Surgical treatment of the extratemporal epilepsies


Address correspondence to Steven N. Roper, M.D., 100 South Newell Drive, Room L2-100, Gainesville, FL 32610. U.S.A. E-mail:


Epilepsy that originates outside of the temporal lobe can present some of the most challenging problems for surgical therapy. These epilepsies can be broadly categorized as lesional or non-lesional, with the nonlesional cases being the most difficult to localize. Lesional cases can result from malformations of cortical development, tumors, vascular malformations, or areas of old injury. Some lesions, such as focal cortical dysplasia, can be challenging, in that the boundaries of the pathology can be difficult to define. Presurgical goals include defining the structural lesion, the physiologic abnormality, and normal function in the area. These goals can be achieved using a variety of noninvasive and invasive tests. Surgical techniques vary depending on location and pathology but they always include removal of the epileptic brain tissue while preserving en passage vessels and underlying white matter tracts. Surgical outcomes vary depending on the underlying pathology. Surgeries are usually planned with a goal of no expected postoperative deficits, although temporary deficits may be anticipated in some areas, such as the supplementary motor cortex. Extratemporal epilepsy can be managed well with surgical treatment; but proper patient selection, evaluation, and discussion of expected outcomes and risks are critical in this challenging patient population.

It is estimated that 30% of patients with epilepsy will be refractory to medical therapy and should be considered for surgical treatment. Of these, the majority are for seizures arising in the anteromedial temporal lobe. The remainder comprise a heterogeneous group of patients with extratemporal lobe (or neocortical) epilepsy. Historically, among large surgical series, about 30–40% of resections are extratemporal (Williamson et al., 1993). These range from focal lesionectomies to large tri-lobar resections. Among resections limited to a single lobe, frontal resections are the most common (Janszky et al., 2000; Hosking, 2003; Jeha et al., 2007), followed by neocortical temporal, parietal, and occipital resections. For the sake of brevity and focus, this discussion concentrates on focal surgical resections for neocortical epilepsy (topectomy) rather than hemispherectomy or disconnection procedures, such as corpus callosotomy.

The extratemporal epilepsies present special challenges for a number of reasons. The first is that of defining the boundary of the resection. In medial temporal lobe epilepsy, the core structures that need to be removed are fairly consistent from one patient to another (i.e., hippocampus, parahippocampal gyrus, the uncus, and the basolateral amygdala). This is not the case for neocortical epilepsy. The center of the epileptogenic zone (EZ) can be anywhere on the cortex, and the boundaries are unique from one patient to the next. This means that the anatomic and physiologic measures that we use to set these boundaries are put to the test with every case. It also requires that the examiners have a fairly accurate idea of the location of the EZ prior to invasive monitoring, since you can sample only from areas where you have placed electrodes. Even after the EZ has been localized and boundaries have been defined, normal function of involved cortex may limit what can be achieved with surgical resection. In the following pages, we will examine some of the tools and methods that can be used to successfully address the special challenges of the extratemporal epilepsies.

Preoperative Work-up

The first goal of preoperative work-up is to determine if a patient has focal epilepsy and if it is extratemporal. This can be determined from history, semiology, interictal and ictal scalp electroencephalography (EEG), and imaging studies. Each test discussed in this article can be helpful in localizing the EZ in extratemporal epilepsy, but no single method is completely sensitive or specific. Therefore, final decision making always depends on assessing concordant results from several different diagnostic modalities (and sometimes ignoring tests that seem to be outliers). Successful surgical decision making is greatly enhanced by the experience of the team, an ability to grasp the gestalt from a heterogeneous assortment of data, and basic common sense. It also requires an appreciation of the limitations of our current methods for localizing extratemporal epilepsy and the risks involved in some of the required invasive tests. This may sound intellectually unsatisfying, but it is simply recognition of the current limitations in our understanding of some of the processes that produce extratemporal epilepsy. Someday, there may be a very clear and concise algorithm for the diagnosis and surgical management of all cases of extratemporal epilepsy, but we have not yet reached that point.

History can be very helpful in differentiating frontal from temporal lobe epilepsy. Some features of frontal lobe epilepsy include nocturnal predilection, brief duration, and absence of febrile convulsions in childhood. Semiology of observed seizures can also be helpful. Compared to temporal lobe seizures, frontal seizures are more likely to have an abrupt onset, rapid progression, complex postures [e.g., the fencer posturing associated with supplementary motor area (SMA) seizures], bipedal automatisms, loud, nonspeech vocalizations, and bizarre, hypermotor manifestations (Salanova et al., 1995b; Jobst et al., 2000; So, 2006). Interictal and ictal scalp EEG is also critical in localizing the site of seizure onset. Common problem areas for scalp EEG include the medial frontal lobe (e.g., the SMA), the orbitofrontal cortex (which can be indistinguishable from temporal lobe seizures), and small foci in the primary motor or sensory cortex (due to the small volume of tissue required to produce a clinical seizure). Parietal and occipital lobe seizures are much less common. Occipital seizures typically have primary visual auras and demonstrate a wide variety of spread patterns (Williamson et al., 1992b). Parietal lobe seizures may produce somatosensory sensations or other phenomena that may represent spread outside the parietal lobe (Williamson et al., 1992a; Salanova et al., 1995a).

Once extratemporal epilepsy is established, major goals, from a surgical perspective, are to define the pathology, define the region of abnormal physiology, define areas of normal function, and formulate a plan for surgical resection. Magnetic resonance imaging (MRI) is the foundation for detecting and defining structural pathology in epilepsy. This is very important, since lesional epilepsy has a much better outcome from surgery than cases without lesions. Foreign tissue lesions such as tumors, vascular malformations, and areas of old injury or stroke are usually well imaged with modern MR techniques. However, focal cortical dysplasia can be very difficult to identify and requires up-to-date MR protocols and experienced radiologists who are familiar with this disorder (Widdess-Walsh et al., 2006). Sometimes focused MRI scans with surface coils can help to identify subtle areas of cortical dysplasia. This may require interaction between the epilepsy team and the imaging team prior to the test so that the radiologists know where to focus their studies.

Despite every attempt to identify a structural lesion, some cases of extratemporal epilepsy will have normal MRI (Siegel et al., 2001). In those cases a second tier of noninvasive testing may be needed. These second-tier tests often require expensive equipment and specially trained personnel for their acquisition and interpretation. Therefore, not all epilepsy centers will have all of these tests available. This has made it difficult to perform direct comparisons of the relative utility of each test. Positron emission tomography (PET) has been well established as an alternative method for diagnosis of medial temporal lobe epilepsy. But it also has utility in diagnosing extratemporal lesions as well. These include areas of cortical dysplasia and cortical tubers of tuberous sclerosis (Salamon et al., 2008). PET scanning is usually performed as an interictal test and looks for baseline reduction in glucose metabolism (using 2-[18F]fluoro-2-deoxyglucose, or 18FDG). Single-photon emission computed tomography (SPECT) uses Tc-99m hexamethylpropyleneamine-oxime (HMPAO) to image blood flow in the brain. It is used to identify an area of increased blood flow shortly after the onset of a seizure. This has been very effective in localizing extratemporal foci (O’Brien et al., 2000, 2004). Ictal SPECT is often analyzed using SISCOM, a computerized method for subtracting the ictal from interictal SPECT and superimposing these data onto an anatomic MR image (O’Brien et al., 1998, 1999). This testing is labor intensive because it requires that the patient be in the hospital when the seizure occurs and that the tracer be injected within seconds of the seizure onset. Magnetoencephalography (MEG) is another important second-tier test used at some epilepsy centers. This technique uses magnetic rather than electrical sources to monitor interictal (and sometimes ictal) activity in the brain (Knowlton, 2008). It has several differences from EEG that allow it to provide information that may not be detectable in the electrical signal. It is also analyzed in such a way that the source of the magnetic transients can be presented on an anatomic MRI scan. This can be very helpful in directing electrodes coverage in patients that require invasive monitoring. Knowlton and colleagues have completed a prospective study comparing MEG, PET, and SPECT and concluded that each test can contribute to seizure localization in difficult surgical candidates (Knowlton et al., 2008a, 2008b).

Invasive Monitoring

Invasive monitoring is required more frequently in extratemporal epilepsy than medial temporal lobe epilepsy. Cases with foreign tissue lesions usually can go directly to surgery with intraoperative electrocorticography (ECoG). Cases with no abnormality on MRI will usually require invasive monitoring using subdural grids, sometimes with the addition of depth electrodes. The goals of invasive monitoring are to localize the site of seizure onset and to provide an avenue for mapping of normal cortical function. This involves a detailed review of all preoperative data so that all cortical areas under suspicion can be covered. In the United States, most centers use subdural electrodes in the form of grids and/or strips in extratemporal epilepsy. This often requires a large craniotomy for coverage of the area of cortex under suspicion. Despite careful planning, it is still possible to conclude invasive testing only to find that the epileptic focus is not completely bounded by the subdural electrodes, or even that the focus lies outside the grid coverage. The second-tier tests discussed previously are often used to direct subdural electrode coverage in an effort to avoid this situation. Because many of the subtle lesions seen on MRI are not visible at surgery, intraoperative stereotactic localization is often utilized to ensure proper placement of subdural electrodes. This also applies to nonlesional cases where localization of the EZ has been determined by PET, SISCOM, or MEG. Grids themselves have a significant complication rate, and there are various strategies for ameliorating the mass effect of the electrodes on the brain and for managing the slowly accumulating subdural hematomas that often accompany the grids. These include the placement of dural expansion grafts, loosely tying or hinging the patient’s bone flap, and leaving the bone flap out entirely during the invasive recording period. One large series reported an overall complication rate of 26%, with infection being the most common single complication at 12% (Hamer et al., 2002). These authors found that increased complication rates were associated with left hemisphere surgery, increased number of electrodes, and longer duration of monitoring. Despite the best available diagnostic methods and the efforts of expert epileptologists and surgeons, almost 40% of patients in one large series who had invasive EEG and normal MRIs ultimately had the electrodes removed without a resection (Wetjen et al., 2008).

Because subdural electrodes record only from the surface of the cortex, it can be difficult to localize seizures when they arise from deeper structures. For this reason, depth electrodes (thin, tubular electrodes that are placed within the brain parenchyma) are sometimes used in combination with subdural electrodes. Common sites for depth electrode implantation (in extratemporal epilepsy) would include the insula, the cingulate gyrus, areas of cortical dysplasia that reside deep within a sulcus, and areas of heterotopic gray matter. Depth electrodes are usually placed with the assistance of intraoperative stereotactic localization. In several European centers, there is a long tradition of using multiple arrays of precisely placed depth electrodes (stereo-EEG) instead of subdural electrodes for extratemporal lobe epilepsy, and these techniques have been used to yield very successful results in the hands of experienced practitioners (Munari et al., 1994; Cossu et al., 2005).

Surgical Resection

The most common extratemporal resection is lesionectomy. Lesions that can produce epilepsy include areas of cortical dysplasia, tumors (low grade gliomas, dysembryoplastic neuroepithelial tumors), areas of old stroke or traumatic injury, and vascular malformations (cavernous malformations) (Frater et al., 2000). In most of these lesions, excellent results can be achieved by complete removal of the lesion and some of the adjacent cortex. This peri-lesional resection is often directed by intraoperative ECoG. These specific techniques vary depending on the locations and types of lesion being removed. Cortical dysplasia is an exception to this rule. It is well recognized that resective surgery for cortical dysplasia has a lower success rate than surgery for other lesions. This is probably due to difficulties in defining the boundaries of the structural abnormality (for an illustrative case, see Fig. 1) and a relatively frequent involvement of cortical dysplasia in eloquent areas that cannot be resected. Therefore, it is not uncommon to perform invasive monitoring with subdural (and/or depth) electrodes for extraoperative recording of interictal and ictal EEG data and well as stimulation mapping of eloquent cortex (Widdess-Walsh et al., 2007).

Figure 1.

Imaging studies of a 37-year-old man who developed seizures at age 7. (A) He was found to have a presumptive focal cortical dysplasia (CD) in the right supplementary motor area (SMA) on initial magnetic resonance imaging (MRI) (arrow). He underwent subdural electrode implantation (B) and an initial resection in 1999 (C, boundaries outlined in white), which showed CD with balloon cells. He was seizure-free for 4 years but then seizures recurred and gradually worsened. In 2008, he underwent a second subdural electrode investigation that showed seizures originating in cortex just posterior to the old resection margin. A second resection was performed that extended back to the primary motor cortex (D, boundaries outlined in white). The patient had a transient contralateral hemiparesis (SMA syndrome) that resolved over the course of 3 weeks after surgery. He is currently seizure-free on short-term follow-up.

Topectomy based on physiologic and metabolic data is performed in cases where no structural lesion can be identified. The specific approach will vary depending on the location of the EZ. But there are some general rules for these resections. They include preservation of arteries and veins that are clearly passing through the EZ to supply normal brain tissue (en passage vessels). The importance of this rule is clearly demonstrated in resections that involve the opercular cortex around the sylvian fissure. Resections in these areas involve opening the crown of the gyrus and removing the gray matter in a subpial fashion down to the base of the adjacent sulci. This is because the major blood vessels of the cortex run within the sulci. It is also important not to carry the resection into the deep white matter tracts that lie below the cortex. This can result in deficits from adjacent eloquent cortex. This can sometimes be challenging in malformations of cortical development, where sulci may be abnormally deep and heterotopic neurons at the base of the sulci may blur the margin between gray and white matter. As mentioned earlier, intraoperative stereotaxis may be required to ensure removal of subtle structural or metabolic defects in addition to the physiologically defined areas demarcated by the implanted electrodes. As a general rule, if the electrical and structural/metabolic areas do not perfectly overlap, it is best to remove all presumptive abnormal cortex unless the resection is constrained by the presence of eloquent cortex.

Surgical Outcomes

Outcomes from extratemporal resections vary widely across the literature, but seizure-free status is clearly less common with these cases compared to resections for medial temporal lobe epilepsy. Because lesionectomies have much better outcomes than nonlesional cases (Janszky et al., 2000), the results from an individual center are often dependent upon the preponderance of lesional cases within that center’s experience. Two articles have used evidence-based approaches to review the literature and have provided similar findings for outcomes from extratemporal resections. Engel and colleagues (Engel et al., 2003) reviewed outcome data on 298 neocortical resections and found a seizure-free rate of 50% and an improvement rate of 79%. Tellez-Zenteno and colleagues performed a review and meta-analysis of surgical outcomes and found long-term seizure-free rates of 27% for frontal, 46% for occipital and parietal, and 34% for grouped extratemporal resections (Tellez-Zenteno et al., 2005). However, it should be noted that results for occipital and parietal resections were based on single reports and that the results for frontal resections, based on seven reports, varied from 9–80% seizure-free. Results from the Cleveland Clinic show that the seizure-free rate for frontal lobe resections drop from 56% at 1 year follow-up to 30% at 5 years (Jeha et al., 2007). Independent predictors of poor outcome in this study were MRI-negative malformations of cortical development, MRI abnormalities outside the frontal lobe, and generalized or nonlocalized ictal EEG patterns. MRI-negative patients have the poorest outcomes from surgery. However, one study reported a seizure-free rate of 57% in frontal resections with normal MRI, where invasive monitoring demonstrated a focal ictal onset (Siegel et al., 2001). The Mayo Clinic reported similar results, with 50% of MRI-negative extratemporal resections having an Engel Class I outcome (free from disabling seizures) if they had focal ictal EEG onset by invasive monitoring (Wetjen et al., 2008). However, this rate dropped to 27.5% if all implanted patients (including patients that were implanted but ultimately had no resection) were considered. However, one provocative study has shown that ultimate outcome from surgery is not significantly different between MRI-positive patients and those with normal imaging, although the majority of these patients had temporal lobe epilepsy (Alarcon et al., 2006).

Of the lesional cases, cortical dysplasia has the poorest outcomes for seizure control. The Cleveland Clinic reported a 49% Engel Class I rate in 35 patients with resections for cortical dysplasia, with no significant difference between temporal and extratemporal resections (Edwards et al., 2000). However, a series from Bonn, Germany reported a 72% seizure-free rate in 53 patients with resections for cortical dysplasia. Again, no difference was seen between extratemporal and temporal resections, but large, multilobar resections did not do as well (Kral et al., 2003). Completeness of resection is an important factor in seizure control with cortical dysplasia (Wyllie et al., 1987). There is some encouraging evidence that surgical outcomes for cortical dysplasia may be improving with the advent of better imaging techniques (Salamon et al., 2008) and better understanding of the clinical characteristics of the different types of cortical dysplasia (Lerner et al., 2009).

Complications from extratemporal resections are rare but can be serious. The risk of complications from invasive EEG recordings was discussed previously and must be included in assessing the overall risk of surgical management. In their large literature review, Engel and colleagues found a 0.4% risk of perioperative death, 6% risk of new neurologic deficit, 5% risk of infection, and 6% risk of cognitive or behavioral changes (Engel et al., 2003). This analysis did not differentiate temporal from extratemporal resections. Behrens and colleagues reported complications from 429 consecutive therapeutic surgeries for epilepsy and found an infection rate of 4.9% and a neurologic complication rate of 5.4%, with 3% being transient (Behrens et al., 1997).

Complications specific to extratemporal resections are related primarily to proximity to eloquent cortex—which begs the question, “What areas can and cannot be removed?” Resections within the primary motor and sensory cortex, Wernicke’s language area, and Broca’s speech area are generally considered off-limits. The only exceptions to this rule would be cases with severe preexisting deficits, cases where frequent seizure activity has made an extremity functionally unusable (e.g., epilepsia partialis continua), or if the seizure disorder itself is truly life-threatening. Fortunately, these instances are rare and surgical decisions must be made carefully and in close communication with the patient and his/her family on an individual basis. Another exception is the nondominant motor cortex for the face. Resections in this area can often be performed with little or very mild facial weakness (Rasmussen, 1975), and this may be acceptable if the chances of surgical success are high. The SMA represents a special area where resections can produce profound initial deficits that resolve over the course of 2–3 weeks. The SMA resides in the medial frontal lobe, just anterior to the foot primary motor cortex and appears to be involved in planning of motor activity (Fried et al., 1991). Resections in this area can produce contralateral hemiplegia (although ipsilateral weakness can also be seen) with normal tone, mutism (most common in dominant resections), and neglect (Laplane et al., 1977; Rostomily et al., 1991; Zentner et al., 1996). The symptoms gradually resolve over 2–3 weeks (41 days in one case) in most cases (Zentner et al., 1996). Resections can be safely performed in the SMA, but patients and their families must be counseled on expected, transient postoperative deficits.


Extratemporal epilepsy presents the most challenging problems in the field of epilepsy surgery. These relate to difficulties identifying certain pathologies and their boundaries, imperfect understanding of the relationship between certain pathologies and the EZ, and limitations imposed by functional cortex that may participate in seizure initiation. Our ability to manage these patients has dramatically improved with newer imaging techniques such as high resolution MRI, PET, ictal SPECT, and MEG. There is also a growing understanding of ictal EEG patterns and their predictive value for surgical resection. Therefore, the future of surgery for extratemporal epilepsy is bright, but there are clearly many issues that need to be addressed to provide optimal care for these challenging patients. More than any other patient population, they require the coordinated attentions of epilepsy experts from multiple disciplines including clinical neurology, structural and metabolic imaging, neurophysiology, and experienced surgeons with stereotactic systems at their disposal.


Disclosure: The author declares no conflict of interest.