Purpose: Stereo-electroencephalography (SEEG) enables precise recordings from deep cortical structures, multiple noncontiguous lobes, as well as bilateral explorations while avoiding large craniotomies. Despite a long reported successful record, its application in the United States has not been widely adopted. We report on our initial experience with the SEEG methodology in the extraoperative mapping of refractory focal epilepsy in patients who were not considered optimal surgical candidates for other methods of invasive monitoring. We focused on the applied surgical technique and its utility and efficacy in this subgroup of patients.
Methods: Between March 2009 and May 2011, 100 patients with the diagnosis of medically refractory focal epilepsy who were not considered optimal candidates for subdural grids and strips placement underwent SEEG implantation at Cleveland Clinic Epilepsy Center. Demographics, noninvasive clinical data, number and location of implanted electrodes, electrophysiologic localization of the epileptic zone, complications, and short-term seizure outcome after resection were prospectively collected and analyzed.
Key Findings: Mean age was 32 years (range 5–68 years); 54 were male and 46 female. The mean follow-up after resection was 15 months. In total, 1,310 electrodes were implanted. Analyses of the SEEG recordings resulted in the electrographic localization of the epileptogenic focus in 96 patients. In the group of 75 patients who underwent resection, only 53 had at least 12 months follow-up. From this group, 33 patients (62.3%) were seizure-free at the end of the follow-up period. The presence of abnormal pathologic finding was strongly associated with postoperative seizure control (p = 0.005). The risk of hemorrhagic complications per electrode was 0.2%.
Significance: In patients who are not considered to be ideal candidates for subdural grids and strips implantation, the SEEG methodology is a safe, useful and reliable alternative option for invasive monitoring in patients with refractory focal epilepsy, providing an additional mean for seizure localization and control in a “difficult to localize” subgroup of patients.
The two major goals governing the presurgical evaluation of patients with medically intractable focal epilepsy are the localization/definition of the spatial extent of the epileptic focus and its relation with potentially testable functional cortex/structures (Rosenow & Lüders, 2001). To achieve these goals, various noninvasive tools are available, including analysis of seizure semiology, prolonged video-scalp electroencephalographic (video-EEG) recordings, magnetoencephalography (MEG), magnetic resonance imaging (MRI), other neuroimaging modalities (functional MRI [fMRI], ictal single photon emission computed tomography [SPECT], positron emission tomography [PET]), and neuropsychological testing (Winkler et al., 1999; Rosenow & Lüders, 2001). Not surprisingly, the success in localizing and completely resecting the epileptic focus has been shown to be the most important predictor of seizure freedom in patients who undergo surgical resections (Jeha et al., 2006, 2007; Lüders et al., 2006; Najm et al., 2006). Nevertheless, when noninvasive data are insufficient to localize and/or define the epileptic zone (EZ) and its overlap/proximity to eloquent cortex, invasive monitoring may be indicated.
Subdural grids and strips are the most common invasive method used in the United States (Risinger & Gumnit, 1995; Najm et al., 2002; Widdess-Walsh et al., 2007). Despite the high spatial resolution provided by the subdural methodology, which allows for accurate mapping of superficial cortical areas, relatively deep epileptic foci cannot be sampled with adequate spatial and temporal resolution. In addition, subdural grids require relative large craniotomies and are, in general, limited to exploration of one hemisphere. In the face of these relative limitations, we explored alternative/complementary methods for invasive monitoring, revisiting the concepts and the techniques of the stereo-electroencephalography (SEEG) methodology (Bancaud et al., 1970; Kahane et al., 2006; Cossu et al., 2008; Devaux et al., 2008).
We report on our preliminary clinical experience with the SEEG methodology in the extraoperative mapping of refractory focal epilepsy, focusing on the surgical technique, efficacy, and the method’s utility in the treatment of medically refractory focal epilepsy in patients who were not considered optimal candidates for subdural evaluation.
We studied all patients with the diagnosis of medically refractory focal epilepsy who underwent SEEG implantation at the Cleveland Clinic Epilepsy Center between March 2009 and April 2011. Procedures related to the SEEG methodology, which included implantation/removal of electrodes and SEEG-guided resections, were performed by a single surgeon (JGM). Functional mapping was performed at the end of the monitoring period. Bipolar stimulation was applied in selected electrode contacts. Stimulation pulses were delivered in biphasic mode, 30-s trains, at a frequency of 25 Hz, with 0.3-msec of pulse width and current intensity varying from 1–10 mA.
Data regarding age, gender, history, seizure semiology, noninvasive EEG, neuropsychology testing, PET, ictal SPECT, MEG, number and location of implanted electrodes, electrophysiologic localization of the EZ, complications, and seizure outcome after resection were prospectively collected and analyzed. Postoperative clinical data were collected through patients’ interviews, during regular scheduled clinic visits, or by telephone contacts. For the seizure outcome analysis, only patients with a minimum follow-up of 12 months following resection were included. Postoperative seizure outcome was classified in a dichotomous manner as seizure-free or exhibiting persistent seizures. The occurrence of auras after surgery was counted as recurrent seizures. Acute postoperative seizures (up to 1 week after surgical resection) were not counted as evidence of recurrent epilepsy (Jehi et al., 2010). All adverse events within a period of 30 days following SEEG implantations were counted as complications. This study was approved by the Cleveland Clinic Institutional Review Board. All surgeries were part of standard patient care and no procedures were performed for research purposes.
Selection criteria for SEEG implantation
In addition to the general selection criteria for invasive extraoperative monitoring, as absence of precise anatomic delineation of the EZ and its anatomic relation with functional cortical areas (Jayakar, 1999; Najm et al., 2002; Cossu et al., 2005; Widdess-Walsh et al., 2007), additional specific indications were used to choose SEEG (vs. other methods of invasive monitoring such as subdural grids/strips) as the recommended method of invasive monitoring. These criteria included:
1 The possibility of a deep-seated or difficult-to-cover location of the EZ in areas such as the mesial structures of the temporal lobe, opercular areas, cingulate gyrus, interhemispheric regions, posterior orbitofrontal areas, insula, and depths of sulci.
2 A failure of a previous subdural invasive study to clearly outline the exact location of the seizure onset zone.
3 The need for extensive bihemispheric explorations.
4 Presurgical evaluation suggestive of a functional network involvement (e.g., limbic system) in the setting of normal MRI.
SEEG implantation method
After an anatomofunctional localizing hypothesis was formulated, a tailored implantation strategy was planned, with the goal of confirming or rejecting the preimplantation hypothesis. The exploration was focused to sample the anatomic lesion (if present), the more likely structure(s) of ictal onset, and the possible pathway(s) of propagation of seizures within a functional network (Figs. 1 and 2). The desired targets were reached using commercially available depth electrodes (AdTech, Racine, WI, U.S.A.; Integra, Plainsboro, NJ, U.S.A.) implanted using conventional stereotactic technique through 2.5-mm drill holes. (Details regarding the implantation technique are explained in the Data S1).
We assessed the statistical correlation between seizure outcome, complications, and categorical variables using univariate and multivariate analyses. For all analyses, significance was set at a p-value of 0.05.
One hundred patients with refractory focal epilepsy who underwent extraoperative invasive monitoring using the SEEG methodology were included in the study. Most of the patients had at least two sessions of video-EEG monitoring. The mean age was 32 years (range 5–68 years); 54 of patients were male and 46 female. The group included 17 pediatric patients (ranging in age from 5–12 years, mean age 8 years). All the patients had the diagnosis of refractory focal epilepsy, with an average failure of five antiepileptic drugs per patient. The MRI in 61 patients was either normal or showed extensive bilateral abnormalities. Focal MRI abnormalities were found in 39 patients: these included subtle blurring of gray–white matter transition with increased T2 and fluid-attenuated inversion recovery (FLAIR) signals, decreased volume of the hippocampal formation associated with increased T2 and FLAIR signals, complex and bilateral congenital cortical abnormalities, previous surgical resections, bilateral tubers, and periventricular nodules. Detailed demographic information, along with seizure classification, severity of seizures, and noninvasive EEG data are depicted in Table 1. Twenty-seven patients had previous resection or previous invasive monitoring procedures with subdural grids and strips (rarely in combination with depth electrodes), which resulted in a nonlocalizable ictal pattern or no improvement in seizure outcome after resection.
In 65 patients, the choice of SEEG over subdural grids was influenced by the presurgical hypothesis of the presence of the EZ in areas such as the mesial structures of the temporal lobe, cingulate gyrus, posterior orbitofrontal areas and insula, or focal epilepsy arising from a structure within a functional system (e.g., limbic network) in the setting of normal MRI. Twenty-seven patients underwent previous resections or prior invasive monitoring using subdural grids and strips, which failed to conclusively localize the EZ or led to surgical resections with no improvement in seizure outcome. In addition, bilateral SEEG electrodes were implanted in 40 patients for seizure lateralization and localization.
On average, 13 depth electrodes were implanted per patient (range 7–22 electrodes). Implantations were bilateral in 40 patients), right hemispheric implantations were performed in 33 patients and left hemispheric placements in 27 patients. Implantations involved at least two lobes (at times noncontiguous), which included a combination of frontal, temporal, parietal, occipital, and/or insular lobes explorations. In total, 1,310 electrodes were implanted. Most electrodes (n = 1,269) were inserted in orthogonal orientation in relation to the sagittal plane, with the remaining 41 electrodes implanted in oblique orientation (targeting the insula cortex, posterior orbitofrontal cortex, superior frontal gyrus, or superior parietal lobule).
Electrographic localization of seizure onset, surgical resections, and postresection seizure outcome
Analyses of the SEEG recordings led to potential electrographic localization of the epileptogenic focus in the majority of patients (n = 96). In four patients, the epileptic focus could not be electrographically identified. From this group, two patients had previously failed surgery/invasive monitoring with subdural mapping. In one patient, a second implantation with SEEG electrodes was recommended but the patient declined. Diffuse ictal onset was recorded in the fourth patient, possibly indicating an erroneous preimplantation hypothesis or an underlying multifocal epilepsy.
Seventy-five patients underwent surgical resection following SEEG evaluation. Surgical resections were not performed in 25 patients for the following reasons: (1) nonelectrographic localization of seizures (four patients); (2) involvement of eloquent cortical areas within the potential EZ (three patients); (3) improvement of seizures after SEEG implantation without resection in two patients; and (4) bilateral and/or noncontiguous areas of seizure onset in 16 patients.
From the group of 75 patients who underwent surgical resections, 53 patients met the inclusion criteria of minimum postresection follow-up of 12 months. In this subset of patients, mean follow-up after resection was 17 months (range 12–36 months). Twenty-two of these patients (41.5%) had unilateral MRI abnormalities and 31 patients (58.5%) had either normal MRI studies (28 patients) or bilateral abnormalities (3 patients). Fifteen patients (28%) failed previous resective surgery, or a localization of ictal-onset zone was not made during a prior subdural grid evaluation. Thirty patients (56.6%) were operated on the right side, and 23 (43.4%) underwent left hemispheric resections. Temporal lobe resections were performed in 22 patients (41.5%) and extratemporal resections in 31 patients (58.5%). In the extratemporal resection group, 15 patients underwent frontal resections (28.5%) and 7 underwent parietal resections (13.5%). Multilobar resections were performed in nine patients (17%), and included frontotemporal resections in five patients, frontoinsular resections in two patients, and insular-perisylvian resections in two patients.
Pathologic evaluation of the resected tissue did not revealed any specific abnormalities in seven patients. No specimen was sent for analysis in one patient. Pathologic abnormalities included type 1 focal cortical dysplasia (31 patients), type 2 (seven patients), hippocampal sclerosis (three patients), encephalomalacia with gliosis (two patients), and cortical dysplasia associated with hippocampal sclerosis or meningoangiomatosis in two patients.
In this group of 53 patients, 33 patients (62.3%) were seizure-free at the end of the follow-up period. Statistical analyses of the possible positive or negative predictors of seizure-free outcome in patients who underwent surgical resection are presented in Table 2. The absence of clear pathologic changes upon histopathologic examination was the only analyzed variable statistically associated with an unfavorable postoperative seizure outcome (p = 0.005). In this group, the presence of pathologic findings—most commonly mild forms of cortical dysplasia (type 1) or hippocampal sclerosis—was strongly associated with postoperative seizure control. Bilateral implantations (probably indicating an insufficient preimplantation localization hypothesis) and younger age at surgery showed a trend toward a worse seizure outcome, without reaching statistical significance. Other variables as sex, previous surgeries, type of resection, duration of epilepsy, and side of resection did not predict seizure outcome following surgical resection guided by SEEG data. Of interest, the preoperative MRI findings groups (normal, abnormal with unilateral lesion, and abnormal with bilateral lesions) did not show statistical difference for postoperative seizure-free outcome (p = 0.2). Similarly, extratemporal or multilobar resections were not statistically associated with worse seizure outcome when compared with the temporal resection group (p = 0.7).
Table 2. Statistical association between each predictor variable and seizure freedom
Seizure free patients (n)
Side of implantation
Side of resection
Type of resection
Complications related to the SEEG procedures were observed in three patients: one patient had an asymptomatic subdural hematoma and two developed intraparenchymal hemorrhages. There were no angiographic complications. All three complications were treated conservatively. Only one hemorrhage corresponding to a small cerebral contusion located in the leg motor cortex was symptomatic. The patient developed a foot paresis, which completely recovered 2 weeks later. All hemorrhages were located in the frontal lobe following placement of the following: (1) an oblique electrode targeting the insula, resulting in a subdural hematoma with minimal local mass effect centered within the left frontal region; (2) a frontal electrode targeting the orbitofrontal cortex resulting in a hemorrhagic hematoma in the lateral orbital gyrus, at the electrode entry point; and (3) another frontal electrode, targeting the posterior mesial frontal cortex, resulting in a small hemorrhagic hematoma at the target point. No mortality or major morbidity occurred. No other complications were observed in this series. Given the total number of implanted electrodes (n = 1,310), the calculated risk of complications per electrode in this series was 0.2%.
For >30 years, subdural mapping techniques have been the hallmark of extraoperative invasive monitoring techniques for refractory epilepsy in the United States (Engel et al., 1990; Silberbusch et al., 1998; Najm et al., 2002; Widdess-Walsh et al., 2007). Despite its efficacy and spatial accuracy in mapping the superficial cortex, invasive monitoring using the subdural methodology has limitations. Disadvantages may include a relatively high surgical morbidity, inherent limitations in accessing deep or bilateral cortical structures, and in patients with a normal MRI, electroclinical features suggestive of functional network involvement (Hamer et al., 2002; Onal et al., 2003; Simon et al., 2003; Johnston et al., 2006; Widdess-Walsh et al., 2007). To overcome these limitations, the SEEG methodology was applied in a highly selected group of patients who were not considered optimal candidates for subdural electrode placement.
Although formal volumetric studies were not performed, it is our impression that the SEEG-guided resections were consistently limited and smaller when comparing the current series with our historical subdural-guided resection series (Najm et al., 2002), where resections tended to be lobar, or near lobar. This preliminary impression could suggest a higher specificity of the SEEG method in mapping the EZ. In the current series, most of the performed resections corresponded to sublobar resections (frontal pole, temporal pole, superior frontal gyrus, and so on). As illustrated in Fig. 2, seizure onset was recorded in contacts located in the vicinity of the MRI-visible lesion (electrode L), with early spread of the pathologic electrical activity to the posterior cingulate electrodes (located in the isthmus of cingulate gyrus, electrode Y), and anteriorly (in the subcentral cingulate gyrus, electrode X). A restricted resection of the posterior cingulate area, including the lesion and the adjacent cingulate gyrus, was performed. Based on our previous experience, a subdural exploration would likely have resulted in diffuse distributed ictal recordings, guiding a larger resection, or even in nonlocalizable patterns, preventing further surgical interventions. In addition, in the analysis of other important variables as efficacy and safety, the SEEG method also revealed adequate seizure outcome results with minimal morbidity.
Our results show that the use of SEEG methodology is both safe and effective in patients with difficult-to-localize medically intractable focal epilepsies. From a safety perspective, the described implantation technique, which partially departures from the more traditional technique used for SEEG implantation (that mainly uses the double Talairach grid and frame), constitutes a safe option as compared to the “traditional” method of implantation described by European centers (Guenot et al., 2001; Cossu et al., 2005, 2008). In our series, the total complication rate was 3% (approximately 0.2% per implanted electrode), with no mortality or major morbidity associated with persistent neurologic deficits. Cossu et al. reported a morbidity rate of 5.6%, with severe permanent deficits from intracerebral hemorrhage in two patients (1%). All three complications in our series were hemorrhagic, which has been reported in several studies to be the most common complication of depth electrodes placement (Bancaud et al., 1970; Guenot et al., 2001; Hamer et al., 2002; Cossu et al., 2005). Other published series reporting complications across invasive monitoring procedures (subdural grids and depth electrodes) have reported rates ranging from 0% to 26% (Burneo et al., 2006; Widdess-Walsh et al., 2007; Cossu et al., 2008; Wong et al., 2009). Of interest, subdural monitoring with grid electrodes has historically been perceived to have low permanent morbidity (0–3%) compared with depth electrodes (3–6%), since there is no intraparenchymal penetration (Rydenhag & Silander, 2001). Although it is difficult to compare morbidity rates between subdural grids and SEEG due to the variability in patient selection, different institutions, and variable number of implanted electrodes, it is our preliminary impression that the SEEG method provides at least a similar degree of safety when compared to subdural grids or strips.
From an effectiveness perspective, our results show that the SEEG approach is highly effective in electrophysiologically localizing the EZ. From the studied group of 100 patients, the area of seizure onset could not be localized in only four patients. The high efficacy of the SEEG method in defining the EZ can be explained by the use of rigid selection criteria prior to SEEG procedures. Cossu et al. (2005) reported their results in 211 patients who underwent SEEG procedures for further localization, in whom identification of the seizure onset region was accomplished in 204 patients (96.7%), results quite similar to this series. But despite the promising percentage demonstrated in our studied group, a fraction of these patients underwent resective surgery and achieved seizure freedom.
Seizure freedom following resective surgeries in patients with normal preoperative MRI has been significantly worse, with seizure-free rates as low as 18% in some series (Scott et al., 1999; Siegel et al., 2001; Chapman et al., 2005; Jeha et al., 2007; Jehi et al., 2009). Similarly, seizure outcomes in patients with extratemporal lobe epilepsy (and in particular frontal lobe epilepsy) are considerably worse in comparison to those with temporal lobe epilepsy (reported success rates could vary from 13% to 80% of patients with frontal lobe epilepsy) (Rasmussen, 1991; Talairach et al., 1992; Williamson & Jobst, 2000; Zaatreh et al., 2002). Of interest, in this series, when analyzing possible predictors variables related to seizure outcome, neither normal MRI nor extratemporal lobe resections were statistically associated with unfavorable seizure outcome. This may be due in part to a rigorous patient selection process during which only patients with a clear and precise preimplantation hypothesis were considered for SEEG exploration. Another possible explanation may be related to the possibility of multilobar or bihemispheric sampling using SEEG (that is technically difficult with subdural grids). In addition, one third of our patients had previously undergone subdural implantations, with acquired information that could have contributed favorably to the formulation of a better anatomoelectrophysiologic hypothesis, leading to a more successful SEEG implantation. Nonetheless, it is important to mention that our results are (1) based on a relatively short postoperative seizure outcome period and (2) report on a relatively small number of patients. Nonetheless, these impressions need to be further tested and validated in a larger patient population with longer follow-up periods.
Although the follow-up period of only 1 year can be considered relatively short, the current data strongly suggest that SEEG evaluation provides adequate localization of the current EZ in a sizable number of patients with highly complex epilepsies. Seventy-five patients underwent surgical resection guided by SEEG electrodes. From this group, 53 patients had a minimum postresection follow-up of 12 months and a mean follow-up period of 17 months. At the end of the follow-up period, 62.3% of the patients remained seizure-free. The highly complex features of our population can be demonstrated by the fact that almost 60% of these patients had a normal MRI or exhibited bilateral MRI abnormalities, approximately one third (28%) of them had previous resections or had failed prior subdural grid implantation, and the majority of these patients (almost 60%) were found to have an extratemporal localization and underwent extratemporal lobar or multilobar resections, all variables frequently associated with unfavorable seizure outcome (Jeha et al., 2007; Jehi et al., 2009). Despite the challenging clinical scenario, seizure outcome results were promising.
In summary, our studied population corresponded to a highly difficult-to-localize group of patients who were not considered to be ideal candidates for subdural grid evaluation due to the complexity of their epilepsy syndrome: (1) almost two thirds (65%) of our patients had a deep-seated EZ with diffuse/nonlocalizing seizure onset on the scalp EEG, suggesting that subdural implantations would have been technically challenging and probably inaccurate; (2) approximately 30% had failed subdural grid evaluation due to the presence of “non-localizable” ictal pattern recordings, indicating that the onset was probably located outside the area of coverage, in deep cortical areas, and/or in locations that are inaccessible for subdural electrodes. In addition, 40% of the studied group needed bihemispheric explorations, which would have been technically challenging with the use of subdural electrodes. We concluded that the SEEG method is an adequate alternative invasive monitoring technique that yields localizing the following information. (1) In patients with deep-seated or difficult to cover region(s) such as depths of sulci, mesial structures of the temporal lobe, opercular regions, cingulate gyrus, interhemispheric regions, posterior orbitofrontal cortex, and insula. (2) Following a failure of a previous subdural invasive study to clearly outline the exact location of the seizure onset zone. (3) In patients with multiple multilobar or bihemispheric lesions with a need for extensive bihemispheric explorations. (4) In patients in whom the presurgical evaluation showed findings consistent with an anatomofunctional network involvement (e.g., limbic system) in the setting of a normal MRI. In performing SEEG in this highly selected group, we were able to overcome the relative limitations related to the current standard method of invasive monitoring, offering to these challenging patients, an additional opportunity for seizure freedom, which would not likely be possible with subdural monitoring. Our early outcome results compare favorably with previously published outcome data. To validate the current data and confirm our conclusions, a long-term longitudinal study will be necessary.
The author would like to thank the Congress of Neurological Surgeons for supporting Jorge Gonzalez-Martinez through the Cushing Fellowship Award.
The authors have no conflict of interest. We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.