratemporal Neocortical Resection

The strategy of resecting identified epileptogenic cortex while preserving normal function similarly applies to surgery on the frontal, parietal, and occipital lobes. Individualized investigation is generally required, given the absence of analogous epileptic syndromes and standardized surgeries, with the possible exception of some epilepsy cases associated with well-defined lesions. Although recognition of structural pathology on imaging studies may guide the approach, appreciation of the extent of the surrounding resection and the proximity of eloquent cortex usually benefits from extraoperative or intraoperative recording with or without functional mapping. MRI-negative or nonlesional epilepsy necessitates such an approach.

Frontal lobe epilepsy resection requires preservation of primary motor cortex and, on the language-dominant side, speech cortex. Although variability in the primary motor areas has been noted, it is considerably less variable than the critical language sites. Under image guidance, planned surgical approaches well anterior to the precentral gyrus may be performed in accordance with the structural anatomy; incorporation of tractography or functional MRI (fMRI) data provides only further reassurance. Resections that are planned closer to eloquent brain regions, however, are facilitated by intraoperative mapping and monitoring. Language preservation with frontal resection requires sparing of Broca’s area in the posterior inferior frontal gyrus, but recent intraoperative studies have documented more extensive and variable frontal lobe language representation as well.^16,17

Parietal resections must preserve primary sensory cortex and, on the language-dominant side, perisylvian language areas. Optic radiations, generally deep to the epileptogenic cortex, are at risk with larger resections but are not generally mapped intraoperatively. When operating within the occipital lobe, the primary visual cortex must be preserved. Although the accumulated surgical experience in posterior brain regions is much more limited than that in the temporal and frontal lobes, surgical interventions in these posterior regions are being carried out with increasing frequency and encouraging results.^18–23

Multilobar Resection

Multilobar resection for the treatment of medically intractable epilepsy has understandably been used far less frequently than well-recognized interventions such as anterior temporal lobe resection or single-lobe cortical resection. The extent of surgery, its potential morbidity, associated functional consequences, and not least, the lower seizure control efficacy rate underlie the smaller surgical numbers, and the published experience with multilobar surgery is limited. Engel’s 1991 survey described comparative seizure control outcomes, with multilobar resection achieving seizure-free outcomes in 45.2% of patients and improvement in an additional 35.5%; anterior temporal lobe resection and lesionectomy, in contrast, achieved seizure-free outcomes in 67.9% and 66.6% of patients, respectively.⁴ Patients undergoing these different procedures, however, are distinctly different patient populations, and it would be a misinterpretation of that previous experience to deny multilobar epilepsy patients consideration of possible surgical intervention. Since the time of Engel’s survey, developments in both evaluation—neuroimaging as well as intracranial investigation—and surgical technique have improved the risk-benefit ratio for these patients.^24–27

Preoperative Evaluation

Surgical evaluation of an epilepsy patient today requires the multidisciplinary team approach of a dedicated epilepsy service. In addition to a general and epilepsy-focused history and physical examination, routine evaluation will include scalp electroencephalography (EEG) and epilepsy-protocol MRI; in those in whom surgery appears to be a reasonable preliminary consideration, further neuroimaging with positron emission tomography or single-photon emission computed tomography (SPECT), or both, inpatient video-EEG monitoring, ictal SPECT, and neuropsychological testing generally follows. Not all patients require all of these studies. In patients with MRI-negative epilepsy, subtraction SPECT (ictal minus the coregistered baseline) has often proved invaluable in directing subsequent intracranial investigative studies.^28,29 Review and consideration of all this information by a multidisciplinary epilepsy team in a regular working conference subsequently ensue. At this point, patients without surgically treatable epilepsy, those who require additional medical management, and those with straightforward conditions such as medial temporal lobe epilepsy or structural lesions confined to a single lobe will have been identified. More challenging patients, such as those with discordant data, apparent multifocal or bilateral disease, MRI-negative epilepsy, or a more extensive pathophysiologic substrate, require further evaluation.

Intraoperative Mapping and Electrocorticography

When used, intraoperative ECoG may be performed with either awake or general anesthesia; in either instance, attention to anesthetic considerations is critically important. Wide exposure of the suspected cortical region is essential. After visual inspection of the cortical surface, electrodes of a variety of types may be used. A combination of grids and strips can be used to obtain the desired number of contacts, regularity of spacing, and access to less exposed surfaces such as the inferior temporal or orbitofrontal cortex. Large 4 × 8 or 8 × 8 subdural grids with 1-cm spacing of contacts may be placed easily over the convexity (Fig. 64-1); smaller subdural grids and strips (most typically 1 × 8) may supplement the larger grids. In lesional cases, image guidance may be helpful in directing manual placement of these electrode arrays. Intraoperatively placed depth electrodes may be used, particularly electrodes placed into the medial temporal lobe structures, but this has become less common. Once placed, ample recording is performed, and in collaboration with electrocorticographers in the operating room, the epileptogenic zone that would ideally be resected is defined. After the initial recording, mapping (see the next paragraph), and subsequent resection, further ECoG is often performed, although the significance of postresection spikes and the appropriate surgical response are debated. Advocates of this technique cite reports of better outcomes associated with absence of residual epileptiform activity,^30–32 although resection in pursuit of residual spike activity is controversial and not universally advised.⁹

FIGURE 64-1 Intraoperative electrocorticography is being performed in this patient with complex partial epilepsy and a right temporal cavernous angioma. A 4 × 8 subdural grid has been placed over the right temporal lobe convexity (anterior is top left; superior is bottom left).

Intraoperative mapping, most commonly for language, motor, or sensory function (or any combination of these functions), typically follows the initial ECoG. For the primary sensorimotor cortex, mapping has a long and successful history. Integration of preoperative MRI or ideally fMRI data through the use of image guidance methods greatly facilitates this activity. Placement of a subdural strip perpendicular to the presumed central sulcus is a method that can be used to obtain somatosensory evoked potentials and assist in localization of the precentral and postcentral gyri. Direct motor mapping may be performed with either a handheld stimulator or a 1 × 4 subdural strip, with appropriate electromyographic recording increasing the sensitivity. Once the motor cortex has been identified, an extremely useful practice involves placing a subdural strip electrode over that area, delivering repeated electrical stimuli, and monitoring motor evoked responses to examine the integrity of the corticospinal tract during the resection.^33–35

Language mapping requires an awake craniotomy, and with an experienced team, this approach can be used routinely in an efficient and safe manner. Numerous descriptions of awake craniotomy techniques have been published.^13,16 A handheld bipolar stimulator consisting of 1-mm electrodes with a separation of 5 mm is used to apply a 5-second, 60-Hz, biphasic square-wave constant-current or constant-voltage stimulus at sequential sites across the exposed cortex of interest, beginning with low-stimulation parameters and gradually increasing the stimulation intensity until either language is disrupted or the maximal subthreshold for afterdischarges is reached (Fig. 64-2). Adjacent electrodes monitor the cortex for afterdischarges, and cold irrigation or intravenous propofol is available as needed to manage stimulation-induced seizures. Broca’s area is most easily identified, with arrest of speech during counting being readily elicited; speech interference by a direct motor response must be excluded. Beyond Broca’s area, naming is assessed by using pictures of objects presented on an appropriately positioned computer screen during application of the stimulus. The cortex is investigated in this manner at approximately 1-cm intervals, with each site typically tested three times. Numbered labels are used to identify sites that have been tested, and intraoperative photographs can be used to document the results of the functional mapping procedure. Subsequent resections are generally kept a minimum of 1 cm from positive language sites, although this conventional dictum does not take into account sulcal patterns that may warrant its modification.

FIGURE 64-2 Functional mapping of language in the left temporal lobe with a handheld bipolar stimulator. Numbered tags identify previously tested sites. The green outline, superimposed on the surgical field with the heads-up display of a neuronavigation system, represents the boundary of an underlying lesion.

Extraoperative Intracranial Investigation

Localization of the area of seizure onset, as well as cortical function, can also be achieved extraoperatively with implanted electrode arrays. Although this strategy necessitates an additional operative procedure, there are also advantages related to intraoperative recording. ECoG data recorded outside the operating room are obtained in a setting that more closely resembles the patient’s normal environment, and these recordings can be obtained for periods of days or even weeks. Actual seizure onset may be recorded rather than relying on interictal spike activity as its surrogate. Repeated seizures can be captured, and when a consistent pattern is observed across all events, it is likely that these data accurately reflect the patient’s usual pattern of epilepsy. With respect to mapping normal function, the extraoperative setting enables more extensive and repeated testing without the constraints of anesthetic requirements or operative time. Extraoperative methods are most valuable in a patient in whom the actual region of seizure onset has not been confidently defined. Bilateral, multilobar subdural strip and depth electrode placement, for example, may identify seizure onset in patients who would otherwise be deemed unsuitable for surgical resection. Particularly in patients with MRI-negative epilepsy, placing multiple electrode types and large grid arrays over the general regions of interest may be the only effective strategy for identifying a resectable seizure focus. The utility and low morbidity associated with subdural grid electrodes have been well documented.³⁶

Intracranial Electrode Implantation Procedure

The actual array that may provide the most information in any given case is individualized. Relatively restricted, superficial cortical lesions may require subdural grid coverage only. If there is a question of involvement of the hippocampus or other deep structures or if some coverage of the opposite hemisphere is also desired, the addition of depth electrodes can be helpful. The craniotomy exposure required for grid placement readily accommodates multiple subdural grids, as well as placement of subdural strip electrodes. Determination of an individual electrode implantation plan is optimally a multidisciplinary process.

Implantation is generally performed with the patient under general anesthesia, and in most instances, use of an image guidance system is helpful in ensuring accurate electrode placement. After induction of general anesthesia, the patient is positioned with the head secured in three-point pin fixation, the head is coregistered with the neuronavigation system by using scalp fiducials and preoperative MRI, and the scalp is shaved, prepared, marked, and draped in conventional manner. Incisions are planned in anticipation of having sufficient exposure for subsequent resection. Attention to preparing a wide field, with planning of electrode exit sites, is essential. During this preparation the patient is administered prophylactic antibiotics, and if the array will entail a more difficult exposure requiring retraction, such as for an interhemispheric array, mannitol is also administered. If the array involves depth electrodes in a region apart from the subdural grid array, they are usually placed (whether by frame-based or frameless stereotactic methodology) before the craniotomy. If depth electrodes will be placed in the same region as the subdural grid, they are most easily placed via a frameless methodology, and in that instance accuracy is best ensured by implantation of the depth electrodes through the craniotomy exposure but before grid placement.

The craniotomy is performed with standard neurosurgical technique. There are two schools of thought regarding management of the bone flap. Many centers in the past have replaced the bone flap at the close of the implantation, in the same manner as for any craniotomy. Alternatively, in an attempt to reduce the risk for subsequent infection by bacterial colonization along the electrode leads and at the same time to simplify the procedure, the bone flap may be left out during the interval while the patient is monitored, with replacement at the time of the subsequent procedure for electrode removal and resection. We keep the bone flap sterile in a freezer and have not experienced significant morbidity or other difficulties related to using this approach. Particular attention to hemostasis during the opening as well as throughout the procedure reduces the risk for development of a symptomatic epidural or subdural hematoma; the postoperative appearance of some blood in the region of the electrodes has been specifically reported but is without clinical significance.³⁷

If placement of depth electrodes in the craniotomy site is planned, they are inserted immediately after opening the dura. We have placed them through cut openings in an already placed subdural grid, as well as before grid placement, with a right-angle bend at the pial surface held in place by the overlying grid, and found the latter preferable. Subdural grid placement is planned preoperatively so that it covers the suspected areas of seizure involvement, in addition to any relevant areas of potential functional eloquence. More difficult electrode placements are performed first to minimize displacement of others, and more easily displaced electrodes, including subdural strip electrodes, are placed last. Besides these electrodes, a four-contact subdural electrode is placed away from the cortex of interest, with the contacts facing the dura, to serve as a reference electrode during recording.

All electrode leads are brought through separate exit sites with the use of temporarily placed, large-bore catheter sheaths; typical tunneling distances are several centimeters. Electrodes are secured to the scalp at their exit sites with a single suture, which also serves to tighten the track and prevent leakage of cerebrospinal fluid. The dura is reapproximated around the electrode leads, but no attempt is made to render the closure watertight at the level of the dural openings. With hemostasis ensured, the galea is then reapproximated and the scalp closed with running nylon or Prolene suture. Bacitracin ointment, Xeroform, and sterile dressings are applied. Most importantly, the electrodes are individually and securely labeled in the recognition that a mislabeled electrode is far worse than none.

Postoperatively, a fine-resolution CT scan of the head is obtained before the patient goes to the epilepsy monitoring unit, and this study is then fused with the patient’s preoperative MRI for optimal understanding of the final electrode positions (Fig. 64-3). The patient is now monitored until (ideally) at least three typical seizures have been recorded, during which time prophylactic antibiotics are administered, although this is a recognized controversial precaution. Use of anticonvulsants is weaned if necessary. Recording is usually performed for 2 days up to 4 weeks, with an average time of 7 days. Functional mapping, unless not required, is performed during this interval. The data are reviewed by the multidisciplinary team, and surgical recommendations are developed and discussed with the patient and family.

FIGURE 64-3 This composite graphic using coregistered preoperative magnetic resonance imaging (for surface anatomy) and postoperative computed tomography (for electrodes) shows the electrode contacts of two contiguous 4 × 8 subdural grid electrodes overlying the left frontal and temporal lobes. This array was used both for electrocorticographic localization of seizure onset and for functional mapping.

Resective Procedure

With a treatment plan in place, the extraoperatively investigated patient is returned to the operating room and generally reregistered with an image guidance system using either the remaining scalp fiducials or natural landmarks. In addition, the intracranial electrodes in place may be used to guide the resection. With the patient and head positioned, the sutures securing the exiting electrode leads are cut, and then the previous incision line and adjacent scalp are prepared and draped so that the exit sites are left outside the sterile field. The incision is now reopened, the electrode leads divided at the level of the dura, and the now divided distal electrode leads withdrawn from the exposed surgical field from under the drapes. The dura is reopened, and the exposed cortex and subdural electrodes are inspected for any evidence of displacement. Electrodes away from the area to be resected are removed, and if image guidance is not reliably available, the locations of the contacts of the subdural grid overlying the margins of the planned resection are marked on the cortical surface with bipolar cautery.

The actual resection is performed under the operating microscope with standard microsurgical technique. If there is any question of functional cortex being at risk, functional mapping or monitoring, or both, may be performed at the time of resection. Should recordings implicate cortex whose functionality precludes resection, that cortex is generally preserved; alternatively, multiple subpial transections in this portion of the cortex may be combined with resection elsewhere. With large resections, care must be taken to preserve large arteries and veins within the resection margins that may be associated with brain tissue outside the planned resection. On completion of the resection, the dura is closed over Gelfilm, and the bone flap is secured in place with a standard plating system. The remainder of the closure follows standard technique. The patient is mobilized immediately after the procedure, and attention is directed at ensuring appropriate anticonvulsant levels. Discharge follows in 1 to 3 days.