Revisiting the role of the insula in refractory partial

epilepsy∗Dang Khoa Nguyen, ∗Dong Bach Nguyen, yRamez Malak, zJean-Maxime Leroux,∗xLionel Carmant, ∗Jean-Marc Saint-Hilaire, ∗Normand Giard, ∗Patrick Cossette,

and yAlain Bouthillier

∗Service de Neurologie, yService de Neurochirurgie, and zDepartement de Radiologie, Hopital Notre-Dame du

CHUM, Universite de Montreal, Montreal, Quebec, Canada; and xService de Neurologie, Hopital Sainte-Justine,

Universite de Montreal, Montreal, Quebec, Canada

SUMMARY

Purpose: Recent evidence suggesting that some

epilepsy surgery failures could be related to unrec-

ognized insular epilepsy have led us to lower our

threshold to sample the insula with intracerebral

electrodes. In this study, we report our experience

resulting from this change in strategy.

Methods: During the period extending from

October 2004 to June 2007, 18 patients had an intra-

cranial study including 10 with insular coverage.

The decision to sample the insula with intracere-

bral electrodes was made in the context of (1) non-

lesional parietal lobe-like epilepsy; (2) nonlesional

frontal lobe-like epilepsy; (3) nonlesional temporal

lobe-like epilepsy; and (4) atypical temporal lobe-

like epilepsy.

Results: Intracerebral recordings confirmed the

presence of insular lobe seizures in four patients.

Cortical stimulation performed in 9 of 10 patients

with insular electrodes elicited, in decreasing

order of frequency, somatosensory, viscerosen-

sory, motor, auditory, vestibular, and speech

symptoms.

Discussion: Our results suggest that insular cor-

tex epilepsy may mimic temporal, frontal, and

parietal lobe epilepsies and that a nonnegligeable

proportion of surgical candidates with drug-resis-

tant epilepsy have an epileptogenic zone that

involves the insula.

KEY WORDS: Insular cortex epilepsy, Refractory

epilepsy, Intracranial study, Cortical stimulation,

Epilepsy surgery.

Recent evidence suggest that failure to recognize theinsula as the epileptogenic zone may be responsible forsome surgical failures in patients diagnosed with temporallobe (TL), parietal lobe (PL), and frontal lobe (FL) epilep-sies (Ryvlin & Kahane, 2005; Ryvlin, 2006). Using intra-cerebral electrodes, insular seizures were demontrated infive patients with TL-like epilepsy (Isnard et al., 2000,2004) and four patients with FL-like epilepsy (Ryvlinet al., 2006; Dobesberger et al., 2008). Furthermore,Aghakhani et al. (2004) suspected the contribution of theinsula in six patients with PL- and/or TL-like epilepsieswho failed TL and PL surgeries. These insightful reports

have led us to modify our presurgical approach of refrac-tory epileptics in 2004, particularly those requiring an inva-sive study, as our threshold to sample the insula withintracranial electrodes is now much lower. For example,most patients with nonlesional refractory epilepsy withsemiological features suggestive of FL, TL, or PL epilepsynow have the insular region covered by depth electrodes.We report here our experience resulting from this changein strategy. The data shown here support the view that theinsula is a great mimicker and plays a nonnegligeable rolein the rate of failure currently encountered in epilepsysurgery.

Methods

Cohort of patientsDuring the period extending from October 2004 and

June 2007, 18 patients had an intracranial study in ourepilepsy service. Among this cohort of 18 patients, 10 had

Accepted June 10, 2008; Early View publication August 20, 2008.Address correspondence to Dang K. Nguyen, Service de Neurologie,

H�pital Notre-Dame du CHUM, 1560 rue Sherbrooke Est, Montr�al,Qu�bec, Canada. E-mail: d.nguyen@umontreal.ca

Wiley Periodicals, Inc.ª 2008 International League Against Epilepsy

Epilepsia, 50(3):510–520, 2009doi: 10.1111/j.1528-1167.2008.01758.x

FULL-LENGTH ORIGINAL RESEARCH

510

an intracranial study that included electrodes sampling theinsular region (among other regions). Cortical structures tobe explored by intracerebral electrodes were chosen basedon findings from a typical presurgical evaluation andagreed upon by members attending the epilepsy surgeryconference. This presurgical evaluation included a detailedquestionnaire and physical examination, a neuropsycho-logical evaluation, high-resolution cerebral magnetic reso-nance imaging (MRI), video electroencephalography(EEG) recording of seizures, ictal single-photon emissioncomputed tomography (SPECT), and 18F fluorodeoxyglu-cose positron emission tomography (PET).

Electrode implantation and intracerebral recordingsIn general, the decision to sample the insula with intrace-

rebral electrodes was made in the context of (1) nonlesionalPL-like epilepsy; (2) nonlesional FL-like epilepsy; (3) non-lesional TL-like epilepsy; and (4) atypical TL-like epilepsyas suggested by early occurrence of laryngeal discomfortwith thoracic oppression or dyspnea, unpleasant paresthe-sia, or warmth sensation focused on the perioral region orextended to a large somatic territory (as suggested byIsnard et al., 2000). In order to maximize the yield of eachspecific intracranial study, a different combination of sub-dural grid, strip, and/or depth electrodes were used for eachpatient capitalizing on the singular advantages of each typeof electrodes. The insular cortex was investigated by meansof depth electrodes placed under direct vision after micro-surgical opening the Sylvian fissure. In short, a frontotem-poral craniotomy was performed. After opening the duramatter, the Sylvian fissure was identified with the help ofneuronavigation and dissected using microsurgical tech-niques. Care was taken to spare most of the veins crossingthe Sylvian fissure. The fissure was opened to expose theinsular cortical area of interest: Anterior, posterior, or both.This was determined by the noninvasive preoperativeinvestigation. Once exposed, the surface of the insula wasinspected to localize the insular arteries (the M2 branchesof the middle cerebral artery). Safe areas between the arter-ies were then chosen for electrode placement. This wasdone after a small incision of the pia matter with a micro-scalpel. The depth electrode was then inserted under directvision. The depth electrodes used (Spencer depth elec-trodes; Ad-Tech Medical Instrument Corporation, Racine,WI, U.S.A.) had four contacts, each of 1.1 mm length, sepa-rated by 2.3 mm. In most cases, two contacts were insertedin the insular cortex. Each electrode was sutured to sub-dural electrodes placed on the lateral surface of the hemi-sphere and to the dura matter. At least two depth electrodeswere implanted in each patient. We also took advantage ofthe craniotomy to place other depth and subdural elec-trodes, depending on the preoperative investigation. Thescheme of implantation and explored areas for each patientare listed in Table 1. Postimplantation MRI was always per-formed to determine the exact position of the electrodes.

One hundred twenty-eight channels of simultaneous EEGrecordings were available for adequate display and inter-pretation of the information.

Cortical stimulation protocolIntracerebral electrical stimulation was performed to test

the excitability and functionality of the investigated brainregions. The parameters for extraoperative cortical map-ping using the Grass S88 Stimulator (Grass Instruments,W. Warwick, RI, U.S.A.) were as follows: Stimulus fre-quency of 50 Hz with a pulse width of 100 ms, a stimulusintensity of 1–10 mA, and a stimulus train duration of 5 s.Stimuli were administered with an intervening rest intervalof at least 1 min. Afterdischarges (ADs) were assessed ateach stimulation. For language mapping, our procedureincluded object naming, sentence completion, reading, andverbal commands tasks.

Insular site locationIn order to create a cartography of implanted insular

contacts, the patients’ postimplantation MRI files wereretrieved. Using SPM99 software package (WellcomeDepartment of Cognitive Neurology, London, U.K.),dicom files were converted with our Matlab toolbox(MathWorks, Natick, MA, U.S.A.) into analyze format(*.img, *.hdr) files to be opened with MRIcro. The ninepatients’ brains were normalized to fit the spatially normal-ized single-subject high-resolution T1 volume provided bythe Montreal Neurological Institute (Collins et al., 1998).Once normalized to fit the standardized brain, eachpatient’s contacts were manually detected and placed onthe template according to their MNI coordinates. Contactsthat were found not be in the insula area were removed.

Results

The clinical and paraclinical features of all 10 patientswho had insular electrodes are summarized in Table 1.There were four men and six women. Age varied between19 and 41 years of age. Apart from case 1, none hadobvious seizure risk factors. None had experienced febrileseizures except possibly patient 5. MRI was reportednormal except for two patients: Patient 1 had a right hemi-spheric atrophy with right hippocampal sclerosis followingmeningitis in childhood, and patient 10 had a noncongruentmild left hippocampal atrophy contralateral to suspectedepileptogenic zone. Intracerebral recordings confirmed thepresence of insular lobe seizures in four patients (patients1, 2, 5, and 6). There were no clear distinguishing featuresbetween the four patients with insular seizures and the sixpatients without (demographics, age of onset, scalp EEG,functional studies). We report below two sets of data help-ful in determining the role of the insula in refractoryepilepsy: First, the clinical and intracranial EEG presenta-tion of four patients with insular cortex seizures and

511

Intracerebral Study of Refractory Insular Cortex Epilepsy

Epilepsia, 50(3):510–520, 2009doi: 10.1111/j.1528-1167.2008.01758.x

Tab

le1.

Pre

surg

icalevalu

ati

on

,su

rgic

altr

eatm

en

t,an

dsu

rgic

alo

utc

om

e

Seiz

ure

Outc

om

eFU

Cas

eA

geSe

xEtiolo

gySe

mio

logy

MR

IFD

GPET

SISC

OM

ICele

ctro

des

Ispik

es

Idis

char

ges

focu

sC

ort

ect

om

y(E

nge

l)(m

onth

s)Pat

holo

gy

123

FM

enin

gitis

Lhan

dplu

sth

roat

par

est

hesi

as,r

isin

g

epig

astr

ic

sensa

tion,f

eelin

g

ofs

tran

gula

tion,

±al

tere

dcs

c

Rhem

ispheri

c

atro

phy

incl

u-

din

gR

HS

Notdone

RT

,RT

-I>

LF

1G

rid

(RF-

P-T

);

2st

rips

(RT

);

3dep

ths

(RA

,

RH

,RI)

+rd

sor

lvfa

RT

-IR

T-I

Ia38

Glio

sis

(HS)

236

MU

nkn

ow

nR

han

dfi

Rlim

b

par

est

hesi

as/p

ain

fiR

arm

ele

vation

Norm

alN

orm

alL

F-P,I

,cin

g1

Gri

d(L

P);

2

dep

ths

(LI)

;

5st

rips

(LT

,LF,

inte

rhem

ispheri

c)

+lv

faL

IL

post

IIII(

Iapost

-

GK

S)

35

n.a

.

325

FU

nkn

ow

nA

rousa

lfi

thora

cic

and

thro

at

oppre

ssio

nfi

tras

hin

g,pedal

ing

Norm

alN

otdone

Multip

leN

-L1

Gri

d(R

F);2

dep

ths

(RI)

;5st

rips

(Rorb

ito-F

,lat

F)

))

Rm

esi

o-F

Rm

esi

o-F

IVb

26

Norm

al

419

FU

nkn

ow

nA

rousa

lfi

min

imal

low

er

limb

mvm

ts

±m

ildpedal

ing

±al

tere

dcs

c

Norm

alfl

RF

Multip

leN

-L1

Gri

d(R

latF)

;1

dep

th(R

I);1

1

stri

ps

(LF,

LT

,

Lpar

asag

,

Lin

terh

em

ispheri

c,

Roperc

ulo

-F,

Rorb

ito-F

,RT

)

))

Rm

esi

o-F

Lm

esi

o-F

Ia22

Glio

sis

537

FU

nkn

ow

nA

rousa

lfi

asym

tonic

post

uri

ng,

grim

acin

g,

voca

lizat

ion,

pedal

ing,

kic

kin

g,ro

ckin

g;

diu

rnal

lipan

d

Rhan

dpar

esth

esi

as

Norm

al(n

on-

speci

fic

LIm

illi-

metr

icsi

gnal

)

flL

IM

ultip

leN

-L2

Depth

s(L

I);1

1

stri

ps

(Lorb

ito-F

,

Lla

tF,

LT

,

Lin

terh

em

ispheri

c,

LP)

+lv

faL

IPen

din

gn.a

.n.a

.n.a

.

626

FU

nkn

ow

nT

hora

cic

plu

sth

roat

plu

slo

wer

jaw

swelli

ng

sensa

tion

fife

ar,a

nxie

ty

±deja

vu±

alte

red

csc

plu

s

auto

mat

ism

splu

s

spitting

Norm

alfl

RT

RT

>R

P5

Depth

s(L

I,L

A,

LH

),4

stri

ps

(LT

)

)rd

sor

lvfa

RT

-IR

T-I

Ia17

Glio

sis

(CA

4);

ect

opic

gian

t

neuro

ns

Continued.

512

D. K. Nguyen et al.

Epilepsia, 50(3):510–520, 2009doi: 10.1111/j.1528-1167.2008.01758.x

Tab

le1.

Co

nti

nu

ed

Seiz

ure

Outc

om

eFU

Cas

eA

geSe

xEtiolo

gySe

mio

logy

MR

IFD

GPET

SISC

OM

ICele

ctro

des

Ispik

es

Idis

char

ges

focu

sC

ort

ecto

my

(Enge

l)(m

onth

s)Pat

holo

gy

741

MU

nknow

nEpig

astr

icbuzz

,

nau

sea,

oeso

phag

eal

heat

,

bac

kch

illfi

alte

red

csc

Norm

alfl

Rhem

isphere

RT

,F,P

5D

epth

s(R

A,

RH

,RI)

,9

stri

ps

(RT

,

Rorb

ito-F

,

Rin

terh

em

i-

spher

ic

))

Rm

esi

o-T

RA

ITL

Ia16

Glio

sis

(HS)

825

MU

nknow

nA

rousa

lfi

fear

,

pedal

ing,

kic

kin

g,

tras

hin

g;ra

re

diu

rnal

cephal

icau

ra

Norm

alN

orm

alM

ultip

le

N-L

3D

epth

s(R

A,

RH

,RI)

,11

stri

ps

(Rin

ter-

hem

ispheri

c,

Rorb

ito-F

,

Rla

tF,

2T

)

))

Rm

esi

o-F

Rm

esi

o-F

n.a

.n.a

.n.a

.

919

FU

nknow

nR

arm

par

est

hesi

as/p

ain

Norm

al(s

usp

ici-

ous

Lfr

onta

l

sulc

us)

Norm

alM

ultip

le

N-L

2G

rids

(LF,

L

FP),

2depth

s

(LF2

,LI)

+)

LF

dors

o-lat

era

lL

F2Ia

11

Cort

ical

dys

pla

sia

10

24

MU

nknow

nH

otflas

h,d

izzy

fial

tere

dcs

cfi

±

frequentG

TC

Lm

ildhip

po-

cam

pal

atro

phy

flR

TM

ultip

le

N-L

1G

rid

(LF-

T),

4depth

s(R

I,R

H),

8st

rips

(LT

subd,R

T,

Rorb

ito-F

))

Rm

esi

o-T

RA

ITL

Ia9

n.a

.

Tab

lesu

mm

ariz

ing

dem

ogr

aphic

s,pre

surg

ical

findin

gs,i

ntr

acra

nia

lEEG

resu

lts,

and

surg

ical

outc

om

eofth

e10

pat

ients

with

insu

lar

intr

acere

bra

lele

ctro

des.

Pat

ients

with

dem

-

onst

rate

din

sula

rse

izure

sar

ehig

hlig

hte

din

gray

.L,l

eft

;R,ri

ght;

csc,

consc

iousn

ess

;H

S,hip

poca

mpal

scle

rosi

s;F,

fronta

l;T

,te

mpora

l;I,

insu

lar;

P,p

arie

tal;

A,a

myg

dal

a;H

,hip

po-

cam

pus;

N-L

,nonlo

caliz

ing;

par

asag

,par

asag

itta

l;rd

s,re

cruitin

grh

ythm

icsp

ikes

or

spik

ean

dsl

ow

wav

es;

lvfa

,lo

wvo

ltag

efa

stac

tivi

ty;C

ort

ect

om

y:T

,te

mpora

llo

bect

om

y;I,

insu

lect

om

y;A

ITL,an

teri

or

inte

rnal

tem

pora

llo

bect

om

y;F2

,re

sect

ion

ofF2

sulc

us;

mesi

o-F

,re

sect

ion

ofci

ngu

late

gyru

s,F1

±F2

±su

pple

menta

rym

oto

rar

ea;

GK

S,c-

knife

sur-

gery

;FU

,follo

w-u

p;n

.a.,

notap

plic

able

.

513

Intracerebral Study of Refractory Insular Cortex Epilepsy

Epilepsia, 50(3):510–520, 2009doi: 10.1111/j.1528-1167.2008.01758.x

second, the behavioral responses to electrical stimulationof the insula.

Clinical and intracranial EEG presentation of seizuresoriginating in the insula

Case 1, with a past history of meningitis, suffered fromdaily to weekly seizures characterized by left hand andthroat paresthesias, followed by a rising epigastric sensa-tion and a feeling of strangulation with or without alteredconsciousness. Evaluation disclosed right hemisphericatrophy on MRI, right mesiotemporal activation on a firstictal SPECT and right temporoinsular activation on a sec-ond, and right hemispheric rhythmic h ictal activity onscalp EEG. The invasive study included a grid over theright frontoparietotemporal carrefour, depth electrodes in

the right insula, amygdala, and hippocampus, and subduralstrips for the right lateral and inferior temporal regions(Fig. 1). Several episodes of left hand paresthesias wereassociated with rhythmic spiking involving synchronouslytwo contacts in the insula and two contacts in the superiortemporal gyrus. In addition, one complex partial seizureoriginating from mesiotemporal structures and spreadingto the insula was recorded. The patient underwent a righttemporal lobectomy plus insulectomy and has remainedseizure-free since (follow-up 38 months).

Case 2 developed seizures at age 30 characterized by anintermittent burning, swelling, wave-like sensation overthe right hand for a few seconds, recurring multiple times aday in a crescendo fashion and building up to a brief(10–20 s) but intense pain over the right arm/groin/leg

Figure 1.

The first row shows the three-dimensional (3D) visualization of each patient’s brain along with the intracranial

electrodes used. Point of entrance of depth electrodes or interhemispheric strips are represented by an asterisk.

The second and third rows show, on axial and sagittal planes, respectively, the location of insular contacts that were

involved at seizure onset or considered to belong to the epileptogenic zone for all four patients with insular cortex

epilepsy.

Epilepsia ILAE

514

D. K. Nguyen et al.

Epilepsia, 50(3):510–520, 2009doi: 10.1111/j.1528-1167.2008.01758.x

areas up to 20 times a day. Evaluation revealed discrete in-terictal EEG irregularities over the left frontocentrotempo-ral regions, no ictal EEG changes during briefdysesthesias, but late semirhythmic h slow waves over thesame regions during longer spells, multiple small areas ofhyperperfusion on ictal SPECT, including the cingulategyrus and left insula, no hypometabolism on PET, and nolesion on MRI. The invasive study comprised a grid overthe left parietal region, depth electrodes into the left insula,and subdural strips reaching the cingulate gyrus (Fig. 1).Interictal spikes were noted over insular contacts. Dyses-thetic spells correlated with low voltage fast activity in theinsula evolving into rhythmic spiking (Fig. 2). A left pos-terior insulectomy was performed that led to a 7-week sei-zure-free period prior recurrence. The following year,complementary c-knife surgery targeting the insular areaanterior to the resection led 18 months later to seizure free-dom for the last 4 months.

Case 5 suffered since age 3 years from daily/weeklynocturnal seizures characterized by sudden arousal, asym-

metric tonic posturing, facial grimacing, vocalization, andhypermotor features. Rare daily spells were identical, butsometimes preceded by dizziness and lip paresthesiasmigrating to the right arm or bilaterally. Evaluationdisclosed left frontal and rare temporal interictal spikes,muscle artifacts on ictal EEG recordings, multiple activa-tion sites on two ictal SPECT, including bilateral insularregions, and no PET abnormality. A nonspecific millimet-ric hyperintense signal on fluid-attenuated inversion recov-ery (FLAIR) over the posterior left subinsular region wasnoted on a 1.5 tesla MRI, but not seen on a 3.0 tesla MRI.Depth electrodes were used to sample the left insula, andmultiple subdural strips were glided over inferior, lateral,and mesial portions of the left frontal lobe as well as partsof the lateral temporal and parietal cortices (Fig. 1).Implantation was complicated by dysphasia from a contu-sion of Wernicke’s area with full recovery over 1 week.Ten seizures were recorded, all associated with low-voltage fast activity originating in the left insula spreadingto temporal, parietal, and frontal regions. Because of the

Figure 2.

Intracerebral recording of interictal insular spikes and insular seizures in patient 2. (A) Preictal stage consisting of

periodic spikes or sharp waves (arrow) occurring in a rhythmic fashion confined to the insula. (B) Transition from a

preictal stage to an ictal pattern characterized by a low-voltage fast activity over the same electrode contacts (arrow)

and ending with rhythmic spike and slow waves. (C) Longer seizure showing the same transition from preictal spiking

to an ictal low-voltage fast activity (arrow) increasing in amplitude and decreasing in frequency. Parietal, electrode

contacts from the subdural grid overlying the parietal lobe; insula, electrode contacts from the insular depth

electrodes; temp, electrode contacts from the subdural strips covering the temporal lobe; front, electrode contacts

from subdural strips covering the dorsolateral frontal region; interhemisph, electrode contacts from subdural strips

reaching the cingulate gyrus.

Epilepsia ILAE

515

Intracerebral Study of Refractory Insular Cortex Epilepsy

Epilepsia, 50(3):510–520, 2009doi: 10.1111/j.1528-1167.2008.01758.x

contusion, electrodes were removed, and the surgerypostponed.

Case 6 had, since age 21 years, seizures characterized byan initial swelling sensation over the thorax, throat, andlower jaw, immediately followed by fear, anxiety, andsometimes d�j� vu, ending with alteration of conscious-ness, verbal automatisms, spitting, and postictal confusion.Evaluation revealed no epileptogenic lesion, no clear inte-rictal spikes, right centrotemporal rhythmic h ictal activitywith rapid spread, significant right temporal and milderright frontal, insular, and parietal activation sites on threeseparate ictal SPECT, and moderate right temporalhypometabolism on PET. Depth electrodes were implantedin the right insula, amygdala, and hippocampus in additionto subdural strips for lateral temporal regions (Fig. 1).Spikes were found over the right insula, superior temporalgyrus, and mesiotemporal structures. Three seizures origi-nated from the superior temporal gyrus, and three seizurescame from a temporoinsular epileptogenic zone with a lowvoltage fast activity seen concomitantly over themesiotemporal and insular contacts. Cortical stimulationof insular contacts reproduced the initial swelling sensa-tion, while stimulation of the amygdala reproduced thefear, anxiety, and d�j� vu. She underwent a right temporallobectomy plus insulectomy with no subsequent seizures(follow-up 17 months). Pathology showed ectopic giantneurons.

Electrical stimulation of the insular cortexNine patients with insular intracerebral electrodes

underwent cortical stimulation. Stimulations were per-formed in 36 insular sites. These 36 insular sites were stim-ulated one to three times each, with a total of 96intracortical insular stimulations. Clinical responses wereevoked in seven of the patients (7 of 9 patients; 78%), in 32sites (32 of 36; 89%). Elicited responses were evoked inthe absence of an AD on 14 sites or a very brief localizedAD, which remained over the stimulated contacts in 21sites. Evoked response at one insular site was associatedwith a diffusing AD and excluded from topographical anal-ysis. Somatosensory symptoms were the most frequentlyencountered responses, representing 62% of all evokedresponses. Sensations were described as a numbing, tin-gling, warmth, electric, or airflow feeling. Areas involvedincluded the nasooropharyngealcervical but also limbareas. When somatosensory symptoms involved limbs,most responses were controlateral (six responses), but asignificant number were bilateral (four responses). Vis-ceral symptoms were next in line, representing 12% of theelicited responses. Half were described as nausea, abdomi-nal buzz, or rising warmth sensation in the digestive sys-tem, while the other half were special taste sensations.Motor symptoms were evoked in 12% of the responsesconsisting of ocular movements causing difficulty to focusor fixate an object and contralateral elevation of the right

arm. Auditory responses represented 9% of the evokedresponses and were portrayed as a distant sound, a feelingof hearing things in echo, or the sensation that his right earwas obstructed. Vertigo (3%) and arrest of speech (3%)were rare. Heart rates prior to and during stimulationrevealed no significant change.

Correlation between the site of stimulation and symptomsevoked

The topographic analysis of our responses was per-formed by pooling sites where responses were obtained ona single sagittal insular image. Viscerosensory-evoked sen-sations were all located in the anterior portion of the insula(Fig. 3B), while somatosensory-evoked responses werewidely distributed (Fig. 3A). Interestingly, somatosensorysymptoms involving the facial area (eyes, nose, mouth,neck) were all located in the anterior insula, while somato-sensory symptoms involving limbs were more posteriorlylocated. Motor responses (Fig. 3C) involving ocular move-ments were situated in the anterior insula, while limbmovements were in the posterior insula. The only vestibu-lar response noted was positioned in the anterosuperiorinsula (Fig. 3D). Auditory symptoms were in the posteriorinsula (Fig. 3D). Finally, the two sites that evoked lan-guage difficulties were in the anterosuperior insula in thedominant hemisphere (Fig. 3D).

Discussion

The work of Guillaume & Mazar (1949a, 1949b), fol-lowed by Penfield & Jasper (1954) paved the way to theconcept of insular cortex epilepsy. Based on the presenceof rich electrocorticographic interictal spiking found in TLepilepsy patients and the similarity of insular intraopera-tive stimulation-evoked symptoms to TL semiology, theysuggested that temporal lobectomy failures could be due tothe lack of recognition of insular seizures mimicking TLseizures (Guillaume et al., 1953; Penfield & Faulk, 1955).This concept eventually fell into disuse when Silfveniusand coworkers (1964) showed that insular resection, whenadded to temporal lobectomy, failed to increase epilepticcontrol, while significantly increasing surgical morbidity.Interest in insular cortex epilepsy was however reignitedby Isnard et al. (2000) who reported the first intracorticalrecording of insular seizures in two out of 21 atypical TLepileptics implanted due to the presence of ictal symptomsor scalp EEG data suggestive of an early spread of seizuresto the opercular cortices. These two patients continued tohave seizures following temporal lobectomy. Apart fromresembling TL epilepsy, insular epilepsy may also mimicPL epilepsy (Cascino & Karnes, 1990; Aghakhani et al.,2004; Isnard et al., 2004). Benefiting from three additionalcases and findings from insular cortical stimulation, thesame group from Lyon later concluded that insular seizureswere typically associated with a sensation of laryngeal

516

D. K. Nguyen et al.

Epilepsia, 50(3):510–520, 2009doi: 10.1111/j.1528-1167.2008.01758.x

Figure 3.

Distribution of stimulated sites according to categorical evoked responses displayed in axial and sagittal views.

Categorized responses are filled with selective colors. Right insular contacts are represented by circles and left

insular contacts by squares. Numbers in semibold italic indicate evoked discharges that were followed by a brief

localized afterdischarge. (A) Somatosensory responses are in green. (B) Visceral responses are in light blue. (C)

Motor association area responses are in navy blue. (D1) Vestibular responses are in purple. (D2) Auditory

responses are in red. (D3) Language or speech arrest responses are in orange.

Epilepsia ILAE

517

Intracerebral Study of Refractory Insular Cortex Epilepsy

Epilepsia, 50(3):510–520, 2009doi: 10.1111/j.1528-1167.2008.01758.x

constriction and paresthesias with or without dysarthria,auditory hallucinations, or motor signs (Isnard et al.,2004). The same year, Aghakhani et al. (2004) alsoreported six very interesting cases with electrographicalfindings suggestive of TL epilepsy but also with somato-sensory auras suggestive of PL epilepsy. Four patients hadpreresection intracranial EEG monitoring, which sug-gested an epileptogenic zone in the posterior temporal andinferior parietal area in two patients, in the TL in onepatient, and was inconclusive in the last patient. A secondintracranial study was performed in three of the patientsafter a first anterior temporal resection and suggested aposterior temporal neocortical localization in two patientsand a posterior temporal-inferior parietal localization inone patient. Overall, only limited improvement occurreddespite one to four operations for each of these six patients,leading the authors to conclude that anteromesial resectionwas ineffective for patients with posterior temporoparietalclinical ictal features. They acknowledged that the insulamight have participated in the complex epileptogenic net-works observed in their patients, although they could notconfirm it due to the lack of intracerebral electrodes placeddirectly within the insula. Finally, recent reports suggestthat insular cortex epilepsy may emulate FL epilepsy.Ryvlin and coworkers (2006) reported three patients withrefractory nonlesional drug-resistant nocturnal hypermotorseizures, whose intracerebral EEG ictal onset primarilyinvolved the insula. Kaido et al. (2006) also reported twosimilar cases of hyperkinetic seizures, but associated witha slight signal change in the right posterior ventral insularcortex, and only cured once the posterior ventral insularand lateral temporal cortices were resected. More recently,Dobesberger et al. (2008) reported a nonlesional patientwith an insular seizure onset and nocturnal hypermotor sei-zures who became seizure-free following a limited resec-tion of the anterior part of the right insula and frontaloperculum. Findings from insular cortical stimulation alsosupport the notion that the insula may generate misleadingsymptoms. From 1945 to 1953, Penfield stimulated 82 sep-arate insular points from 36 awake patients after temporallobectomy (Penfield & Jasper, 1954; Penfield & Faulk,1955). Forty percent of the stimulations produced visceralresponses, while another 40% resulted in various sensoryresponses. The Lyon group has also reported, morerecently, the clinical responses obtained by stimulating upto 144 insular sites in 50 patients with atypical TL epilepsystudied by stereoelectroencephalography (Ostrowskyet al., 2000, 2002; Isnard et al., 2004). They found (indecreasing order of frequency) somatosensory, viscerosen-sitive, and auditory responses, as well as dysarthria andother miscellaneous responses.

In our series of four patients with insular seizures, two hadTL-like epilepsy, one had FL-like epilepsy, and one had PL-like epilepsy. Insular cortical stimulations evoked responsesthat were coherent with data found in the literature, as all

previously reported responses were reproduced: Somatosen-sory including pain, viscerosensitive including gustatory,motor association area-evoked responses, vestibular, audi-tory, and language disturbances. Similar to Isnard et al.(2004), visceral responses were mainly evoked from theanterior insula. However, sites evoking somatosensorysymptoms were widely distributed and not clearly limited tothe posterior three-quarters of the insula as previouslyreported (Isnard et al., 2004). Such diversity in spontaneousseizure semiology and the fact that the insula may generatesuch a variety of responses following cortical stimulationsupport the notion that insular cortex epilepsy is a greatmimicker and that some prior cases of refractory TL-, PL-,or FL-like epilepsy could have had a poor surgical outcomefrom lack of insular lobe intracranial sampling.

Recognizing insular cortex epilepsyAlthough the insula may seem like a great mimicker,

careful analysis of our data and the literature indicate that itis possible to recognize insular cortex epilepsy in mostcases. Due to the confluence of functions in a restrictedregion, as demonstrated by cortical stimulation, we believethat insular seizures should be suspected whenever visceralor motor and especially somatosensory symptoms are com-bined early into a seizure. Hence, independent insular sei-zures should be suspected (1) in TL-like epilepsy patientsif there is early occurrence of somatosensory (e.g., laryn-geal discomfort, throat constriction, limb paresthesias) ormotor ictal symptoms (e.g., arm elevation, trashing, or ped-aling) preceding or concomitant to typical mesiotemporallobe symptoms (e.g., d�j� vu); (2) in PL-like epilepsypatients especially if paresthesias are restricted to perioralor intraoral areas, distributed to a large cutaneous territory,or bilateral; (3) in FL-like epilepsy patients in the presenceof occasional somatosensory symptoms prior to hyper-motor manifestations. Because most patients are usuallyunable to remember any of the subective symptoms experi-enced during nocturnal seizures, the clinician needs to bevigilant about the aura reported during the rare diurnalspells patients might experience. Similar to case 5 in ourseries, patient 1 in Ryvlin’s series (2006) reported an initialtingling sensation, while the two patients reported byRoper et al. (1993) respectively felt throat butterflies andleft arm/leg tingling prior to complex motor behavior.

Investigation of insular cortex seizuresAs seen in our series and the Lyon series (Isnard et al.,

2004), scalp EEG is unable to differentiate insular seizuresfrom TL, PL, or FL seizures. MRI should be carefullyreviewed for any suspicious signal changes, as found inpatient 5 and the two patients reported by Kaido et al.(2006), as they may be easily overlooked or regarded asnonspecific/unrelated to the epileptic disorder. Rather thanproducing unequivocal insular activation, ictal SPECT willmost likely reveal multisite activations including the

518

D. K. Nguyen et al.

Epilepsia, 50(3):510–520, 2009doi: 10.1111/j.1528-1167.2008.01758.x

insula. The latter should however raise the level of suspi-cion in the right clinical context. Once insular seizures aresuspected, and in absence of a congruent epileptogeniclesion, confirmation by intracerebral ictal recordings is nec-essary, since electrocorticography is unreliable (Silfveniuset al., 1964; Isnard et al., 2000). Considering the cost andrisk of intracranial studies, it may be preferable to identifythe subset of patients for whom insular intracerebral elec-trodes would yield best results rather than generallyimplant more patients. Better characterization of insularcortex epilepsy will eventually permit to do so. For themoment, we submit that sampling of the insula should beconsidered in patients with nonlesional PL- or TL-like epi-lepsy exhibiting a combination of visceral or motor, andparticularly somatosensory, symptoms at seizure onset.For patients with nonlesional FL-like epilepsy, identifyingthose with insular origin is more difficult. Surely, patientswith nonlesional FL-like epilepsy reporting occasionalsomatosensory symptoms during diurnal seizures, withinnocuously-looking millimetric MRI signal changes, orwith any type of presurgical functional imaging data sug-gesting insular involvement could benefit from insularintracranial sampling.

For patients with atypical TL-like epilepsy associatedwith hippocampal sclerosis, the decision to proceed to anintracranial study with sampling of the insula is less clearand should at the present time be individualized based onthe strength of the hypothesis produced by the noninvasiveevaluation. One could argue that these patients should firstundergo a temporal lobectomy considering the relativelylower incidence of insular cortex epilepsy and the risk ofan intracranial study. An intracranial study would then onlybe performed in posttemporal lobectomy failures. In theseries by Isnard et al. (2000), all 13 patients with lesional[i.e., hippocampal sclerosis, dysembryoplastic neuroepi-thelial tumor (DNET), or hyperintense T2 signal] atypicalTL epilepsy had a good postsurgical outcome followingtailored temporal lobectomy, anterointernal lobectomy,c-knife internal lobectomy, or temporal lobectomy (with-out insular resection). On the other hand, others wouldfavor an invasive investigation prior to surgery to identifycases of ‘‘temporal plus epilepsy,’’ a term coined by Ryvlin& Kahane (2005) to better delineate specific forms of mul-tilobe epilepsy characterized by a prominent ictal involve-ment of the temporal lobe, electroclinical featuresprimarily suggestive of TL epilepsy, and MRI findings thatare either unremarkable or show signs of hippocampalsclerosis. In our series, two of the patients with insular sei-zures also had a temporomesial epileptogenic zone. In theseries by Isnard et al. (2000), both their patients with insu-lar seizures also had temporal lobe seizures.

Insular cortex epilepsy: Not so rare?By being increasingly aware of the various faces of insu-

lar cortex epilepsy and broadening our indications for insu-

lar intracerebral electrodes, our group found four patientswith insular seizures from a series of 18 consecutive intra-cranial studies. In comparison, Isnard et al. (2004) foundfive cases of stereotactically proven insular epilepsy froma series of 50 patients. The discrepancy in prevalence couldbe related to differences in patient selection for intracranialstudies. The latter series included only atypical TL epi-lepsy patients as in ours, all intracranial studies (temporaland extratemporal) were reviewed. Out of our 18 patients,four (patients 3, 4, 5, and 8) were suspected of having a me-siofrontal epileptic focus, and implantation was requireddue to the lack of congruent noninvasive investigations.Insular intracerebral electrodes were justified due toreports of insular nocturnal hypermotor seizures (Kaidoet al., 2006; Ryvling et al., 2006; Dobesberger et al., 2008)and because patient 3 and 5 reported occasional thoracicoppression and rare diurnal hand paresthesias, respec-tively, prior to hypermotor features. Four patients (patients1, 6, 7, and 10) were suspected to have a TL epileptic focus,but implantation of intracerebral electrodes with samplingof the insula was justified by early throat paresthesia inpatient 1, absence of an epileptogenic lesion in patients 6and 7, and left hippocampal atrophy noncongruent with ic-tal recordings in patient 10. Two patients (patients 2 and 9)were suspected to have PL epilepsy based on ictal semiol-ogy. Implantation of intracerebral electrodes was neces-sary for patient 2, due to inadequate localization fromnoninvasive tests. As for patient 9, ictal semiology sug-gested a parietal focus rather than a frontal focus, wherethe suspected dysplastic lesion was found.

Despite confirmation by intracerebral recordings in fourof our patients, we acknowledge that final proof of aninsular focus by seizure freedom postinsular resection wasnot demonstrated; one patient declined insular resection,another became seizure-free only after adjunctiveradiosurgery with a limited 4-month seizure-free status atthis time, and two patients also had anterior temporallobectomies in addition to insular resections. It is uncertainwhat would have happened to these patients without theinsular resection, but because both patients with indepen-dent insular and TL seizures reported by Isnard et al.(2000) continued to have insular seizures following tempo-ral lobectomies, we elected to proceed with the additionalinsular resection.

Conclusion

A nonnegligeable proportion of surgical candidates withdrug-resistant epilepsy have an epileptogenic zone thatinvolves the insula. Our observations made from the evalu-ation of four patients with insular cortex epilepsy andextraoperative cortical stimulation in nine subjectsimplanted with insular intracerebral electrodes add to thegrowing body of literature of this neglected localization-related syndrome. Increased awareness of clinical

519

Intracerebral Study of Refractory Insular Cortex Epilepsy

Epilepsia, 50(3):510–520, 2009doi: 10.1111/j.1528-1167.2008.01758.x

characteristics of insular cortex epilepsy should hopefullyreduce the number of epilepsy surgery failures.

Acknowledgments

The authors would like to thank all the EEG technicians (N. L�vesque,J. Forand, H. Cossette, S. Lebrun, and G. Imbeault). This work wassupported by a Clinical Scientist Research Bursary from the Centre deRecherche du CHUM awarded to D.K.N.

Conflict of interest: We confirm that we have read the Journal’s positionon issues involved in ethical publication and affirm that this report is con-sistent with these guidelines. The authors have no conflicts of interest todisclose.

References

Aghakhani Y, Rosati A, Dubeau F, Oliver A, Andermann F. (2004) Patientswith temporoparietal ictal symptoms and inferomesial EEG do notbenefit from anterior temporal resection. Epilepsia 45:230–236.

Cascino GD, Karnes WE. (1990) Gustatory and second sensory seizuresassociated with lesions in the insular cortex seen on magnetic reso-nance imaging. J Epilepsy 3:185–187.

Collins DL, Zijdenbos AP, Kollokian V, Sled JG, Kabani NJ, Holmes CJ,Evans AC. (1998) Design and construction of a realistic digital brainphantom. IEEE Trans Med Imaging 17:463–468.

Dobesberger J, Ortler M, Unterberger I, Walser G, Falkenstetter T,Bodner T, Benke T, Bale R, Fiegele T, Donnemiller E, Gotwald T,Trinka E. (2008) Successful surgical treatment of insular epilepsywith nocturnal hypermotor seizures. Epilepsia 49:159–162.

Guillaume MMJ, Mazars G. (1949a) Cinq cas de foyers �pileptog�nesinsulaires op�r�s. Soc FranÅaise de Neurol 766–769.

Guillaume MMJ, Mazars G. (1949b) Technique de r�section de l’insuladans les �pilepsies insulaires. Rev Neurol 81:900–903.

Guillaume MMJ, Mazars G, Mazars Y. (1953) Indications chirurgicalesdans les �pilepsies dites ‘temporalis.’ Rev Neurol 88:461–501.

Isnard J, Guenot M., Ostrowsky K, Sindou M, Maugui�re F. (2000) Therole of the insular cortex in temporal lobe epilepsy. Ann Neurol48:614–623.

Isnard J, Guenot M, Sindou M, Maugui�re F. (2004) Clinical manifesta-tions of insular lobe seizures: a stereo-electroencephalographic study.Epilepsia 45:1079–1090.

Kaido T, Otsuki T, Nakama H, Kaneko Y, Kubota Y. (2006) Complexbehavioral automatism arising from insular cortex. Epilepsy Behav8:315–319.

Ostrowsky K, Isnard J, Ryvlin P, Gu�not M, Fischer C, Maugui�re F.(2000) Functional mapping of the insular cortex: clinical implicationin temporal lobe epilepsy. Epilepsia 41:681–686.

Ostrowsky K, Magnin M, Ryvlin P, Isnard J, Gu�not M, Maugui�re F.(2002) Representation of pain and somatic sensation in the humaninsula: a study of responses to direct electrical cortical stimulation.Cerebral Cortex 12:376–385.

Penfield W, Jasper WW. (1954) Epilepsy and the functional anatomy ofthe human brain. Little Brown, Boston.

Penfield W, Faulk ME. (1955) The insula: further observations on itsfunction. Brain 78:445–470.

Roper SN, L�vesque MF, Sutherling WW, Engel J Jr. (1993) Surgicaltreatment of partial epilepsy arising from the insular cortex. J Neuro-surg 79:266–269.

Ryvlin P, Kahane P. (2005) The hidden causes of surgery-resistant tem-poral lobe epilepsy: extratemporal or temporal plus? [editorialreview]. Curr Opin Neurol 18:125–127.

Ryvlin P. (2006) Avoid falling into the depths of the insular trap. Epilep-tic Disord 8(Suppl 2):37–56.

Ryvlin P, Mnotu L, Demarquay G, Hirsch E, Arzimanoglou A, HoffmanD, Gu�not M, Picard F, Rheims S, Kahane P. (2006) Nocturnal hyper-motor seizures, suggesting frontal lobe epilepsy, can originate in theinsula. Epilepsia 47:755–765.

Silfvenius H, Gloor P, Rasmussen T. (1964) Evaluation of insular abla-tion in surgical treatment of temporal lobe epilepsy. Epilepsia 5:307–320.

520

D. K. Nguyen et al.

Epilepsia, 50(3):510–520, 2009doi: 10.1111/j.1528-1167.2008.01758.x