SEARCH

SEARCH BY CITATION

Keywords:

  • Magnetoencephalography;
  • Insula;
  • Refractory epilepsy;
  • Epilepsy surgery

Summary

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Disclosure
  7. References
  8. Biography

Purpose

To study the utility of magnetoencephalography (MEG) in patients with refractory insular epilepsy. Covered by highly functional temporal, frontal, and parietal opercula, insular-onset seizures can manifest a variety of ictal symptoms falsely leading to a diagnosis of temporal, frontal, or parietal lobe seizures. Lack of recognition of insular seizures may be responsible for some epilepsy surgery failures.

Methods

We retrospectively reviewed and analyzed MEG data in 14 patients with refractory insular seizures defined through intracranial electroencephalography (EEG) or by the presence of an epileptogenic lesion in the insula with compatible seizure semiology. MEG was performed as part of the noninvasive presurgical evaluation, using a 275-channel whole head MEG system. MEG data were analyzed using a single equivalent current dipole model. MEG localization was compared to interictal positron emission tomography (PET) and ictal single photon emission computed tomography (SPECT) results and to the resection margin.

Key Findings

Three patterns of MEG spike sources were observed. Seven patients showed an anterior operculoinsular clusters and two patients had a posterior operculoinsular cluster. No spikes were detected in one patient, and the remaining four patients showed a diffuse perisylvian distribution. Spike sources showed uniform orientation perpendicular to the sylvian fissure. Nine patients proceeded to insular epilepsy surgery with favorable surgical outcome. Among patients with anterior operculoinsular cluster who proceeded to have surgery, MEG provided superior information to ictal SPECT in four of six patients and to interictal PET in five of six patients.

Significance

MEG is useful in identifying patients who are likely to benefit from epilepsy surgery targeting the insula, particularly if a tight dipole cluster is identified even if other noninvasive modalities fail to produce localizing results.

In the past decade, there has been growing interest in revisiting the role of the insula in refractory focal epilepsy. The extensive connections of the insula with other potentially epileptogenic areas, like the mesial temporal structures or frontal and parietal cortices, can account for variability in seizure semiology and nonlocalizing or misleading findings in electrophysiologic and imaging investigations (Isnard et al., 2000; Ostrowsky et al., 2000; Isnard et al., 2004; Nguyen et al., 2009a,b). Insular lobe seizures have been reported to have semiology that mimics frontal, temporal, and parietal lobe epilepsies (Ryvlin et al., 2006; Nguyen et al., 2009b). Failure to resect insular foci has been implicated in surgical failures after temporal and frontal lobe surgeries (Harroud et al., 2012).

Noninvasive localization methods using positron emission tomography (PET) or single photon emission computed tomography (SPECT) in patients with insular seizures yielded conflicting data, often with nondiagnostic, misleading results or demonstrating extensive multifocal abnormalities (Ryvlin et al., 2006; Nguyen et al., 2009b). The distance of the insular cortex to the surface and the overlying cortex can lead to imprecise results of surface electroencephalography (EEG) or even subdural invasive EEG recordings, and can make it difficult to differentiate seizures originating from the insula from seizures originating from the overlying temporal, frontal, or parietal opercula (Nguyen et al., 2009b). The implantation of depth electrodes or a combination of surface and depth electrodes is preferred by most investigators to evaluate the involvement of the insular cortex (Afif et al., 2008; von Lehe et al., 2009; Surbeck et al., 2011). Implantation requires dissection through the sylvian fissure or finding a trajectory for depth electrodes through the rich vasculature of the insula, and is not without risks. Even so, only a few contacts end up sampling the insula itself, and extensive coverage is rarely possible. For these reasons, accurate noninvasive presurgical localization technology is needed to better recognize patients with suspected insular seizures and to guide the strategy for intracranial electrode placement, especially in the absence of localizing epileptogenic lesions with high-resolution MRI.

The utility of magnetoencephalography (MEG) in presurgical evaluation of epilepsy has been extensively studied. MEG can be helpful in detecting epileptogenic zones even if several other techniques have failed (Knowlton et al., 2006; Mohamed et al., 2007; Knowlton et al., 2008a,b; Stefan et al., 2011). MEG offers the advantage of directly localizing neuronal activity rather than relying on the accompanying hemodynamic or metabolic changes. It is disputed in the literature whether MEG is able to localize epileptic activity from deep cerebral structures, such as the sylvian fissure and insula, as the magnetic signal decays the farther the source from the MEG sensors. However, Paetau et al. (1999) recorded intrasylvian MEG spikes in children with Landau-Kleffner syndrome that resolved after multiple subpial transection. Insular and periinsular MEG sources were also detected in small numbers of patients in other studies (Heers et al., 2012; Park et al., 2012).

In this study, we examine the utility and additive value of MEG in the localization of the epileptogenic zone in patients with insular epileptogenic foci.

Methods

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Disclosure
  7. References
  8. Biography

Patients

We reviewed MEG results in patients who had confirmed insular seizures based on either an intracranial EEG (ICEEG) study demonstrating ictal onset from the insula or the presence of epileptogenic structural lesion in the insula and compatible seizure semiology.

The decision to undergo an intracranial study sampling the insula, among other regions, was based on findings from a standard presurgical evaluation, which included a detailed history and physical examination, a neuropsychological evaluation, high-resolution cerebral MRI, long-term video-EEG monitoring, ictal SPECT, and 18F-fluorodeoxyglucose positron emission tomography (FDG-PET). MEG data were collected as part of a research protocol between 2006 and 2010, and the results were not included in the surgical decision. From 2010 onward, MEG data were considered in the surgical decision. The MEG data were analyzed by an epileptologist who had access to only a sample of the morphology of scalp-EEG recorded spikes but not the clinical history. The decision to sample the insula and the technique for surgical implantation were published previously (Surbeck et al., 2011).

Simultaneous MEG and EEG recordings, coregistration of MEG with MRI

MEG recordings were performed using a whole-head CTF 275-channel axial gradiometer system (MISL, Coquitlam, BC, Canada). We collected EEG data simultaneously with MEG recording, using either 19 scalp electrodes or a 64-channel head cap. Typically 10 segments of spontaneous data, each five minutes in duration, were recorded in each patient. The sampling rate for data acquisition was 480 Hz. MEG studies were generally performed while patients were admitted for long-term video-EEG monitoring as part of their presurgical evaluation. In that context, antiepileptic drugs were frequently tapered or doses lowered to increase the chances for detecting epileptiform discharges. Furthermore, patients were in a state of mild sleep deprivation for the MEG study.

Equivalent current dipole analyses

A band pass filter of 3–70 Hz was applied to the raw MEG data. The high-pass filter was adjusted to eliminate high voltage slow waves; hence, a high-pass filter of 7 Hz was used in one pediatric patient (patient 11). We applied a single moving equivalent current dipole model (ECD) with a multisphere conductor model, individually fitted to each patient based on the patient's MRI. Epileptic discharges were identified by examining the MEG recordings and cross-referencing them with the simultaneous EEG recording. We marked the earliest peak of each epileptic discharge with stable magnetic field topography. One dipole source was selected for each spike, with the following criteria: residual error between measured and calculated magnetic fields <30%, dipole moment 50–300 nAm and stable dipole localization, moment and magnetic field topography over minimum 10 msec period. Because the ECD model can be misleading when localizing sources of large cortical patches, we analyzed MEG spikes if they were detected only on MEG or accompanied by low amplitude spikes on simultaneous EEG recordings.

We classified MEG spike sources (MEGSSs) according to their number and spatial density: “clusters” consisted of six or more MEGSSs with 1 cm or less between adjacent sources and “scatter” consisted of six or more scattered spike sources >1 cm apart.

Magnetic resonance imaging

All patients had MRI scans as part of the presurgical evaluation. MRI was performed on an Achieva Dual 3.0 T system (Philips Medical Systems, Best, The Netherlands). All 3.0 T studies included (a) a three-dimensional (3D) T1-weighted gradient-echo acquisition of the whole brain (repetition time [TR]/echo time [TE], 24/6; flip angle, 258; field of view, 256 × 256 mm; matrix, 256 × 192, thickness 1 mm); (b) axial T2-weighted (TR/TE, 24/6; flip angle, 258; field of view, 256 × 256 mm; matrix, 256 × 192) and fluid-attenuated inversion recovery (FLAIR; TR/TE, 24/6; flip angle, 258; field of view, 256 × 256 mm; matrix, 256 × 192) acquisitions of the whole brain; and (c) coronal T2-weighted and FLAIR acquisitions perpendicular to the longitudinal axis of the hippocampus.

We applied quantitative MRI analysis in patients with negative MRI to enhance the detection of subtle cortical dysplasias. The quantitative analysis consisted of a measure of cortical thickness performed with the FreeSurfer software (Dale et al., 1999; Fischl et al., 1999). The presurgical 3T MRI T1-weighted images were processed with FreeSurfer to obtain a 3D segmentation of the cortical surface, a cortical thickness map, and an individual versus control-group node-to-node correspondence to assess regions with increased cortical thickness. The control group was composed of 40 healthy right-handed subjects between the ages of 18 and 35. A z-score map was created for each patient to compare the two measures, using a standard deviation of two and a minimum of 10 voxels.

Results

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Disclosure
  7. References
  8. Biography

Table 1 summarizes clinical features, magnetic resonance imaging (MRI) findings, and MEG findings in all 14 patients. Ages ranged from 8 to 46. Somatic sensory symptoms were the most common seizure semiology, occurring ipsilateral to the epileptogenic zone in one patient (patient 8), contralateral in six patients (patients 1, 7, 9, 12–14) and bilateral in one patient (patient 11). Auditory illusions or hallucinations were seen in three patients (patients 4, 5, 8). Complex motor behavior was seen in seven patients (patients 2–5, 7, 11, 12). Hypersalivation, throat constriction, and visceral symptoms were seen in four patients. Ictal scalp EEG localized to the temporal region in five patients (patients 1, 7, 8, 13, 14), lateralized with multilobar involvement in three patients (patients 3, 4, 11), and was nonlateralizing in the six remaining patients.

Table 1. Clinical, neurophysiologic, and surgical data for all patients
Patient no.AgeSeizure semiologyInterictal EEGIctal EEGMRIIctal SPECTRelation of SPECT to resectionaPETRelation of PET to resectionaMEGType of MEG findingsbICEEG (y/N)Interictal ICEEGIctal ICEEGSurgeryOutcomeFollow- up (months)
  1. AH, auditory hallucinations; ant, anterior; bi, bilateral; C, central; F, frontal; G, gyrus; GKS, gamma knife surgery; IFG, inferior frontal gyrus; ins, insula; R, MFG, middle frontal gyrus; MTG, middle temporal gyrus; med, medial; lat, lateral; L, left; OF, orbitofrontal; P, parietal; post, posterior; R, right; SFG, superior frontal gyrus; sup, superior; STG, superior temporal gyrus SSS, somatic sensory symptoms; T, temporal; UE, upper extremity; 2GTC, secondarily generalized tonic–clonic seizures. [RIGHTWARDS ARROW], spread pattern.

  2. a

    Relation of ictal SPECT and interictal PET to resection or presumed epileptogenic zone: NL, nonlocalizing; Lat, lateralizing; C, concordant; D, discordant; patients 10–12 did not have surgery, hence the relation to resection cannot be determined.

  3. b

    Types of MEG; ant O-I, anterior operculoinsular; post O-I, posterior operculoinsular.

  4. c

    Patient 1, statuspost failed selective amygdalohippocampectomy and anterior temporal lobectomy; patient 7, statuspost unsuccessful L parietal operculum resection; patient 10, statuspost failed R frontopolar resection guided by ICEEG without insular coverage; patient 13, statuspost R partial operculoinsulectomy; patient 14, status post L posterior insulectomy guided by ICEEG and then GKS targeting insula (Engel class IC).

  5. d

    In patient 2, insular depth electrodes covered only the posterior insula. Anterior insular resection was based on MEG results.

  6. e

    Patients 11 and 12 had subtle cortical dysplasia identified on quantitative MRI.

151Hypersalivation, dysgeusia, LUE SSSRTRTSurgical cavitycNo activationNLR TLatClusterR residual ant T, ant ins, OFant O-IYR ant ins, R OF, R STG, R IFGR ant ins[RIGHTWARDS ARROW]post insComplete R insulectomyEngel IA23
237Gelastic ± complex motor behaviorRF, biFDiffuseF atrophyMultifocalNLL ant T, L OFDClusterR IFG, ant insant O-IYR IFG, OF, CingulateR IFG[RIGHTWARDS ARROW] OF + med FGdR IFG corticectomy + partial insEngel IA15
327Gelastic ± complex motor behaviorRFTRFTNMultifocalNLNNLClusterR ant ins, post OFant O-IYR OF, F operculum, ant insJunction of R OF, F operculum, ant insR OF, ant insEngel IA13
437AH, complex motor behaviorLFTLFTNL IFG + ant ins + T poleCL T pole + med TLatClusterL ant ins, IFGant O-IYL ant ins, IFG, MFG, STGL ant ins[RIGHTWARDS ARROW]IFG, MFGL ant ins + limited resection of L IFG, STGEngel IA6
520Viscerosensory, AH ± complex motor behaviorLT, LFTDiffuse or RTMultiple tubersR T-insCR MFG, R insCClusterR ant insant O-INR insulectomyEngel IA6
637Dyscognitive ± 2GTCSL FT, biFbiFT L > RNL IFG, SFG, STG, Fpolar, OFLatL IFG + lat TCClusterL sup T pole, ins, F operculumant O-IYL ant ins, post ins, med T, IFG, OFL ant ins [RIGHTWARDS ARROW]med T, IFG, OFAnterior insulectomy (epileptogenic Broca could not be removed)Engel IIA29
746R facial SSS, complex motor behaviorLTLTSurgical cavitycMultifocalNLNNLClusterL post STG, post insPost O-IYL post ins, P operculum, T operculum, MTGL post ins[RIGHTWARDS ARROW]P operculum, MTG, STGL post ins + L P operculum + Heschl's gyrusEngel IA7
841LUE SSS, AH, dysphasiaLTLTNL med T, L Lat TCL med T + T poleLatScatterL ant T vertical, ins, post T vertical, STGDiffuse perisylvianYL ins, med and lat T, IFGa) L ins or b) F operculum or c) T operculum or STG[RIGHTWARDS ARROW] medial TL lat T (sparing Wernicke) + insulectomy (epileptogenic Broca's area could not be removed)Engel IIA40
942RUE SSS, HypermotorLT, LFT, biFTDiffuseNMultifocal (including bi ins)NLNNLScatterL post ins, parietal operculumDiffuse perisylvianYL post ins, OF, Fpolar, ant med FGL post ins[RIGHTWARDS ARROW] ant ins, med FGKS targeting L post insulaEngel IID36
1045Visceral, flushing, confusion ± rare GTCSbiFDiffuseSurgical cavitycMultifocalNLR Fpolar, R med TClusterR ant ins, IFGant O-IN  
118Bilateral SSS, complex motor behaviorRT, RCT, RFCTRCTN (R ins subtle CD)eR insCNNLClusterPost 1/3 sylvian fissure, post insPost O-IN  
1231RUE SSS, complex motor behaviorNo spikesDiffuseN (L post ins subtle CD)eMultifocalNLL insCNo spikesN  
1337Vertigo, LUE SSS, laryngeal constrictionRTRTR temporoinsular oligoastrocytomacMultifocalNLNLScatterR T, R insDiffuse perisylvianN  
1445R SSSLTLTSurgical CavitycMultifocalNLNNLScatterL T, L STGDiffuse perisylvianN  

MRI findings

Five patients had normal MRI studies (patients 3, 4, 6, 8, 9). Subtle insular T2 abnormal signal or insular abnormality on quantitative MRI was seen in two patients (patients 11, 12) where structural MRI was interpreted originally as normal, frontal atrophy was seen in one patient (patient 2), and multiple tubers were seen in one patient with tuberous sclerosis (patients 5). MEG was performed after unsuccessful epilepsy surgery in five patients (patients 1, 7, 10, 13, 14). Patient 10 had two MEG studies, the first prior to a failed frontal polar resection guided by invasive electrodes with no insular coverage and the second postoperatively because of persistent seizures.

MEG findings

Three patterns of MEG spike sources were observed: anterior operculoinsular cluster, posterior operculoinsular cluster, and diffuse perisylvian scatter (Fig. 1). A tight anterior operculoinsular cluster was seen in seven patients (patients 1–6, 10). MEG showed a tight dipole cluster with uniform orientation over the frontal operculum, anterior insula (Fig. 2), and also overlying the adjacent inferior frontal gyrus in three patients (patients 2, 4, 10) and the adjacent orbitofrontal cortex in two patients (patients 1 and 3). Most of the spike sources were oriented perpendicular to the sylvian fissure. Two patients had posterior operculoinsular clusters with vertically oriented dipoles overlying the posterior third of the sylvian fissure and posterior insula (patients 7 and 11) (Fig. 1). Anterior and posterior operculoinsular clusters were named based on where the majority of spikes were localized, even though some of the spikes were located beyond the operculoinsular region (Fig. 3). Diffuse perisylvian spikes were seen in the remaining four patients over the insula, superior temporal gyrus, and the temporal and parietal opercula (patients 8, 9, 13, 14). MEG was performed after failed previous epilepsy surgery in five patients (patients 1, 7, 10, 13, 14). One patient (patient 10) had two MEG studies before and after an unsuccessful right frontal polar resection guided by ICEEG coverage that did not include the insula. His first MEG study, performed prior to epilepsy surgery, showed a tight anterior operculoinsular cluster. MEG results were not included in the clinical decision at that time, and the data were analyzed after the failed surgery. A second review of the ICEEG data demonstrated subtle high frequency activity at ictal onset over the right inferior frontal gyrus that was missed at the time of surgery. The seizure semiology in this patient was more suggestive of insular localization, although the insula was not sampled during the first surgery. Repeat MEG was performed after the failed surgery and showed fewer spikes with diffuse perisylvian distribution.

image

Figure 1. MEG patterns in patients with insular seizures. Upper row demonstrates patients with anterior operculoinsular clusters (patients 3, 2, 10, 4). Lower row, diffuse perisylvian (patients 13, 8, 7) and posterior operculoinsular cluster (patient 11). Most dipoles show a uniform orientation vertical to the sylvian fissure.

Download figure to PowerPoint

image

Figure 2. EEG (band pass 1–50 Hz), selected MEG channels over the left frontal and temporal regions (band pass 5–70 Hz), topography map and dipole localization in patient no 4. EEG showed low amplitude spikes over F7; MEG demonstrated repetitive spikes with dipole localization over the anterior insula and inferior frontal gyrus. The patient is seizure-free after anterior insular and limited inferior frontal resection guided by intracranial EEG.

Download figure to PowerPoint

image

Figure 3. Simultaneous recording of scalp EEG and MEG in patient 5 with right anterior operculoinsular cluster EEG (band pass 1–50 Hz), selected MEG channels over the right temporal region (band pass 5–70 Hz), topography map and dipole localization are demonstrated. (A) two types of spikes were demonstrated; black arrow marks MEG spikes with no definitive EEG correlates that localized to the posterior insula (B) and MEG spikes (green arrow) with simultaneous EEG spikes that localized to the anterior temporal lobe (C). The most common type of epileptiform discharges in this patient were repetitive runs of low amplitude MEG spikes (D; red arrows) with or without EEG correlates that localized to the anterior insula (E). A tight anterior operculoinsular cluster is demonstrated with scattered spikes over the posterior insula and anterior temporal lobe (F). The patient became seizure-free after right insular resection.

Download figure to PowerPoint

Correlation between MEG, PET, SPECT localization, ICEEG and surgical outcome

Ictal SPECT was performed in all patients, and interictal PET scan was performed in 13 patients. Ictal SPECT and interictal PET were compared to the ictal onset zone on ICEEG or to the resection area if direct surgery was performed in the group of patients who had epilepsy surgery. Ictal SPECT and interictal PET were classified as concordant if hyperperfusion or hypometabolism involved focal lobar or sublobar contiguous areas that overlapped the ictal onset zone or the resection margin, lateralizing if it showed diffuse or multifocal involvement within one hemisphere concordant with the epileptogenic hemisphere but not overlapping with the ictal onset, nonlocalizing if it showed multifocal involvement of both hemispheres, and discordant if it localized contralateral to the ictal onset.

Ictal SPECT provided localizing information in four patients (patients 4, 5, 8, 11), whereas 10 patients had a nonlocalizing ictal SPECT demonstrating either no, diffuse, or multifocal bilateral hyperperfusion. PET scan showed normal glucose metabolism in five patients (patients 3, 7, 9, 11, 14). Focal area of glucose hypometabolism was seen in three patients (patients 5, 6, 12). Among the nine patients who had subsequent epilepsy surgery, ictal SPECT and interictal PET provided concordant localizing information in three and two patients, respectively.

Nine patients underwent insular epilepsy surgery. ICEEG was performed in eight patients. One patient (patient 5) had extensive insular resection without invasive monitoring. In this patient, ictal SPECT, interictal PET, and MEG showed concordant localizing information pointing to the right insula. ICEEG captured seizures arising from the insula in five patients (patients 1, 4, 6, 7, 9), from the insula and adjacent orbitofrontal region in one patient (patient 3) and several seizure types arising from the insula, frontal operculum, and superior temporal gyrus in one patient (patient 8). One patient had inadequate insular coverage and insular resection was performed based on the MEG results (patient 2). Surgical resections were limited to the insula in four patients (patients 1, 5, 6, 9). Additional resections were performed on the adjacent orbitofrontal cortex in one patient (patient 3), inferior frontal gyrus in two patients (patients 2 and 4), parietal operculum and Heschl's gyrus in one patient (patient 7), and lateral temporal cortex in one patient (patient 8). All nine patients had a favorable surgical outcome (Engel class I or II). Epilepsy surgery in six of seven patients with anterior or posterior operculoinsular cluster resulted in seizure freedom, whereas both patients with scattered perisylvian spike sources had Engel class II outcome. In both patients, incomplete resection of the ictal onset zone on ICEEG was performed to preserve language areas on the left hemisphere. Complete resection of the MEG cluster region was performed in five patients; all became seizure-free (patients 2–5, 7). In seizure-free patients, MEG provided superior information to ictal SPECT in four of six patients and to interictal PET in five of six patients (Fig. 4).

image

Figure 4. Intracranial EEG, SPECT, and MEG findings in patient 3. Ictal SPECT showed multifocal areas of hyperperfusion and interictal PET was normal. Interictal MEG showed a tight cluster over the anterior insula and posterior orbitofrontal regions with uniform orientation. Intracranial EEG using extensive subdural strips over the frontal and temporal lobes as well as insular depth electrodes captured seizures arising from the right insula and orbitofrontal cortex. Rhythmic spikes and low amplitude fast activity was seen at ictal onset over insular depth contact U22 and electrode F45 overlying the orbitofrontal region (red circle). The patient became seizure-free after insular and orbitofrontal resection.

Download figure to PowerPoint

Five patients did not undergo epilepsy surgery. Two patients were hesitant to undergo epilepsy surgery (patients 11 and 12). One patient had coexisting nonepileptic seizures and surgery was delayed (patient 13). One patient achieved seizure control after modifying antiepileptic drug therapy (patient 14). One patient awaits surgery (patient 10).

Discussion

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Disclosure
  7. References
  8. Biography

The goal of this study was to assess the utility of MEG in the presurgical workup of refractory insular epilepsy and evaluate the additive value of MEG compared to other noninvasive localization methods. There are few reports describing MEG findings in suspected opercular and insular epilepsy in small numbers of patients. Park et al. (2012) described insular MEG sources in two patients with insular lesional epilepsy secondary to cavernous angioma of the insula. Combined MEG and EEG spikes localized to the anterior temporal lobe, whereas MEG-only spikes localized to the perilesional insular cortex. Heers et al. (2012) described MEG findings in three patients with occult insular or periinsular epilepsy. MEG failed to detect sufficient spikes in one patient and detected suprasylvian and posterior insular sources in the two remaining patients. Both patients underwent epilepsy surgery with good seizure outcomes. Kakisaka et al. (2012) described frontoparietal opercular and perirolandic MEG spike sources in four patients where EEG failed to detect any spikes. Two patients became seizure-free after epilepsy surgery including resection of the frontoparietal operculum. Our data provide further supportive evidence for the utility of MEG in the presurgical evaluation of refractory insular epilepsy. Moreover, insular localization was confirmed in our patients through intracranial recording demonstrating seizures arising from the insular cortex and a good postoperative outcome following insular resections. Our data are also in concordance with previous reports demonstrating the importance of MEG in presurgical localization (Otsubo et al., 2001; Knowlton et al., 2008a) and that complete resection of tightly clustered MEG spike sources is associated with a favorable epilepsy surgery outcome (Oishi et al., 2006).

The variability in the semiology of insular seizures makes it difficult to differentiate between frontal, temporal, and parietal lobe epilepsy. Ictal scalp EEG, as in our cases, would not differentiate insular seizures from temporal, frontal, or parietal lobe epilepsy (Isnard et al., 2000; Nguyen et al., 2009b). Failure to recognize or suspect an insular epileptogenic focus would often result in surgical failure. An intracranial study is often necessary to confirm insular seizures, especially in the absence of an epileptogenic structural lesion. It is desirable to identify through noninvasive investigations the subset of patients with a high probability of insular seizures, thus preventing unnecessary insular implantations or guiding a targeted minimally invasive approach toward the operculoinsular structures to reduce the cost and risk of intracranial studies. In our cases, MEG proved to be superior to both ictal SPECT and interictal PET in identifying patients with insular seizures, particularly in patients with anterior operculoinsular cluster. In this subgroup, insular involvement was suggested by interictal PET in one of seven patients and ictal SPECT in two of seven patients. Six patients with anterior operculoinsular cluster achieved favorable surgical outcome after insular epilepsy surgery. We routinely now sample the insula during invasive monitoring when an anterior operculoinsular cluster is detected.

There is debate on whether MEG can detect insular activity. Theoretically, MEG may fail to detect insular activity, as it detects only currents tangential to the scalp. Currents generated from the insular cortex will be oriented radially. Park et al. (2012) speculated that detection of insular spikes was facilitated by a geometrical change due to the presence of insular lesion. More recently, Kakisaka et al. (2013) described simultaneous MEG and insular stereotactic EEG (SEEG) recordings in an adolescent patients with nonlesional insular epilepsy. MEG detected 83% of all insular and periinsular SEEG spikes, and MEG dipoles were localized in the insular and periinsular region. Our study and those of others (Heers et al., 2012; Park et al., 2012; Kakisaka et al., 2013) confirm that MEG can localize insular spike sources and identify specific MEG patterns that are associated with insular epileptogenic foci, even in the absence of insular structural lesions.

In addition to the insula, some dipoles were also localized to other periinsular structures, particularly to the frontal operculum and anterior part of the sylvian fissure in patients with anterior operculoinsular clusters and along the superior sylvian plane and parietal operculum in patients with posterior operculoinsular clusters. Intracranial stereotactic ictal studies sometimes demonstrated simultaneous involvement of both insular cortex and opercular structures at seizure onset (Proserpio et al., 2011). The superior temporal cortical plane and opposing inferior frontal and parietal opercular planes are ideally oriented to generate MEG fields (Ebersole & Ebersole, 2010). Therefore, some insular spike sources in our patients could have propagated along the sylvian fissure resulting in a periinsular distribution. To decrease the chance of erroneous source localization due to spike propagation, we choose spikes detected only on MEG or accompanied by low amplitude spikes on simultaneous EEG recordings. MEG-only spikes likely represent more restricted focal epileptic discharges, whereas MEG spikes accompanied by high amplitude EEG activity could represent activation of more extended cortical patches compared to MEG-only spikes and therefore may represent propagated activity (Park et al., 2012).

The localization accuracy of MEG depends on the presence of a large number of spikes in the data with a good signal-to-noise ratio. We performed our MEG studies in most cases while patients were admitted for long-term continuous video-EEG monitoring as part of their presurgical evaluation. In that context, antiepileptic drugs had frequently been tapered. Furthermore, patients were in a state of mild sleep deprivation for the MEG study to increase the chances of detecting interictal spikes. Smaller number of spikes in the data will limit the localization accuracy of MEG. Therefore, the absence of insular or periinsular dipoles in an MEG study does not necessarily exclude the possibility of insular epilepsy.

It is important to point out the limitations of this study. Some of our patients did not proceed to epilepsy surgery. Insular epileptogenic zone localization in those cases becomes presumptive based on the presence of an epileptogenic structural lesion and compatible seizure semiology. The duration of follow-up is also relatively short in some patients who had insular surgery. In those patients, however, insular localization is confirmed through intracranial EEG.

Disclosure

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Disclosure
  7. References
  8. Biography

None of the authors has any conflict of interest to disclose. 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.

References

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Disclosure
  7. References
  8. Biography

Biography

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Disclosure
  7. References
  8. Biography
  • Image of creator

    Dr. Mohamed is a pediatric epileptologist at the Department of Pediatrics, Dalhousie University.