5Journal ofNeurology, Neurosurgery, and Psychiatry 1994;57:58-65

Utility of stereoelectroencephalography inpreoperative assessment of temporal lobe epilepsy

C D Binnie, R D C Elwes, C E Polkey, A Volans

AbstractOf 269 consecutive patients entered intoa preoperative assessment programmefor possible surgical treatment ofepilepsy, 33 had intracranial recording(SEEG) with combined subdural anddepth electrodes for the purpose of local-ising a suspected temporal site of seizureonset. The findings in these patients areanalysed with particular reference to: 1)the criteria of selection for SEEG andtheir validity; 2) information on SEEGcompared with that obtained by lessinvasive means, including foramen ovaletelemetry; 3) information on the use ofintracerebral electrodes compared withsubdural placements; 4) possible predic-tors of failure of localisation by SEEGand of surgical outcome. It was con-cluded that SEEG had usefully con-tributed to the management of 690/o ofthe patients in whom it was used, estab-lishing a previously unidentified site ofseizure onset in 33%, correcting anerroneous localisation in 15%, and estab-lishing inoperability in 21% of patients.No predictors of failure of SEEG or ofsurgery emerged; thus there was noevidence of unnecessary use of this pro-cedure. Five patients were found withincorrect lateralisation of seizure onseton foramen ovale recording (of a total of192 foramen ovale telemetries).Localisation of the ictal onset zoneeither by the distribution of inter-ictaldischarges or by the initial ictal changesat subdural electrodes was unreliable,confirming the need for ictal, depthrecordings.

(7 Neurol Neurosurg Psychiatry 1994;57:58-65)

Departments ofClinicalNeurophysiology andNeurosurgery,Maudsley Hospital,London, UKC D BinnieR D C ElwesC E PolkeyA VolansCorrespondence to:Dr Binnie, NeurosciencesDepartment, MaudsleyHospital, Denmark Hill,London SE5 8AZ, UKReceived 26 June 1992 andin final revised form9 November 1992.Accepted 19 November 1992

Successful treatment of epilepsy by resectivesurgery depends on removal of tissue which isboth structurally and functionally abnormaland within which ictal activity is initiated.Despite recent developments in functionalneuroimaging, electrophysiology remains themost important means of identifying suchlocalised dysfunction.

In 1986, an international survey of centresengaged in surgical treatment of epilepsy'indicated marked variations in the use ofelectrophysiology in preoperative assessment.Techniques have long been available foridentifying sites of seizure onset by

stereoelectroencephalography (SEEG), usingstereotactically positioned intracerebral elec-trodes, sometimes in combination with sub-dural contacts. However, in the past manycentres have achieved good results withoutintracranial recording, except acute electro-corticography24; some continue to do so' oruse SEEG rarely. Differences of approachmay reflect differing patient populationsrather than conflicting philosophies. Resectiveprocedures in programmes making little or nouse of invasive assessment are, or were, per-formed mainly on the temporal lobes, formesial temporal sclerosis in particular.48 Thefeasibility of identifyfing the epileptogeniczone from the scalp EEG in these patients isclearly demonstrated, both by the veryacceptable surgical results reported from cen-tres not using SEEG, and by confirmation ofEEG localisation by intracranial ictal record-ing in similar subjects.9 Other workers haveemphasised rather the lack of agreementbetween scalp and depth recording."011Recently, however, there has been some con-vergence of practice, and most centresemploy SEEG in 10-40% of patients. 12

Uncertainty remains about criteria forSEEG and concerning the choice of technol-ogy, specifically whether complex partialseizures should be investigated with subduralor depth electrodes, or with a combination ofthese.

Identification of the epileptogenic zone isonly one requirement for successful treat-ment; surgical outcome is the criterion bywhich preoperative assessment should bejudged; in practice this is difficult. Because ofthe differing patient populations and surgicaltechniques it is hard to compare results fromcentres with different programmes; indeed afeed-back seems to operate whereby theselection criteria, and hence the case-mix, indifferent centres are adjusted in such a way asto obtain very similar results overall.However, a single centre controlled trial ofalternative programmes presents insuperableethical problems, as this would necessarilyrequire the surgeon concerned to act againsthis own judgment, by performing apparentlyunnecessary invasive recording, or operatingwith electrophysiological data consideredinadequate.The best evidence concerning indications

for invasive neurophysiology is probably thatgained by clinical audit in centres followingdefined preoperative protocols, in a variedand specified patient population. Such astudy is reported here.

58

on Novem

ber 26, 2020 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.57.1.58 on 1 January 1994. Dow

nloaded from

in preoperative assessment of temporal lobe epilepsy

Material and methodsThis report concerns a consecutive series of33 patients with seizures of presumed orproven temporal lobe origin investigated witha combination of depth and subdural elec-trodes. In 26 patients an ictal onset zone inone temporal lobe was established byrepeated ictal SEEG recording; in seven sub-jects a single site of electrographic seizureonset was not established. During the studyperiod, the five years ending December 1991,the preoperative assessment protocol was, asfar as possible kept constant, and operativeprocedures for temporal lobe resectionschanged little.

During the same period, 269 patients withmedically intractable epilepsy were assessedfor surgery. Forty one had hemispherectomyor partial callosal section, following a protocoldifferent from that for possible local resec-tions. All of the remaining 228 had wake andsleep EEGs. Detailed neuropsychologicalassessment was performed in all subjects, anda carotid amytal test (following the proceduredescribed by Powell et al3) in all those con-sidered for temporal resections. Ictal scalprecordings were performed in 199, in sevenwith sphenoidal, and in 192 with foramenovale electrodes. Forty-eight had more inva-sive recordings with at least subdural elec-trodes, and all in the present study groupwere investigated by SEEG, with a combina-tion of subdural and depth electrodes (table1). All 33 study group subjects had under-gone a Wada test and had temporal lobe ori-entated CT scans, nine had been investigatedby MRI.Of the 180 candidates for resection in

whom more invasive recording was not con-sidered to be indicated, 140 had resectivesurgery, and 30 were deemed inoperable. Tenwithdrew themselves from the preoperativeassessment programme.The algorithms for selecting patients for

subdural or depth (SEEG) recording aresummarised in tables 2a-c.

Assessment of scalp EEG andforamen ovaletelemetryThe EEG and telemetry studies used theMaudsley electrode placement system, whichresembles the 10-20 system but providesmore extensive cover, particularly of the ante-rior temporal region.'4 Sleep recording wasperformed in all subjects, induced by sodium

Table 1 Patient population

AssessedCallosotomy/hemispherectomyAssessed for local resectionTelemetry performed (n = 199)

with foramen ovale electrodeswith sphenoidal electrodes

Subdural/depth recording not performed (n = 180)operatedinoperablepatient withdrew from programme

Subdural/depth recording performed (n = 48)extratemporal ictal onset zonecps ofpresumed temporal origin (n = 33: study group)no single ictal onset zone foundsingle temporal ictal onset zone

26941228

1927

1403010

15

7

26

quinalbarbitone, unless they slept sponta-neously. A localising focus of epileptiformactivity was considered to be present if spikesor spike wave activity occurred with a consis-tent maximum at a single electrode or adja-cent electrodes, exceeding the discharge rateat any other site in a ratio of at least 4/1.

Assessment of foramen ovale (FO) teleme-try, for purposes of this study was confined toictal events. A localised electrographic seizureonset was considered to have been detected ifthe initial changes (spikes, spike-and-wave, ora new rhythmic activity of theta, alpha or betafrequency) were consistently of maximumamplitude over one lobe of the brain.

Table 2a General criteria for admission to preoperativeassessment programme

1 Reliable diagnosis of intractable epilepsy:-Attacks are epileptic, no pseudoseizuresFailure of appropriate medicationPatient is compliant

2 Seizures of such a frequency and nature that they aredisabling, having regard to the patient's life style

3 No other contraindication, for example, a coagulationdefect

Table 2b Criteria for investigation for resective surgeryandlor multiple subpial transection

1 Partial seizures, whether or not secondarily generalised2 Does not meet criteria for hemispherectomy:-

a) Long-standing unilateral hemispheric damageevidenced by hemiplegia and brain-imaging changes

b) Partial seizures, whether or not secondarily generalisedapparently arising from the diseased hemisphere

3 Full scale IQ not less than 704 Age not more than 55 years5 Patient has the emotional resources to withstand the

procedures, including the possibility of being found inop-erable

Table 2c Specific criteria for subdural or SEEG recording

In addition to meeting general admission requirements andspecific criteria for assessment with a view to resective surgery,patients selected for invasive recording satisfy all of thefollowing:-A) The patient does NOT meet any of the followingcriteria which would allow surgery to be offered withoutsubdural or SEEG recording:-1 (For lesionectomy) A discrete cerebral lesion, demonstratedby imaging,

concordant withSeizure patternSite of interictal EEG dischargesAbnormalities of background interictal EEGNeuropsychological findings.

and notAt an eloquent site which demands functional mapping.

2 (For temporal resection) A consistent, single, temporal siteof electrographic seizure onset, at or preceding clinical onseton foramen ovale telemetry,

concordant withSeizure patternAbnormalities of background EEGBrain imaging findingsNeuropsychological findings.

B) The patient does NOT show the following featureswhich would exclude resective surgery:-3 Predominantly generalised interictal EEG discharges in the

absence of a discrete lesion on neuroimaging.4 Independent sites of electrographic seizure onset

demonstrated in more than one lobe by telemetry withforamen ovale electrodes.

5 Generalised epileptiform discharges at or preceding clinicalseizure onset.

6 A probable site of seizure onset, which cannot be resectedwithout unacceptable complications, and is consideredunsuitable for multiple subpial transection.

59 on N

ovember 26, 2020 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.57.1.58 on 1 January 1994. D

ownloaded from

Binnie, Elwes, Polkey, Volans

Introduction of electrodes and recordingThe electrode technology employed isdescribed in detail by Van Veelen et al." Inbrief, all electrodes were inserted throughbilateral fronto-central trephine holes. Thoseplaced in the subdural space had seven 5-mmcontacts on 10 mm centres and were of twotypes: thin (0'3 mm) flexible bundles, and flat"reeds" 3 mm in width (considerably nar-rower than conventional subdural strips)which were compliant in only one dimensionand therefore sometimes easier to insert inthe required direction. From 7 to 10 subduralbundles or reeds were placed under fluoro-scopic control, in all subjects, including aminimum of one mid-temporal, one lateralfrontal and one fronto-polar-orbital place-ment on each side.The procedure for placement of depth

electrodes differed from the method of VanVeelen et al," in that CT controlled stereo-taxy was used. Targets within mesial tempo-ral structures were identified using CT scansorientated along the axes of the temporalhorns, the gantry being tilted, typically at 200,from the horizontal. Temporal target coordi-nates were calculated taking account of theangle of the gantry during imaging.'6 Six-con-tact depth electrode bundles, 4 to 8 in num-ber, were implanted bilaterally, through thesame fronto-central trephine holes as wereused for the subdural electrodes. They wereplaced in amygdala, anterior and/or posteriorhippocampus and, as indicated, tangentiallythrough the medial frontal cortex and ante-rior cingulum to the gyrus rectus.

Recording started when the patient hadrecovered from the operation three to sevendays following implantation and continuedfor up to three weeks, mean 8-6 days.Antiepileptic medication was reduced orwithdrawn in all cases. Cable telemetry of 32or 64 channels was used (TelefactorMODAC or Beekeeper systems) with simul-taneous video monitoring. During the last 18investigations of the series automatic seizurerecognition was employed using an algorithmbased on that of Gotman'7 (Telefactor SeZacsystem).

Surgery and assessment of outcomeThose patients in the study group who weretreated surgically had either en bloc temporallobe resection or amygdalohippocampectomyby the trans-Sylvian route.'8 Outcome wasassessed by the system of Engel,' whichdivides seizure relief into four grades (eachwith two to four subgroups, a-d), and grade5, fatality. In brief: grade 1 seizure-free; 2rare seizures; 3 worthwhile improvement(75% seizure reduction or prolonged seizure-free intervals of more than 2 years); 4 unim-proved; 5 died. For purposes of the presentanalysis, grades 1 and 2 are regarded as a suc-cessful outcome.

ResultsThe 48 patients investigated with subdural ordepth recording included 15 with a proven

extra-temporal onset, which had beenexpected before SEEG. They are excludedfrom the study population as they do notform an homogeneous group, and are too fewto analyse as sub-groups. The study group istherefore restricted to the patients with a pre-sumed temporal seizure onset, and reasonsfor undertaking invasive recording will beconsidered in this sample of 33. Results con-cerning the utility of SEEG and comparisonof SEEG with other data relate only to the 26subjects with a single proven temporal ictalonset zone.

Reasons for intracranial recordingThe case records were reviewed to determinewhat considerations had led to SEEG in these33 patients (table 3). None met criteria 1-6of table 2c. Seven with localised radiologicalabnormalities (two frontal, one thalamic, oneparietal, one occipital and two temporal)failed to meet criterion 1 (for lesionectomy),as the site of the demonstrated lesion was notconcordant with other evidence. The remain-der failed criterion 2 (consistent temporalonset) because of:- a) Bilateral electrographicseizure onset on FO telemetry.'4 b) "Delayedelectrographic onset",'2 that is, the appear-ance during foramen ovale telemetry of clini-cal ictal phenomena before the first localisedelectrographic change (including an electro-decremental event or other "non-epilepti-form" change). c) "Cross-over"9 that is, rapidappearance of bilateral or contralateral dis-charges, within 1 second of an apparently lat-eralised seizure onset at scalp or foramenovale electrodes. d) Apparent extra-temporalelectrographic onset.5 e) Non-concordancebetween electrographic ictal onset and con-sensus of other evidence.'0 f) Clinical seizurepattern non-concordant with consensus ofother evidence.3 Multiple reasons were citedin 10 patients.

Localisation of seizure onset by SEEGDetailed SEEG findings will be presentedonly in the 26 patients with reliably localisedtemporal ictal onset. Two hundred and sixtysix seizures were captured in these subjects;sites of ictal onset on SEEG are summarisedin table 4.

Fourteen patients showed, in addition toictal onsets at one main site, a minority ofseizures starting at adjacent electrodes in thesame temporal lobe ("minor sites" in table 4).

Table 3 Reasons for selection for SEEG; see criteria 1-2of table 2c. No patient was clearly unsuitable for resectivesurgery by criteria 3-6. These reasons are not mutuallyexclusive; 10 patients failed on multiple criteria

Reasons for selection for SEEG1 Lesion on imaging not concordant with consensus of

clinical, psychological and electrophysiologicalevidence

2 Ictal onset on telemetry:-not located:

bilateral onsetrapid "cross-over"

"delayed onset"not concordant with other evidencenot temporal

3 Clinical findings not concordant with other evidence

7

14912753

60 on N

ovember 26, 2020 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.57.1.58 on 1 January 1994. D

ownloaded from

Utility ofstereoelectroencephalography in preoperative assessment of temporal lobe epilepsy

Table 4 Sites ofseizure onset in SEEG.

Sites ofonset

Main Minor Subclinical

Amygdala 2 1 0Anterior hippocampus 11 6 9Posterior hippocampus 2 1 1Anterior and posterior hippocampus 1 0 0Hippocampus and amygdala 5 1 4Lateral temporal 2 1 1Lateral temporal and hippocampus 1 0 0Parietal and posterior hippocampus 1 1 1Amygdala, anterior hippocampus

and lateral temporal 1 0 0Frontal and anterior hippocampus 0 1 0Frontal 0 2 0

Seizure onset appeared to be localised to asingle contact in 18 patients. The initial ictaldischarge was confined to the anterior hippo-campus in 11 and detected simultaneously inanterior hippocampus and amygdala in five.Anterior hippocampal onsets of some or allseizures occurred in 19 subjects, amygdaloidin eight. There was an initial lateral neocorti-cal involvement in the majority of seizures infive patients, three of whom showed aregional onset in neocortex and hippocam-pus.The initial discharge consisted of fast activ-

ity at 16 to approximately 50 Hz in 15patients, and of a rhythmic build up of spikesstarting at less than 13 Hz in six. Slow peri-odic spikes preceding a faster rhythmic ictaldischarge, as described by Spencer et al,'9were rare, and occurred in a majority ofstwures in only two patients, both with aregional seizure onset including neocortex.Neuropathological findings are now availablein 19 patients: in 13 out of 15 with initial fastactivity a clear abnormality was found,whereas specific pathology was present inonly one of four patients who did not showthis type of ictal onset, a finding in accor-dance with the observations of Spencer et al.19

Three subjects suffered from one to threeseizures arising contralateral to the main ictalonset zone (not included in the analysisabove). In all instances, these occurred late inthe course of the investigation, several daysafter antiepileptic drug withdrawal, andwithin 30 minutes of a preceding seizure, thepostictal electrophysiological sequelae ofwhich had not yet resolved.

Sixteen patients exhibited sub-clinical elec-trographic seizures, which corresponded inboth topography and electrographic pattern

Table 5 Lateralisation of interictal and ictal discharges with respect to seizure onset inSEEG

Lateralisation with respect to ictal SEEG

Not ContralateralPredominandy lateralisedipsilateral or absent Predominant Exclusive

InterictalSubdural discharges 7 11 8 0Depth discharges 19 1 6 0

Ictal onset, subdural 16 4 4 2Mid-ictal (depth and

subdural electrodes) 14 4 4 4Late ictal (depth and

subdural electrodes) 10 8 4 4Postictal (depth and

subdural electrodes) 3 18 1 4

to the initial events of overt seizures at theictal onset zone. Thirteen of these subjectseach exhibited over 50 such episodes in 2-21days recording.

Subdural electrodes were involved atseizure onset in five patients. Discharges ofmesial temporal origin propagated first to theipsilateral neocortex in 11 subjects but to thecontralateral convexity (via the hippocampus)in six. This early propagation to contralateralsubdural contacts occurred consistently intwo patients, and in the majority of seizuresof four subjects (table 5). In three subjectssubdural discharges appeared simultaneouslyover both hemispheres, following an earlypropagation to the contralateral hippocam-pus. In one patient the ictal discharges wereconfined to depth electrodes. Thus lateralisa-tion relying on the first ictal events detectedat subdural electrodes would have been incor-rect in six patients and impossible in four.

Mid-ictal discharges at the stage of maxi-mum generalisation, if any, and late ictal dis-charges in the last 10 seconds of the seizure,were with the exception of one patient, gener-ally present at both subdural and deep elec-trodes, predominating ipsilateral to seizureonset in 14 out of 26 subjects (table 5). Infour patients, following an early propagationto the contralateral mesial temporal struc-tures, the subsequent mid- and late ictal dis-charges showed a consistent contralateralpreponderance. Postictal slowing was cor-rectly lateralising to the side of seizure onsetin only three subjects, and usually contralat-eral in five.

Interictal epileptiform activity appearedbilaterally at deep electrodes in all patients:this showed a clear lateralised preponderancein all but one subject, and was maximal ipsi-lateral to the site of seizure onset in 19,contralateral in six. Lateralising interictal sub-dural discharges, by contrast, were present inonly 15 patients, with a maximum ipsilateralto seizure onset in only seven.

Evidence of site of seizure onset before SEEGIn this group of patients, lateralisation of thesite of SEEG seizure onset by less invasivemeans was conspicuously unsuccessful (table6). The carotid amytal test showed the great-est reliability, the onset zone being in themore dysfunctional hemisphere in 10instances and contralateral in only one.Lateralising abnormalities on general neu-ropsychological assessment were uncommon,but were also fairly reliable (six implicatingthe ipsilateral, and one the contralateralhemisphere).

Lateralisation by routine wake and sleepEEG was random: 18 patients had lateraliseddischarges, but the lateralisation agreed withthe findings on SEEG in only nine.Lateralisation by foramen ovale telemetryagreed with SEEG findings in 10 patients anddisagreed in five. These five failures of lateral-isation by FO telemetry were explicable bythe SEEG findings. This group included bothof the patients with an ictal onset zone con-fined to the amygdala, which has a closed

61 on N

ovember 26, 2020 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.57.1.58 on 1 January 1994. D

ownloaded from

Binnie, Elwes, Polkey, Volans

Table 6 Lateralisation by non-invasive methods with respect to seizure onset in SEEG

Lateralisation by non-invasive methods

Ipsilateral Not lateralised Contralateral

Clinical ictal pattern 6 16 4CT and/or MRI 3 23 0Psychometrics 6 19 1Carotid amytal test 10 15 1Non-invasive neurophysiology:Wake and sleep EEG 9 8 9FO telemetry 10 11 5

Concordant findings:Telemetary and EEG 4 NA 0Clinical and EEG 4 NA 1Clinical and telemetry 3 NA 2Clinical and amytal 3 NA 0EEG and amytal 4 NA 1Telemetry and amytal 3 NA 0

NA = not applicable.

electrical field not detectable from without.20All five showed rapid propagation to the con-

tralateral hippocampus. FO telemetry hadshown clinical ictal onset before the firstapparent electrographic changes and rapidcross-over (see above). These, indeed, were

the features which had led to selection forSEEG.

Concordant lateralisation on interictalEEG and FO telemetry was found in onlyfour instances, but agreed with SEEG in all ofthese. Clinical and EEG lateralisation were

concordant in five subjects, and disagreedwith SEEG findings in one of these.

Lateralising abnormalities on neuroimag-ing were found ipsilateral to seizure onset inthree instances, but implicated an extra-tem-poral site in all cases.

Outcome of surgery in relation toelectrophysiologicalfindingsTwenty one patients have undergone surgery,and one to four years follow up have beencompleted in 15. Two were deemed inopera-ble despite localisation of ictal onset on

neuropsychological grounds: the contralat-eral hemisphere did not support memory on

the Wada test, precluding en bloc temporallobe resection, and the onset zone involved

lateral neocortex so that amygdalohippocam-pectomy was considered inappropriate. Onepatient has refused operation and a decisionhas been deferred in another, due to sponta-neous remission. One has asked to defer a

planned temporal resection. In patients withat least one year's follow up, the median out-come (following the classification of Engel'7)is 2b.The sequence of electrophysiological find-

ings and eventual outcome are summarised intable 7. Interictal epileptiform activity in thewake and sleep EEG was unifocal or showeda clearly lateralised preponderance in 25patients, but this agreed with SEEG laterali-sation, where achieved, in only 9/18. Ictalfindings on FO telemetry were concordantwith interictal scalp EEG in six instances;SEEG was confirmatory in four, but failed toidentify a single ictal onset zone in two. FOtelemetry led to a reversal of lateralisation innine patients. SEEG confirmed this revisedlateralisation in five, led to a further reversalin three, and was non-localising in one. In 10

patients with lateralising EEG findingstelemetry showed bilateral changes at seizureonset. An ictal onset zone was identified bySEEG in six of these. In the eight patientswithout lateralising interictal EEG discharges,telemetry showed a unilateral onset in three,confirmed by SEEG in only one, andreversed in two. In five patients neither inter-ictal EEG nor telemetry provided evidence oflateralisation, SEEG identified the ictal onsetzone in all of these, but in the three who haveso far completed one year's follow up afteroperation the results were poor.

DiscussionThe utility ofSEEGThe patients in our study were selected forSEEG specifically because the site of seizureonset was considered not to have been reli-ably determined by other means. Thus the

Table 7 Sequence of electrophysiological localisation and surgical outcome

Interictal Foramen ovale Surgicalscalp EEG telemetry SEEG outcome Pathology

6 4 confirmed la 4a EO MTS CD MTS?lateralityconfirmed 2 non- N N ? ?

localised5 confirmed la 3a lb MTS MTS NS

E N MTS ?9 3 reversed 2a 2b E MTS MTS NS

25 laterality againlateralised reversed

1 non-localised N ?

10 6 nowdiffuse localised* 2c 5 IA NS MTS MTSbilateral NNE ? ? CT

4non NNNN ????localised ___ __

3 1 confirmed 4b MTS8 not lateralisedlateralised 2 reversed lb E MTS MTS

5 not 5 now 3a 3a 3a NS MTS NSlateralised lateralised N E ? MTS

*Lateralisation in agreement with scalp EEG in 2.Surgical outcome:N: no operation to be performed (8 not localised, I refused, 1 remitted, 2 inoperable on neuropsychological grounds);0: awaiting operation; E: early postoperative, less than 1 year's follow up; 1-5 outcome grades following Engel (1987).Pathology:MTS: mesial temporal sclerosis; CD: cortical dysplasia; CT: low grade cerebral tumour; ? not available (not operated); NS: nosignificant specific pathology found.

62 on N

ovember 26, 2020 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.57.1.58 on 1 January 1994. D

ownloaded from

U in preoperative assessment of temporal lobe epilepsy

lack of agreement between localisation bySEEG and by less invasive methods is not areflection of the general unreliability of thelatter, but evidence rather of the relevance ofthe selection criteria employed. No singlenon-invasive method of assessment achievedcorrect lateralisation in more than one thirdof operable patients; various combinations ofassessments appeared no more reliable.The findings of SEEG are traditionally

taken to be the "gold standard" of preopera-tive localisation.21-2' However, locating ictalonset does not guarantee a successful out-come of technically adequate surgery.Determining the ictal onset zone and study-ing early discharge propagation does not pro-vide an objective means of identifying thehypothetical "epileptogenic zone", theremoval of which is both necessary and suffi-

24 e hrcient to abolish seizures. In some cases theremay exist no zone meeting this definition, thepatient being inherently inoperable, for suchreasons as diffuse cortical hyperexcitability.Development of criteria for identifying suchsubjects should be an urgent priority.Conversely, surgical success is not proof ofaccurate electrophysiological localisation. It isentirely possible that disconnection of seizurepropagation pathways during surgery basedon an incorrect identification of the onsetzone may sometimes give a good result.

In terms of its immediate, limited objec-tives, SEEG may be considered successful if asingle, circumscribed ictal onset zone isfound, and if surgical resection is followed byworthwhile seizure relief and/or results inremoval of tissue containing a discrete lesion.Seven SEEGs were clearly unsuccessful bythese standards, as an ictal onset zone wasnot located, but may be considered to havebeen of value in establishing inoperability andhence avoiding inappropriate surgery. In the18 patients so far operated 16 meet the abovecriteria of successful SEEG; two, however,have achieved only a grade 3a outcome afterremoval of essentially normal tissue. Thesetwo cases arguably represent failures of locali-sation by SEEG, but have no apparent distin-guishing clinical or electrographic features.The contribution of SEEG to preoperative

assessment in this study group may be sum-marised as follows:-1 Confirmatory of previous localisation-10(30%)2 Erroneous localisation corrected-5 (15%)3 Localisation achieved-I 1 (33%)4 Inappropriate operation avoided-7 (21 %)

Spencer25 reviewed published reports of178 patients who had depth recording andclassified the comparative EEG and SEEGfindings into five categories. Our four groupsapproximate to her categories 1, 2, 3, and 4combined with 5, which respectively com-prised 30%, 9%, 36% and 25% of the total.Spencer suggested that such results indicatedthat the number of operative candidates canbe increased by 36% (category 3) through theuse of SEEG. However, in their own materialSpencer et a126 failed to localise ictal onset in16 out of 31 patients (52%) (Spencer's

groups 4 and 5). Neither study analyses theproportions of patients undergoing invasiverecording and the implications of selectioncriteria. As we performed SEEG selectively ina small minority of the 269 patients formingthe parent population and 181 had surgerywithout SEEG, the percentage increase inoperability amounted to only 6%.Alternatively, the proportion of the 269 refer-rals in whom the procedure could be consid-ered to have improved preoperativeassessment (our groups 2, 3, and 4) makes up8.6%.

PARTICULAR CONSIDERATIONS IN

LOCALISATIONSubclinical seizuresSperling and O'Connor 27 reported subclini-cal electrographic seizures in 16 patients withcomplex partial seizures of temporal origin.The sites of onset of subclinical and overtseizures were coincident in 14, but as this wasnot the case in two, they concluded that sub-clinical seizures did not reliably indicate theorigin of the overt attacks. In our materialthere were also 16 patients with subclinicalseizures, but here there was complete agree-ment with the localisation of clinical episodes.Only three patients without subclinicalseizures have so far had surgery, with out-come grades of 3A, 3A, and 4B, whereas themedian outcome in the subclinical seizuregroup is 2b. There is thus a trend in accor-dance with a finding of Sperling andO'Connor,27 that patients without subclinicalseizures have less favourable results.

Anomalous site of onsetThree patients showed a reversal of lateralityof mesial temporal seizure onset after anti-epileptic drug (AED) withdrawal and when asecond seizure supervened within the periodof postictal electrographic change up to 30minutes following a previous attack.Subsequent seizures after a greater interictalinterval, both before and after restoration ofmedication, showed the habitual pattern. Thetypical seizures of all three patients rapidlypropagated to the contralateral hippocampus.It remains uncertain how far the inter-seizureinterval or the withdrawal of medication wasthe determining factor in the occurrence ofatypical seizures. Medication withdrawal isgenerally accepted as a satisfactory means ofpromoting seizure occurrence during teleme-try: apart from an increased tendency to sec-ondary generalisation the seizures follow theirhabitual pattern.28 29 Engel and Crandall30report a single patient with an anomalousseizure onset following AED withdrawal;arguably three further instances can beclaimed here. However, in our patients(unlike that of Engel and Crandall) a consis-tent relationship of atypical onset to a shortpreceding interictal interval provided groundsfor regarding these attacks as anomalous andit was decided to discount them for purposesof preoperative assessment. This decisionappeared justified by surgical outcome (grade1 in two patients, grade 2 in one).

Utility ofstereoelectroencephalography 63 on N

ovember 26, 2020 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.57.1.58 on 1 January 1994. D

ownloaded from

Binnie, Elwes, Polkey, Volans

Interictal SEEG dischargesInterictal discharges were present bilaterallyin all SEEG recordings, but when clearly pre-dominant on one side did not reliably later-alise the ictal onset zone, this was particularlyso at subdural contacts; lateralisation of theside of onset at subdural electrodes was alsoof no predictive value. These results agreewith the experience of some other investiga-tors31-33 but conflict with the findings ofLuders et aP4 who noted a high rate of con-cordance between ictal and interictal findingsin an, admittedly more extensive, array ofsubdural electrodes. Whether or not this is ofpractical consequence will depend on the cri-teria used to plan the resection. In centresperforming tailored procedures intended toinclude, so far as possible, the irritative zone,the delineation of this is obviously important.

Subdural placementsIn ictal recordings too, subdural placementsfailed to provide reliable lateralisation. Therewas initial subdural involvement in a focal orregional onset in five subjects and earlyappearance of discharges at ipsilateral sub-dural electrodes in 11; in the remaining six,propagation occurred first to the contralateralsurface contacts in most or all seizures. Thefailure rate of localisation by subdural elec-trodes is similar to that in comparative studiesof simultaneous depth and subdural record-ing,2735 but again in disagreement with theconclusions of Liiders et al,'34 using subduralelectrodes only. This result also conflicts withthe findings of Spencer et aP6 in eight patientswith depth and bilateral subdural electrodes;they did not observe propagation to contralat-eral neocortex before ipsilateral involvement.The high incidence of false lateralisation bysubdural electrodes is similar to the error ratewith multipolar FO contacts, which may beconsidered to provide a more extensive mesialtemporal subdural coverage than can beachieved either by the subdural reeds whichwe employed or by conventional strip elec-trodes inserted by a lateral approach. Thepurpose of FO recording was to reduce theneed for SEEG, and this is likely- to havefavoured the selection of patients withoutearly mesial temporal subdural discharges.

SELECTION FOR SEEGConcerning the criteria of selection forSEEG, three queries arise:1 Was SEEG performed unnecessarily inpatients in whom reliable localisation hadalready been achieved?2 Was SEEG performed to no useful pur-pose in patients in whom it could have beenpredicted that: a) no single ictal onset zonewould be found; b) surgical outcome wouldbe poor, or c) operation would be contraindi-cated on neuropsychological grounds?3 Was SEEG withheld in patients who wouldhave benefited from its use?

Addressing these issues1 Localisation without SEEGConcordance between any two non-invasive

assessments was found rarely (table 6) anddid not appear to offer reliable evidence oflateralisation. Again, this is unremarkable aslack of consistency between different assess-ments was a major factor in selection forSEEG. The highest specificity may beclaimed for the finding, uncommon in thisgroup, of agreement between routine EEGand telemetric lateralisation. The reasons forproceeding to SEEG in the four subjects withconcordant interictal and ictal EEG findingswere, however, not exceptional: delayedelectrographic ictal onset in three, and frontallocalisation of EEG discharges with clinicaldisparity in one.

2a Failure ofSEEG localisationIn seven patients (21%) no single ictal onsetzone was found, and the procedure conferredno benefit beyond avoidance of a possiblyinappropriate operation. This is a proportionsimilar to the 18% reported by Spencer.25The most conspicuous features of this groupwere that five of the seven had atrophiclesions on neuroimaging (mesial temporal intwo, thalamic in one, frontal in two), onlyone had a unilateral EEG focus, and nonehad lateralising findings on telemetry. In allbut the patient with a thalamic abnormality,the intracranial electrode array included con-tacts in the region indicated by imaging, butseizure onset was not located at this site norelsewhere. By contrast, of 26 patients with ademonstrable ictal onset zone, only one had aconcordant localised abnormality on imaging(unilateral dilatation of one ventricular horn,compatible with MTS), 18 had lateralisedEEG foci and 15 lateralised ictal onset ontelemetry (albeit incorrectly lateralised in nineand five instances, respectively). On suchsmall numbers it would be premature to drawany conclusion. However, these findings maysuggest that lack of lateralisation by bothinterictal EEG and ictal FO telemetry, in theface of clear radiological abnormalities maybe an adverse feature in selection for SEEG.

2b Poor surgical outcomeThe number of operated patients with at leastone year's follow up is too small to permitidentification of any group in whom anunfavourable result could have been pre-dicted and SEEG avoided. However, in thethree subjects without localisation by scalpEEG or foramen ovale telemetry, results werepoor and in two of the three no significantpathology was found. On these small num-bers there is no convincing evidence of anidentifiable group in whom SEEG was super-fluous, but in a larger prospective series itwould be worth testing the present trendswhich suggest that when EEG and telemetryprovide concordant localisation, SEEG isconfirmatory. When both are negative, SEEGtoo may fail to locate the ictal onset zone,even in the presence of abnormalities onimaging (see 2a above), and if the site ofseizure onset is apparently identified, opera-tive results may yet be poor.

64 on N

ovember 26, 2020 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.57.1.58 on 1 January 1994. D

ownloaded from

Utility ofstereoelectroencephalography in preoperative assessment of temporal lobe epilepsy

2c Neuropsychological contraindication to surgeryIn two patients no operation was performedbecause the Wada test suggested a high riskof postoperative memory deficit. However,this could not have been predicted beforeSEEG which lateralised the site of ictal onsetto a hemisphere required to support memory,and demonstrated a neocortical onset, pre-cluding the use of selective amygdalohip-pocampectomy to preserve memory.

Even with hindsight, there are no apparentcriteria by which patients who did notrequire, or could not benefit from, SEEGcould have been identified.

Obviously, this study cannot address thethird issue: the question of whether a greaterproportion of patients could have benefitedfrom SEEG. This will be difficult to resolveexcept possibly by comparing these findingswith a similar analysis of clinical materialfrom centres using less conservative selectioncriteria and therefore employing SEEG in alarger proportion of patients.

1 Engel J. Outcome with respect to epileptic seizures. In:Engel J, ed. Surgical treatmtent of the epilepsies. New York:Raven Press, 1987:553-71.

2 Rasmussen T. Cortical resection in the treatment of focalepilepsy. In: Purpura DP, Penry JK, Walter RD, eds.Neurosurgical management of the epilepsies. Adv Neurol,1975:8:139-54.

3 Davidson S, Falconer MA. Outcome of surgery in 40 chil-dren with temporal lobe epilepsy. Lancet 1975;1:1260-3.

4 Polkey CE. Prognostic factors in selecting patients withdrug-resistant epilepsy for temporal lobectomy. In: RoseFC, ed. Research progress in epiepsy. London: Pitman,1983:500-6.

5 Wright GDS, Oxbury JM, Fenwick P, Oxbury SM,Adams CBT. The possible need for intracranial EEG insurgery for temporal lobe epilepsy. Acta Neurol Scand1990;82:20.

6 Bloom D, Jasper H, Rasmussen T. Surgical therapy inpatients with temporal lobe seizures and bilateral EEGabnormality. Epilepsia 1959;1:351-65.

7 Falconer MA. Anterior temporal lobectomy for epilepsy.In: Logue V, ed. Operative surgery, vol 14. Neurosurgery.London: Butterworths, 1971:142-9.

8 Delgado-Escueta AV, Walsh GO. The selection processfor surgery of intractable complex partial seizures: sur-face EEG and depth electrography. In: Ward AA, PenryJK, Purpura D, eds. Epilepsy. New York: Raven Press,1983:295-300.

9 Morris HH, Kanner A, Luders H, Murphy D, DinnerDS, Wyllie E, Kotogal P. Can sharp waves localized atthe sphenoidal electrode accurately identify a mesio-temporal epileptogenic focus? Epilepsia 1989;30:532-9.

10 Engel J, Crandall PH, Brown WJ. Consistent false laterali-sation of seizure onset with sphenoidal and scalptelemetered ictal EEG recordings in two patients withpartial complex epilepsy. Electroencephalogr ClinNeurophysiol 1980;50:60.

11 Spencer SS, Williamson PD, Bridgers SL, Mattson RH,Ciccetti DV, Spencer DD. Reliability and accuracy oflocalization by scalp ictal EEG. Neurology 1985;35:1567-75.

12 Engel J, Sherman DA. Overview: Who should be consid-ered for surgical treatment? In: Engel J, ed. Surgicaltreatment of the epilepsies, 1993:23-34.

13 Powell GE, Polkey CE, Canavan AGM. Lateralisation ofmemory function in epileptic patients by use of the

sodium amytal (WADA) technique. J Neurol NeurosurgPsychiatry 1987;50:665-72.

14 Margerison JH, Binnie CD, McCaul IR. Electro-encephalographic signs employed in the location of rup-tured intracranial arterial aneurysms. ElectroencephalogrClin Neurophysiol 1970;28:296-306.

15 Van Veelen CWM, Debets RM, Van Huffelen AC, et al.Combined use of subdural and intracerebral electrodesin preoperative evaluation of epilepsy. Neurosurgery1990;26:93-101.

16 Agbi C, Polkey CE. Calculation of coordinates for depthelectrodes placed in temporal lobe structures visualizedby oblique CT scan cuts. Br J Neurosurg 1990;4:517-20.

17 Gotman J. Automatic recognition of epileptic seizures inthe EEG. Electroencephalogr Clin Neurophysiol 1982;54:530-40.

18 Yasargil MG, Teddy PG, Roth P. Selective amygdalo-hippocampectomy. Operative anatomy and surgicaltechnique. In: Symon L, ed. Advances and technicalstandards in neurosurgery vol, 12. Vienna: Springer-Verlag, 1985:93-123.

19 Spencer SS, Guimaraes P, Katz A, Kim J, Spencer D.Morphological patterns of seizures recorded intracra-nially. Epilepsia 1992;33:537-45.

20 Gloor P. Volume conductor principles: their applicationto the surface and depth electroencephalogram. In:Wieser H-G, Elger CE. eds. Presurgical evaluation ofepileptics. Berlin: Springer, 1987:59-70.

21 Gloor P. Contributions of electroencephalography andelectrocorticography to the neurosurgical treatment ofthe epilepsies. In: Purpura DP, Penry JK, Walter RD,eds. Neurosurgical management of the epilepsies. AdvNeurol 1975;8:59-106.

22 Rossi GF. Problems of analysis and interpretation of elec-trocerebral signals in human epilepsy: a neurosurgeon'sview. In: Brazier MAB, ed. Epilepsy: its phenomena inman. New York: Academic Press, 1973:261-87.

23 Walter RD. Principles of clinical investigation of surgicalcandidates. In: Purpura DP, Penry JK, Walter RD, eds.Neurosurgical management of the epilepsies. Adv Neurol,1975;8:49-58.

24 Liuders H, Awad, I. Conceptual considerations. In:Luders H. Epilepsy surgery. New York: Raven Press1991:57-62.

25 Spencer SS. Depth electroencephalography in selection ofrefractory epilepsy for surgery. Ann Neurol 1981;9:207-14.

26 Spencer SS, Spencer DD, Williamson PD, Mattson RH.The localizing value of depth electroencephalography in32 patients with refractory epilepsy. Ann Neurol1982;12:248-53.

27 Sperling MR, O'Connor MJ. Auras and subclinicalseizures: characteristics and prognostic significance. AnnNeurol 1990;28:320-8.

28 Spencer SS, Spencer DD, Williamson PD. Ictal effects ofanticonvulsant medication withdrawal in epilepticpatients. Epilepsia 1981;22:297-307.

29 Marciani MG, Gotman J, Andermann F. Patterns ofseizure activation-after withdrawal of antiepileptic med-ication. Neurology 1985;35:1537-43.

30 Engel J, Crandall P H. Falsely localizing ictal onsets withdepth EEG telemetry during anticonvulsant withdrawal.Epilepsia 1983;24:344-55.

31 Lieb JP, Joseph JP, Engel J, et al. Sleep state and seizurefoci related to depth spike activity in patients with tem-poral lobe epilepsy. Electroencephalogr Clin Neuraphysiol1980;49:538-57.

32 Lieb JP, Engel J, Gevins A, Crandall P. Surface and deepcorrelates of surgical outcome in temporal lobe epilepsy.Epilepsia 1981;22:515-38.

33 Engel J, Rausch R, Lieb JP, Kuhl DE, Crandall P.Correlation of criteria used for localizing epileptic foci inpatients considered for surgical therapy of epilepsy. AnnNeurol 1980;9:215-24.

34 Luders H, Hahn J, Lesser RP, Dinner DS, et al. Basaltemporal subdural electrodes in the evaluation ofpatients with intractable epilepsy. Epilepsia 1989;30:131-42.

35 Van Emde Boas W, Brekelmans GJF, Velis DN, VanHuffelen AC, Van Veelen CWM. Comparison of depthand subdural intracranial electrodes in recordingseizures of temporal and extratemporal lobe origin.Neurophysiol Clin 1992;22:45s.

36 Spencer SS, Williamson PD, Spencer DD, Mattson RH;Combined depth and subdural electrode investigation inuncontrolled epilepsy. Neurology 1990;40:74-9.

65

on Novem

ber 26, 2020 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.57.1.58 on 1 January 1994. Dow

nloaded from