The association between epilepsy and intracranial neoplasms is well established. Seizures occur in 58% of patients with primary brain tumors and 34% of metastases, frequently as the presenting symptom (Lynam et al., 2007). Seizures sometimes persist even after resection, and occasionally they arise de novo after resection (You et al., 2012). Factors that play a role in the pathogenesis of epilepsy in these patients include the histology and location of the lesion, the cellular and molecular surrounding environment, the local blood supply, and the effects of medical and surgical interventions (Shamji et al., 2009; You et al., 2012). Invasive monitoring techniques such as subdural grids and depth electrodes can be helpful to identify the epileptogenic zone (EZ) and its relationship to eloquent brain in this population. In this article, we briefly review the causative factors contributing to epilepsy in patients with intracranial tumors, describe the invasive strategies used to identify the EZ and eloquent tissue, and present the data obtained from a series of patients managed at our institution as an illustrative case series.
Patients with intracranial mass lesions are at increased risk of intractable epilepsy even after tumor resection due to the potential epileptogenicity of lesional and perilesional tissue. Risk factors for tumoral epilepsy include tumor location, histology, and extent of tumor resection. In epilepsy that occurs after tumor resection, the epileptogenic zone often does not correspond precisely with the area of abnormality on imaging, and seizures often arise from a relatively restricted area despite widespread changes on imaging. Invasive monitoring via subdural grids and/or depth electrodes can therefore be helpful to delineate areas of eloquence and localize the epileptogenic zone for subsequent resection. Subdural grids offer excellent contiguous coverage of superficial cortex and allow resection using the same craniotomy, facilitating understanding of anatomic relationships. Depth electrodes offer superior coverage of deep structures, are easier to use in cases where a previous craniotomy is present, are not associated with anatomic distortion due to brain shift, and may be associated with a lower complication rate. We review the biology of focal postoperative epilepsy and invasive diagnostic strategies for the surgical evaluation of medically refractory epilepsy in patients who have undergone resection of intracranial mass lesions.
Factors Contributing to Epileptogenicity
Histology and location
Focal seizures are a common presenting symptom in patients with intracranial lesions, but not all tumors carry the same risk of epileptogenicity. Patients with low-grade, primary tumors of the central nervous system have a higher likelihood of producing seizures than those with high-grade glial tumors or metastases (Lynam et al., 2007; Shamji et al., 2009; You et al., 2012). In one large series, oligodendrogliomas and World Health Organization (WHO) grade 2 astrocytomas were associated with a significantly higher incidence of epilepsy than high-grade astrocytomas (Lynam et al., 2007). One explanation for this finding might be the insidious growth pattern of these tumors, resulting in slow invasion and possibly deregulation of the surrounding brain parenchyma (Shamji et al., 2009). This process of neoplastic infiltration of peritumoral tissue may in part explain the focal pattern of epilepsy that develops in these patients.
The intracranial location of a tumor may also play a role in epileptogenicity. Superficial supratentorial tumors have a greater association with seizures than deeply seated or infratentorial lesions (Shamji et al., 2009). In addition, tumors of the occipital lobe are less likely to result in seizures than those in the frontal or parietal region, and temporal tumors are the most epileptogenic (Shamji et al., 2009), which may be due to the variable thresholds for epileptic activity in different areas of the brain. In particular, temporal lobe epilepsy may result from dual pathology, with hippocampal sclerosis or focal cortical dysplasia that coexists with a tumor in temporal or extratemporal structures (Englot, 2012). The notion that the location of a tumor influences whether a patient develops epilepsy may help to explain why seizures in such patients tend to arise from discrete foci, and removal of these is curative.
Additional factors thought to be involved in the pathogenesis of epilepsy in patients with brain tumors include changes in the microenvironment. Abnormal neuronal migration as seen with tumor growth, variability in neurotransmitter release such as increased levels of glutamate, and imbalances of excitatory and inhibitory processes may all act to alter the peritumoral environment (Vecht & van Breemen, 2006; Shamji et al., 2009). Moreover, tumoral hypoxia due to insufficient blood supply may result in acidosis with sodium channel activation and subsequent sodium influx. These changes, along with increased intercellular communication with nonfunctioning gap junctions between glial cells, increase the risk of seizure propagation (Shamji et al., 2009). Finally, disruption of the blood–brain barrier may further predispose patients to seizure activity, which may be a result of not only the neoplastic process itself but of systemic treatments, such as chemotherapeutic agents (Shamji et al., 2009; Englot, 2012). Each of these changes occurs in the area surrounding the tumor, and focal epilepsy may therefore develop even after removal of the lesion.
Extent of surgical resection
A well-established independent risk factor for persistent postoperative seizures in patients with brain tumors is the extent of tumor resection. A gross total resection is correlated with a higher likelihood of seizure freedom, ascribed to the increased chance of removal of the epileptogenic zone (You et al., 2012). Conversely, a subtotal resection is a major independent risk factor for both early and delayed postoperative seizures (Gump et al., 2013).
Role of Invasive Monitoring
The goal of the presurgical evaluation is to identify the EZ and its anatomic relationship to eloquent tissue. Noninvasive measures, such as surface electroencephalography, may not offer sufficient localization information, whereas electrodes placed in direct contact with the brain allow much higher resolution. In addition, invasive electrodes can be employed both to record electrical activity from the brain during active seizure activity and to map zones of eloquence in a much more comfortable and natural setting than is possible using intraoperative techniques. These techniques involve implantation of intracranial electrodes that are externalized for several days (up to a few weeks) in an inpatient setting to allow continuous physiologic monitoring during seizure activity. Two strategies commonly employed for invasive monitoring include subdural electrodes, which are placed on the surface of the brain, and depth electrodes, in which the recording electrodes are implanted in the region of seizure onset (Fig. 1, Table 1).
|Superficial cortex||Excellent|| |
If carefully planned
|Anatomic relationships||Straightforward||Can be challenging|
|Functional mapping|| |
Regular contiguous pattern
Requires careful planning
|Previous craniotomy||Difficult|| |
Must avoid hardware
|Subsequent resection|| |
Use electrodes to guide
Reopen same craniotomy
Must use stereotaxy
Performed after removal
Can use burrholes for strips
The use of subdural electrode arrays (strips and grids) evolved out of the intraoperative use of electrocorticography pioneered by Foerster in the 1930s (Rosenow & Luders, 2001). Extraoperative use of externalized electrodes lying directly on the surface of the brain allows for observation of the onset and early spread of seizure activity and correlation with interictal activity. Placement of subdural electrodes for chronic monitoring requires exposure of the cortical surface, typically through a generous craniotomy with wide opening of the dura, followed by placement of a grid centered on the area of interest. Grid or strip electrodes can be advanced beneath the dura beyond the extent of the craniotomy, and strips can also be placed in a minimally invasive fashion using a burr hole unilaterally or bilaterally, although the latter can be more technically challenging. In cases in which a previous craniotomy was performed, such as for resection of tumor, subdural placement may be difficult depending on the degree of scar tissue present.
There are numerous advantages offered by subdural electrodes. Given the two-dimensional nature of the arrays, the potential surface area covered by the electrodes is extensive, enabling excellent coverage of the superficial cortex, which can be extremely advantageous in circumstances in which functional mapping is required. Because a significant portion of the primary speech and motor areas are located on the convexity surface, subdural arrays are able to cover these structures, and stimulation of specific electrodes overlying eloquent areas of the brain will unequivocally result in a transient functional deficit. The additional benefit of extraoperative mapping over the course of several days, rather than in a single session in the operating room, is that it allows for more accurate data acquisition because patients are more comfortable, less fatigued, and not influenced by anesthesia. This approach also allows repeated stimulation to ensure that findings are reproducible.
Another advantage of subdural array techniques is that subsequent resection of the EZ is often less demanding following placement of a subdural array, due to the relative ease of reopening the craniotomy used to place the grid, as well as the straightforward anatomic relationship of the grid electrodes to the underlying cortex. Surgeons can localize eloquent tissue according to the findings from the previously performed functional mapping and resect the epileptogenic areas while sparing those contacts responsible for producing deficits, making subdural electrodes a preferable option for many clinicians.
However, subdural grids are associated with a complication rate of 5–30%, with permanent morbidity in 2–5% of patients (Hamer et al., 2002; Onal et al., 2003; Johnston et al., 2006; Wong et al., 2009; Morace et al., 2012). Complications include hemorrhage (particularly subdural hematomas), infection, cerebral edema, venous infarction, cerebrospinal fluid leakage, and transient neurologic deficit in addition to electrode malfunction. Continuous antibiotics may not reduce risk of infection (Wyler et al., 1991). Risk factors for complications include patient age, left-sided placement, and length of evaluation (Hamer et al., 2002). A greater number of electrodes may lead to increased risk (Hamer et al., 2002; Onal et al., 2003), although this claim is controversial (Johnston et al., 2006).
The use of intracerebral electrodes to sample deep tissue was popularized by Jean Talairach, who described use of multiple depth electrodes to cover deep structures in three dimensions for stereo electroencephalography (SEEG; Guenot et al., 2001). As originally reported, SEEG involves placement of electrodes into the brain perpendicular to the sagittal plane, making it possible to identify target structures on a standard atlas and avoid blood vessels using angiography. As imaging and stereotactic technologies have improved, it has become possible to implant depth electrodes from any direction, using contrasted magnetic resonance imaging (MRI) with or without stereotactic angiography to avoid vascular structures. In addition, placement of electrodes using either the orthogonal Talairach approach or from variable angles is often much less invasive than placing subdural grids, as no craniotomy is necessary. Moreover, implantation of electrodes in patients who have undergone previous craniotomies is not limited by scarring, as is the case for subdural arrays. Although each individual depth electrode provides only one spatial dimension, when multiple electrodes are used it is possible to obtain detailed information about seizure onset and spread as well as eloquent tissue, including deep structures and white matter, especially if electrodes are placed in a tightly spaced three-dimensional (3D) SEEG grid, a technique frequently employed at our institution (Fig. 2).
Whereas subdural arrays are particularly useful for depicting superficial cortical anatomy, 3D-SEEG conveys information from subcortical and cortical structures alike. Although subdural arrays can be placed to cover mesial and basal structures by advancing electrodes into the interhemispheric or subhemispheric subdural space (Bekelis et al., 2012), this blind procedure can be risky. However, with depth electrodes, surgeons have the ability to strategically implant electrodes from any trajectory to record from the surface of the cortex, the mesial cortex, and the cortical structures deep within the sulci in addition to the subcortical structures and white matter tracts. With up to 12 electrode contacts on a single electrode, it is possible to sample a sizeable territory of brain per electrode, thus encompassing a large volume of tissue for recording the EZ and for performing functional mapping extraoperatively. Furthermore, there is a greater degree of accuracy with respect to the location of each depth electrode contact than there is for subdural arrays given the stereotactic placement of the electrodes, making mapping somewhat more reliable, largely due to minimal brain shift that occurs with depth electrode insertion in contrast to subdural arrays, which are frequently associated with a high degree of brain displacement and distortion. However, because the anatomic relationship of the electrodes is not as straightforward as that of subdural arrays, a thorough understanding of intracranial anatomy is essential with the use of such grids, underscoring the importance of careful stereotactic planning and a meticulous mapping technique.
Future resection of the EZ can be more demanding following the placement of depth electrodes compared with subdural arrays because there is no craniotomy and the active electrode contacts are not directly visible from the surface of the brain. At our institution, we prefer to remove the electrodes in a separate procedure prior to performing the subsequent craniotomy for removal of the seizure focus, although this task can be done together in a single procedure. With the use of imaging software to contour out the electrode contacts located within the epileptogenic and eloquent areas, a detailed surgical plan can be devised to aid in subsequent tissue resection.
Despite the fact that multiple electrodes are passed through brain tissue, depth electrodes seem to be associated with a relatively low complication rate of 1–5% (Fernández et al., 1997; Guenot et al., 2001; Cossu et al., 2005, 2006; Bekelis et al., 2013). Complications include hemorrhage (usually intracerebral, although epidural and subdural hematomas have been described), infection, transient neurologic deficit, and electrode malfunction. Careful avoidance of vascular structures is essential to safely place depth electrodes. In addition, because the procedure is minimally invasive without direct visualization of the brain surface, intraoperative complications can be difficult to address.
Combined subdural arrays and depth electrodes
It is possible to combine techniques and use both depth electrodes and subdural arrays (Fig. 3). For patients with tumors who have previously undergone surgery, combining techniques affords the surgeon the advantages of subdural arrays such as direct visualization of the tissue and increased ease with subsequent resection of the seizure focus in addition to the benefits of a 3D grid, including coverage of a large volume of brain tissue deep to the surface, and a higher degree of accuracy with respect to the location of the electrode contact. Ideally, the depth electrodes are placed prior to the craniotomy from points outside the planned craniotomy. It is also possible to perform the craniotomy first and then implant the depth electrodes through the grid, although this method may be associated with less stereotactic precision.
Illustrative Case Series
We retrospectively evaluated six patients with intractable epilepsy following resection of an intracranial lesion using subdural grids, with or without 3D-SEEG grids, to identify the epileptogenic zone and determine its location with respect to perilesional and eloquent tissue. This study was reviewed and approved by the University Hospitals Case Medical Center Institutional Review Board . All patients underwent prospective MRI examinations using a 3T magnetic resonance scanner (Siemens, Erlangen, Germany) and interictal positron emission tomography/computed tomography scans (Brilliance iCT, 256-slice; Philips, Amsterdam, The Netherlands). When available, diffusion tensor imaging sequences were performed on a 3T MRI (gradient directions, 30; Siemens). Depth electrode targets and trajectories were planned using imaging software (iPlan Stereotaxy 3.0; BrainLab, Feldkirchen, Germany).
Stereotactic electrode implantation was performed by standard techniques (Talairach & Bancaud, 1973; Munari et al., 1994) using 12-contact depth electrodes (Integra Epilepsy) secured with titanium anchor bolts (Depth Placement Kits and Accessories; Ad-Tech Medical Racine, WI, U.S.A.). The 3D-SEEG implantation consisted of the placement of 9–15 parallel electrodes spaced approximately 1 cm apart in a rectangular pattern (Fig. 2). The 3D grid encompassed both the suspected epileptogenic zone and the closely related eloquent territory. The volume of tissue included in the grid was made to be large enough to include normal brain unaffected by seizure spread in order to accurately delineate the margins of the epileptiform activity. Patients were monitored for 7–14 days to precisely localize the epileptogenic zone. Stimulation was performed when appropriate to map cortical and subcortical eloquent regions. Electrode locations were coregistered for frameless stereotaxy during subsequent resection of the seizure focus.
We used subdural grids and/or intracranial depth electrodes arranged in a 3D-SEEG grid pattern to evaluate six patients (two women, four men) who developed epilepsy months to years (mean 12.3 years) after resection of a primary brain tumor (n = 2), metastatic brain tumor (n = 2), or cavernous malformation (n = 2). Three patients were evaluated with subdural grids: two with a 3D-SEEG grid exclusively, and one with a combination of grids and depth electrodes. In all cases, the irritative and seizure-onset zone was located within a single, restricted area in the abnormal tissue adjacent to the surgical cavity. Subsequent focused resection led to improvement in seizures in all patients at 1 year mean follow-up, with five of six proceeding to Engel class I. In all cases, pathology of the resected tissue did not demonstrate recurrence of tumor but only reactive change with scarring and astrocytosis. Based on this experience, we believe that epilepsy that occurs after tumor resection is often very focal, arising from a single, restricted area and will respond to a limited resection, providing that the epileptogenic zone can be precisely identified using the invasive techniques described.
Patients with intracranial mass lesions are at increased risk of focal, intractable epilepsy, which may develop even after resection. Invasive monitoring via subdural arrays and depth electrodes can play a role in identifying both the EZ and eloquent tissue, allowing for subsequent tissue resection to achieve seizure freedom.
None of the authors has any conflict of interest to disclose. The authors confirm that they have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.