Epilepsia, 52(3):458–466, 2011 doi: 10.1111/j.1528-1167.2010.02910.x
FULL-LENGTH ORIGINAL RESEARCH
The combination of subdural and depth electrodes for intracranial EEG investigation of suspected insular (perisylvian) epilepsy *Werner Surbeck, *Alain Bouthillier, *Alexander G. Weil, yLouis Crevier, zLionel Carmant, zAnne Lortie, zPhilippe Major, and xDang Khoa Nguyen *Service de Neurochirurgie, Hoˆpital Notre-Dame du CHUM, Universite´ de Montre´al, Quebec, Canada; yService de Neurochirurgie, Hoˆpital Sainte-Justine, Universite´ de Montre´al, Quebec, Canada; zService de Neurologie, Hoˆpital Sainte-Justine, Universite´ de Montre´al, Quebec, Canada; and xService de Neurologie, Hoˆpital Notre-Dame du CHUM, Universite´ de Montre´al, Quebec, Canada
SUMMARY Purpose: We present two methods of implantation for the investigation of suspected insular and perisylvian epilepsy that combine depth and subdural electrodes to capitalize on the advantages of each technique. Methods: Retrospective study of all intracranial EEG studies that included insular electrodes from 2004– 2010. Patients were divided according to the implantation scheme. The first method (type 1) consisted of a craniotomy, insertion of insular electrodes after microdissection of the sylvian fissure, orthogonal implantation of mesiotemporal structures with neuronavigation, and coverage of the adjacent lobes with subdural electrodes. The second method (type 2) consisted of magnetic resonance imaging (MRI)– stereotactic frame-guided depth electrode implantation into insula and hippocampus using sagittal axes, and insertion of subdural electrodes through burr holes to cover the adjacent lobes. The combined
Recent observations from our group and others have shown that the insula may generate a variety of ictal symptoms (such as visceral, motor, and somatosensory) that may mimic temporal, frontal, or parietal lobe seizures (Isnard et al., 2004; Ryvlin, 2006; Nguyen et al., 2009). Furthermore, insular seizures may coexist with seizures from other lobes (Isnard et al., 2000; Nguyen et al., 2009). Failure to recognize and identify insular seizures may explain why some patients continue to experience seizures following epilepsy surgery (Isnard et al., 2000; Aghakhani et al., 2004). Because insular resections cannot be based on intraoperaAccepted October 20, 2010; Early View publication January 4, 2011. Address correspondence to Dang K. Nguyen, MD, FRCPC, Service de Neurologie, CHUM Hopital Notre-Dame, 1560 Sherbrooke Street East, Montreal, QC, H2L 4M1, Canada. E-mail:
[email protected] Wiley Periodicals, Inc. ª 2011 International League Against Epilepsy
implantations were developed and performed by one neurosurgeon (AB). Key Findings: Nineteen patients had an intracranial study that sampled the insula, among other regions. Sixteen patients were implanted using the first method, which allowed a mean of 4, 5, 20, 15, and 42 contacts per patient to be positioned into/over the insular, mesial temporal, neocortical temporal, parietal, and frontal areas, respectively. The second method (three patients) allowed a mean of 8, 7, 16, 6, and 9 contacts per patient to sample the same areas, respectively. The four patients in whom transient neurologic deficits occurred were investigated with use of type 1 implantation. Significance: Combined depth and subdural electrodes can be used safely to investigate complex insular/perisylvian refractory epilepsy. Choice of implantation scheme should be individualized according to presurgical data and the need for functional localization. KEY WORDS: Epilepsy, Insula, Perisylvian, Invasive, Intracranial, Electroencephalography.
tive electrocorticographic spikes alone (Silfvenius et al., 1964), confirmation by intracerebral recordings is necessary when insular seizures are suspected, especially if there is no abnormality on high-resolution magnetic resonance imaging (MRI). The Lyon group has mainly used transopercular electrodes implanted perpendicular to the sagittal plane using the classical approach described by Talairach and Bancaud (1973). More recently, the Grenoble group reported their experience with depth electrodes stereotactically implanted within the insular cortex via an oblique trajectory, which obviated the need to pass through the dense vascular wall covering the insula and allowed larger coverage of the insular cortex. Although both methods allow fairly good sampling of the insular and mesial temporal structures, coverage of inferior and lateral neocortices as well as the interhemispheric area is less than what could be achieved by subdural strip and grid electrodes. Less
458
459 Invasive EEG Investigations of Insular Epilepsies extensive cortical coverage may compromise the ability to detect the epileptogenic zone and limit functional mapping. In this study, we report the safety and usefulness of two other methods of sampling the perisylvian region, including the insula, used at our institution over the last 6 years.
Patients and Methods Patients For this study, we selected all patients who underwent an intracranial study that required sampling of the insula among other regions between October 2004 and February 2010. Cortical structures to be explored by intracranial electrodes were chosen on the basis of findings from a typical presurgical evaluation, which included a detailed questionnaire and physical examination, a neuropsychological evaluation, high-resolution cerebral MRI, long-term video–electroencephalography (EEG) monitoring, ictal single-photon emission computed tomography (SPECT), and 18F-fluorodeoxyglucose positron emission tomography (PET). Patients evaluated in the last 3 years also benefited from magnetoencephalographic and combined EEG/functional MRI (fMRI) studies. Consensus on areas to cover was obtained during an epilepsy surgery conference after review of all available data. In general, the insula was sampled in the presence of (1) nonlesional temporal lobe–like epilepsy, (2) nonlesional parietal lobe–like epilepsy, (3) atypical nonlesional frontal lobe–like epilepsy (occasional early occurrence of somatosensory symptoms during diurnal seizures, coexisting temporal interictal spikes, insular activation on ictal SPECT or EEG-fMRI), and (4) atypical temporal lobe–like epilepsy (early occurrence of somatosensory or motor symptoms, insular activation on ictal SPECT or EEG-fMRI, nonspecific millimetric signal change in the insula on MRI) even in the presence of hippocampal atrophy or sclerosis. Electrode implantation Type 1: open microdissection of the sylvian fissure The first method consisted of a unilateral frontotemporoparietal craniotomy (sometimes broken into two smaller craniotomies), insertion of insular depth electrodes by microdissection of the sylvian fissure, orthogonal implantation of depth electrodes into mesial temporal structures using frameless stereotaxy, and coverage of the three adjacent lobes with subdural strip and grid electrodes (Fig. 1). The sylvian fissure was identified by anatomic landmarks and neuronavigation. It was dissected using microsurgical techniques, and care was taken to spare most of the veins crossing the sylvian fissure. The fissure was opened to expose the insular cortical areas of interest (anterior and/or posterior), as determined by the noninvasive preoperative investigation. Safe areas between M2 branches of the middle cerebral artery were then chosen for electrode
Figure 1. Schematic drawing of type 1 implantation: Accomplishement of a unilateral frontotemporoparietal craniotomy. Implantation of one to three insular electrodes under direct vision after microdissection of the sylvian fissure (blue-marked depth electrodes). Orthogonal implantation of three additional depth electrodes into mesial temporal structures using frameless stereotaxy (violet-marked depth electrodes) and coverage of the three adjacent lobes with subdural strip and grid electrodes. Epilepsia ILAE
placement. This was done after a small incision of the pia matter with a microblade. One to three insular electrodes were then implanted in each patient under direct vision (Fig. 2C4). The insular depth electrodes used (Spencer depth electrodes; Ad-Tech Medical Instrument Corporation, Racine, WI, U.S.A.) had a diameter of 1.1 mm, and carried four contacts along their length (each of 2.3 mm in length and spaced 5 mm apart from center to center). Two contacts per electrode were entered into the insular cortex. Each depth electrode was later sutured to subdural electrodes placed on the lateral surface of the hemisphere, to the dura matter, and to the skin surface. In most cases, three additional depth electrodes were then implanted perpendicular to the temporal neocortex through the middle temporal gyrus into the temporomesial structures using frameless stereotaxy (4-contact Spencer depth electrodes; 1.1 mm diameter, 2.3 mm length, 10 mm spacing): one directed into the amygdala, one into the anterior hippocampus, and the last into the posterior hippocampus. Taking advantage of the craniotomy, subdural strip and/or grid electrodes were positioned on the surface of the three remaining perisylvian lobes, depending on the preoperative investigation (Fig. 2). The inferior surfaces of the frontal and temporal lobes were Epilepsia, 52(3):458–466, 2011 doi: 10.1111/j.1528-1167.2010.02910.x
460 W. Surbeck et al.
A1
B1
C1
A2
B2
C2
A3
B3
C3
C4
Figure 2. Examples of type 1 implantations: (A) Patient 2. Nonlesional refractory case with early somatosensory aura mainly requiring parietoinsular coverage; (A1) Intraoperative view; (A2) 3D visualization of intracranial electrode arrangement (lateral view, insular depth electrodes not visible); (A3) Coronal MRI view of the insular electrode. (B) Patient 4. Nonlesional refractory case requiring mainly frontal and insular coverage; (B1) Intraoperative view; (B2) 3D visualization of intracranial electrode arrangement (anterolateral view, insular depth electrode not visible); (B3) Processed sagittal MRI view of the interhemispheric strip electrodes. (C) Patient 14. Nonlesional refractory case with early dysphasic manifestations requiring extensive temporal and insular coverage; (C1) Intraoperative view; (C2) 3D visualization of intracranial electrode arrangement (lateral view; insular and mesiotemporal depth electrodes not visible); (C3) Axial MRI showing the middle and posterior insular electrodes; (C4) Insertion of depth electrodes in the insular cortex under direct vision after microdissection of the sylvian fissure. Epilepsia ILAE
covered in most cases, and putative functional areas were covered for functional mapping. All electrode leads were tunnelled through the dura and skin and purse-string sutures were placed around their exit site on the skin to further minimize the potential for cerebrospinal fluid (CSF) leak. Duroplasty was sometimes necessary for adequate closure of the dural opening. Bone flap was not replaced until electrode removal and, in cases where epileptogenic zone was identified, resective procedure. Type 2: combined Yale-Grenoble stereotactic implantation The second method consisted of MRI-stereotactic frameguided depth electrode implantation sagittally into the insula and the hippocampus through small burr holes using Epilepsia, 52(3):458–466, 2011 doi: 10.1111/j.1528-1167.2010.02910.x
leads with 10 contacts and subsequent insertion of subdural strip electrodes through extended burr holes to cover the adjacent lobes (Fig. 3). After induction of general anesthesia and fixation of the Cosman-Roberts-Wells (CRW) stereotactic frame (Integra Radionics, Burlington, MA, U.S.A.), a computed tomography (CT) scan was obtained. Using the Treon CRW frame compatible navigation software (Medtronic, Minneapolis, MN, U.S.A.), a fusion between the CT images and a threedimensional (3D) magnetization-prepared rapid acquisition with gradient echo (MPRAGE) MRI injected with gadolinium was done. Electrode paths were chosen meticulously on the 3D MRI to avoid injury to vascular structures using the mentioned CRW frame–compatible software on the Treon
461 Invasive EEG Investigations of Insular Epilepsies
Figure 3. Schematic drawing of type 2 implantation: depth electrode implantation sagittally into the insula using an anterior transfrontal and a posterior transparietal trajectory (blue-marked depth electrodes) and the hippocampus using an occipitotemporal trajectory parallel to the hippocampal long axis (violetmarked depth electrode). Subsequent subdural strip electrodes insertion through a temporal burr hole. Epilepsia ILAE
A
D
computer. For each electrode, we obtained the settings for the CRW frame (anteroposterior, lateral, vertical, arc, and ring), the length of the intracranial trajectory, and skull thickness. Initially, we used burr holes made with a compressed air motor for skull perforation. More recent cases were done with a twist drill. A stopper was used adjusted to the skull thickness (measured by the navigation software) to avoid dura penetration. Once the skull was perforated, the dura was opened and a small incision of the pia matter was made. To avoid electrode deviation, an introducing cylinder of 2-mm diameter was placed, with its obturator, precisely at the entrance of the insula or the hippocampus. This distance was measured using the Treon navigation software. The oburator was removed and the electrode, with a stylet inside, was inserted into the cylinder. Once the target was reached, the stylet was removed from the electrode. For exact electrode placement, indelible ink marks were made on the electrode, taking into account the residual length of the cylinder outside the skull. Once inserted, the introducing cylinder was carefully withdrawn. During this step, care was taken not to remove the electrode. Each electrode was tunnelled through the skin using an epidural needle along the long axis of the skin incision, in order to avoid electrode displacement upon skin closure. Because the hippocampal electrodes sometimes penetrate the ventricular system, with thereby the potential to distort the brain anatomy, these electrodes are implanted after the insular ones. The intrainsular electrodes (10-contact Spencer depth electrodes; 1.1 mm diameter, 2.3 mm length, 10 mm spacing) were implanted using an anterior (transfrontal) approach passing through
B
E
C
F
Figure 4. (A) Insertion of depth electrodes in a sagittal plane with the CRW frame and the Treon navigation software (Medtronic). (B) Sagittal MRI view of the posterior insular electrode (entrance through parietal lobe). (C) Sagittal MRI view of the anterior insular electrode (entrance through the frontal lobe). (D) Coronal MRI view of the anterior insular electrode. (E) Axial MRI view of the amygdalohippocampal electrode. (F) Subdural electrodes inserted through a burr hole. Epilepsia ILAE Epilepsia, 52(3):458–466, 2011 doi: 10.1111/j.1528-1167.2010.02910.x
462 W. Surbeck et al. the middle frontal gyrus and a posterior (transparietal) trajectory passing through the inferior parietal cortex as described by Afif et al., 2008 (Fig. 4A). Because the insula is not a flat but rather a pyramidal structure, choosing a trajectory running mediolaterally targeting the anteroinferiorly located tip of the pyramid (apex insulae) can position the contacts close to the insular surface. To sample the hippocampal formation and amygdala, a parietooccipital approach was utilized with a trajectory parallel to the hippocampal long axis as described by Spencer (1987). The electrodes were fixed to the skin after tunnelization. The remaining perisylvian lobes where finally covered by subdural strip electrodes inserted through extended temporal burr holes (Fig. 4). As mentioned previously, electrode leads are tunnelled through dura and skin and purse-string sutures are placed around their exit site on the skin. For both procedures, great care was taken to minimize the potential for CSF leak and associated infections upon closure. Prophylactic antibiotic treatment during long-term intracranial EEG recordings was not routinely used in our institution. Steroid administration to reduce procedurerelated parenchyma inflammation was utilized postoperatively and withdrawn after 3–4 days. Postimplantation high-resolution MRIs with 1-mm–thick slices were always obtained to determine as precisely as possible the position of the contacts. A 3D representation of the electrodes with respect to the patient’s brain was obtained using Gridview software (Stellate Systems Inc., Montreal, Canada). One hundred twenty-eight channels of simultaneous EEG recordings were available for extraoperative continuous monitoring (Stellate Systems Inc.).
Results During the period extending from October 2004 and February 2010, 19 patients (7 male and 12 female) underwent intracranial EEG studies with sampling of the insula among other regions in our epilepsy service. The combined implantations were developed and performed by a single surgeon (AB). Age at implantation varied between 2 and 43 years of age (median age 25 years). Before the invasive EEG study, 10 patients had temporal lobe–like epilepsy, 2 had parietal lobe–like epilepsy, and 7 had frontal lobe–like epilepsy. The first method (open microdissection of the sylvian fissure) was used in 16 patients and the second (combined Yale-Grenoble stereotactic implantation) in 3 patients. The scheme of implantation and explored areas for each patient is summarized in Tables 1 and 2. Coverage With the type 1 implantation method, a mean of 3.5 (range 1–6) contacts were directly positioned into the insular cortex. Mesial temporal structures were sampled by a mean of 4.5 (range 0–11) contacts per patient. Mean number of contacts sampling the neocortex was 20 (range 0–44) for Epilepsia, 52(3):458–466, 2011 doi: 10.1111/j.1528-1167.2010.02910.x
Table 1. Coverage of perisylvian cortex by type 1 implantation Electrodes
Contacts Temporal
Case Grid Strips Depths Insular Mesial Neocortical Parietal Frontal 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Mean Min. Max.
1 1 1 1 0 0 0 0 2 1 1 3 1 1 1 3 1 0 3
2 5 2 8 11 4 9 9 0 4 8 2 5 7 5 0 5 0 11
3 2 2 1 2 5 5 3 2 4 3 2 5 6 4 5 3.5 1 6
1 3 4 2 4 4 4 3 1 4 6 3 4 6 3 3 3.5 1 6
8 0 2 0 2 7 11 8 0 9 0 0 4 10 8 0 4.5 0 11
36 8 0 6 6 25 33 8 0 37 6 18 36 44 28 28 20 0 44
25 48 0 0 12 0 4 2 32 0 27 6 3 18 12 41 14.5 0 48
15 42 42 98 60 0 24 68 33 22 59 91 27 8 42 36 41.5 0 98
Min., minimum; Max., maximum.
the temporal, 14.5 (range 0–48) for the parietal, and 41.5 (range 0–98) for the frontal lobes. With the type 2 implantation method, 2–4 lead-contacts were positioned directly into the insular cortex with the transfrontal electrode and 5–7 contacts for the transparietal electrode, resulting in a mean of 8 contacts per insula. Mean number of depth contacts sampling the mesiotemporal structures were 6.75 (range 5–10). Mean number of subdural contacts sampling the neocortex was 16.25 (range 6–23) for temporal, 6 (range 0–8) for parietal, and 9 (range 0–20) for frontal neocortices. Usefulness Long-term monitoring was performed extraoperatively over 2–4 weeks. Invasive EEG findings from the first 10 patients have been reported previously in detail (Nguyen et al., 2009). Insular spikes were found in 11 (58%) of 19 patients and insular seizures were recorded in 7 (37%) of 19 patients. Five patients (71%) with insular seizures also had independent seizures from temporal or frontal areas. Type 1 implantation allowed mapping of eloquent language areas in three of the four patients implanted on the dominant hemisphere using this technique. Of the seven patients with demonstrated insular seizures, all but one underwent a tailored surgical resection based on invasive EEG findings. Details of the surgery and outcome for three of them have been reported previously (Malak et al., 2009). Epilepsy surgery consisted of a partial insulectomy in six patients (two on the dominant side). This partial resection of the insular cortex was combined with temporal or frontal tailored corticectomy in three and two patients,
463 Invasive EEG Investigations of Insular Epilepsies Table 2. Coverage of perisylvian cortex by type 2 implantation Electrodes
Contacts Temporal
Case
Grid
17 18 Right Left 19 Mean Min. Max.
0
Strips 1
Depths 2
0 0 1 0.25 0 1
5 4 6 4 1 6
3 3 3 2.75 2 3
Insular 5 10 9 8 8 5 10
Mesial 5 6 6 10 6.75 5 10
Neocortical 6 14 22 23 16.25 6 23
Parietal
Frontal
0
0
8 8 8 6 0 8
8 8 20 9 0 20
Min., minimum; Max., maximum.
respectively. Surgical outcome was Engel class I for all operated cases (mean follow-up 24 months) except for one (Patient 2) who eventually attained seizure-freedom (Engel class I) after gamma-knife treatment (Engel et al., 1993). Transient complications attributable to the surgery were noted in three patients: transient dysphasia lasting less than a day in one patient, transient mild limb weakness for