Deep Brain Stimulation in Patients with Refractory Temporal Lobe Epilepsy

Epilepsia, 48(8):1551–1560, 2007Blackwell Publishing, Inc.C© 2007 International League Against Epilepsy

Deep Brain Stimulation in Patients withRefractory Temporal Lobe Epilepsy

∗Paul Boon, ∗Kristl Vonck, ∗Veerle De Herdt, ∗Annelies Van Dycke, ∗Maarten Goethals, ∗LutGoossens, ∗Michel Van Zandijcke, ∗Tim De Smedt, ∗Isabelle Dewaele, †Rik Achten, ‡Wytse

Wadman, §Frank Dewaele, §Jacques Caemaert, and §Dirk Van Roost

∗Reference Center for Refractory Epilepsy (RCRE) and Laboratory for Clinical and Experimental Neurophysiology (LCEN), and†Department of Neurology; Department of Radiology and Medical Imaging, Ghent University Hospital, Ghent, Belgium;

‡Swammerdam Institute for Life Sciences, Amsterdam University, Amsterdam, The Netherlands; and §Department of Neurosurgery,Ghent University Hospital, Ghent, Belgium

Summary: Purpose: This pilot study prospectively evaluatedthe efficacy of long-term deep brain stimulation (DBS) in medialtemporal lobe (MTL) structures in patients with MTL epilepsy.

Methods: Twelve consecutive patients with refractory MTLepilepsy were included in this study. The protocol included in-vasive video-EEG monitoring for ictal-onset localization andevaluation for subsequent stimulation of the ictal-onset zone.Side effects and changes in seizure frequency were carefullymonitored.

Results: Ten of 12 patients underwent long-term MTL DBS.Two of 12 patients underwent selective amygdalohippocampec-tomy. After mean follow-up of 31 months (range, 12–52 months),one of 10 stimulated patients are seizure free (>1 year), one of 10patients had a >90% reduction in seizure frequency; five of 10

patients had a seizure-frequency reduction of ≥50%; two of 10patients had a seizure-frequency reduction of 30–49%; and oneof 10 patients was a nonresponder. None of the patients reportedside effects. In one patient, MRI showed asymptomatic intracra-nial hemorrhages along the trajectory of the DBS electrodes.None of the patients showed changes in clinical neurologicaltesting. Patients who underwent selective amygdalohippocam-pectomy are seizure-free (>1 year), AEDs are unchanged, andno side effects have occurred.

Conclusions: This open pilot study demonstrates the poten-tial efficacy of long-term DBS in MTL structures that shouldnow be further confirmed by multicenter randomized controlledtrials. Key Words: Refractory epilepsy—Neurostimulation—Deep brain stimulation—Temporal lobe.

Epilepsy is the second most common chronic neuro-logic disease after cerebrovascular disorders, affecting0.5–1% of the population (Hauser et al., 1993). More than30% of all epilepsy patients have uncontrolled seizures orunacceptable medication-related side effects despite ade-quate pharmacologic treatment (Kwan and Brodie, 2000).Refractory epilepsy increases the risk of cognitive deterio-ration and psychosocial dysfunction and is associated withexcess injury and mortality (Brodie and Dichter, 1996).Therapeutic options that can be offered to patients withrefractory epilepsy include trials with newly developedantiepileptic drugs (AEDs), resulting in seizure freedomin ∼7% of these patients (Fisher, 1993). Epilepsy surgery

Accepted December 3, 2006.Address correspondence and reprint requests to Dr. P. Boon at Ref-

erence Center for Refractory Epilepsy (RCRE), Laboratory for Clinicaland Experimental Neurophysiology (LCEN), Department of Neurology,Ghent University Hospital, De Pintelaan 185, B-9000 Gent, Belgium.E-mail: [email protected]

doi: 10.1111/j.1528-1167.2007.01005.x

is another option that leads to long-term seizure freedomin an average of 58% of the patients, depending on thelocalization of the seizure focus (Engel et al., 2003; Leeet al., 2005). For the remainder of patients, few options areleft. Neurostimulation, defined as direct administration ofelectrical pulses to nervous tissue to modulate a patho-logic substrate and to achieve a therapeutic effect may besuch an alternative. Which part of the nervous system isbeing targeted and how the stimulation is being admin-istered may be variable. Vagus nerve stimulation (VNS)is an extracranial form of stimulation that was developedin the 1980s and is now routinely available in epilepsycenters worldwide (Ben Menachem, 2002). Deep brainstimulation (DBS) has previously been used for move-ment disorders and pain (Nguyen et al., 2000; Pollaket al., 2002; Volkmann et al., 2004). Moreover, severalnew indications such as obsessive–compulsive behaviorand different headache syndromes are being investigatedwith promising results (Nuttin et al., 1990; Leone et al.,2003, 2005). In the past, DBS of various targets such as

1551

1552 P. BOON ET AL.

the cerebellum, the locus coeruleus, and the thalamus wasperformed in patients with spasticity or psychiatric dis-orders who also had epilepsy, but the technique was notfully explored or developed into an efficacious treatmentoption for patients with epilepsy (Cooper, 1978; Wright etal., 1984; Upton et al., 1985; Feinstein et al., 1989). Thevast progress in biotechnology along with the experiencein other neurologic diseases in the past decade, has led to arenewed interest in DBS for epilepsy. A few epilepsy cen-ters worldwide have recently initiated trials with DBS indifferent intracerebral structures such as the thalamus, thesubthalamic nucleus, the caudate nucleus, and the cerebel-lum (Fisher et al., 1992; Velasco et al., 1995; Chkhenkeliand Chkhenkeli, 1992; Chabardes et al., 2002; Hodaie etal., 2002; Velasco et al., 2005). Two major stimulationstrategies can be pursued.

One approach is to target crucial central nervous sys-tem structures that are considered to have a “pacemaker,”“triggering,” or “gating” role in the epileptogenic net-work, such as the thalamus or the subthalamic nucleus(Iadarole and Gale, 1982). At Ghent University Hospital,the approach chosen was to evaluate potential interfer-ence with the ictal-onset zone. In medial temporal lobe(MTL) epilepsy, the epileptogenic region is believed to bein the medial temporal lobe, as documented by the gold-standard investigation using intracranial electrodes (Kingand Spencer, 1995). It is also supported by the high num-ber of seizure-free patients after resection of this region(Engel, 1993). On the one hand, investigating the potentialefficacy of DBS in patients with MTL epilepsy is inspiredby the search for less-invasive procedures compared withtissue resection. On the other hand, it fits in the search foralternative treatments for unsuitable candidates for resec-tive surgery, such as patients with bilateral MTL epilepsy.

Patients scheduled for invasive recordings because ofdiscrepant findings on noninvasive presurgical evaluationmust undergo an implantation procedure and were con-sidered to represent ideal candidates for the evaluation ofMTL DBS that can be performed by using the electrodesimplanted for diagnostic reasons.

Preliminary findings in three patients studied in ourgroup were reported previously (Vonck et al., 2002).

PATIENTS AND METHODS

Patient selectionThe study protocol and the informed-consent docu-

ments were approved by the Ethics Committee of GhentUniversity Hospital. Patients with refractory epilepsywere enrolled in a presurgical evaluation protocol atthe Reference Center for Refractory Epilepsy at GhentUniversity Hospital, a tertiary neurological referral cen-ter in Belgium. The presurgical protocol was publishedpreviously and includes video-EEG monitoring, opti-mum 1.5- or 3-T magnetic resonance imaging (MRI),

fluorodeoxyglucose-positron emission tomography, andcomprehensive neuropsychological assessment, accord-ing to international standards (EFNS task force, 2000).Twelve patients with medically refractory epilepsy wereincluded in the study (Fig. 1). Inclusion criteria consistedof (a) suspicion of temporal lobe epilepsy on the basis ofvideo-EEG monitoring; (b) seizure frequency of at leastone complex partial seizure (CPS) per month, confirmedduring a prospective preintervention baseline period of6 months; and (c) requirement for invasive video-EEGmonitoring in the bilateral MTL area and other subduralbrain areas because of incongruent findings during nonin-vasive presurgical evaluations to localize the seizure onset.During the preintervention baseline period, all patientswere receiving a stable combination therapy of two ormore AEDs.

Surgical procedureDuring an MRI-guided stereotactic procedure un-

der general anesthesia, two quadripolar DBS electrodes(model 3387; Medtronic, Fridley, MN, U.S.A.) were im-planted in each hemisphere through two parietooccipitalburrholes. Details of this procedure in our center werepublished previously (Vonck et al., 2002). The most an-terior electrode on each side was placed in the amyg-dala. The second electrode was placed in the anteriorpart of the hippocampus on each side. The trajectoryof the second electrode had a small angle with the firstone. Each electrode has four cylindrical electrode con-tacts of 1-mm diameter, 1.5-mm length, and an intercon-tact distance of 1.5 mm. On each electrode, the four elec-trode contacts cover a total length of 10.5 mm. In allpatients, additional subdural grids and/or strips in var-ious combinations were placed on the temporal and/orfrontal neocortex, depending on the results of the presur-gical evaluation during a procedure under general anes-thesia. Patients were allowed to recover in the neuro-surgical unit for a period of 48 h. Subsequently all in-tracranial electrode contacts and 27 scalp-EEG electrodeswere connected with the video-EEG monitoring system(128-channel digital video-EEG, Beehive; Astromed-Grass-Telefactor, West Warwick, RI, U.S.A.). The preciselocation of the intracranial electrode contacts was assessedby performing MRI by using an MPRAGE sequence.

Recording and stimulation paradigmAfter 48 h of video-EEG monitoring, during which

AEDs remained unchanged, AEDs were gradually tapereduntil habitual seizures were recorded (AED tapering con-dition). The finding of a unilateral or bilateral focal orregional MTL ictal onset was the criterion for offeringpatients the choice to undergo continuous MTL DBS. Pa-tients with unilateral MTL seizure onset were stimulatedby using the ipsilateral amygdalar and hippocampal DBSelectrodes. In patients with bilateral MTL onset, bilateralhippocampal stimulation was performed.

Epilepsia, Vol. 48, No. 8, 2007

NEUROSTIMULATION FOR EPILEPSY 1553

Initial DBS was performed by using a temporary exter-nal pulse generator (DualScreen 3628; Medtronic) duringa trial period (acute stimulation condition) before implant-ing patients with an internalized pulse generator. At anytime during the study, patients could make the choice ofinterrupting the ongoing stimulation treatment and under-going resective surgery, when indicated. Immediately be-fore the acute stimulation condition, subdural grids andstrips were removed. The aim was to keep patients onthe tapered AED regimen. In case of an acute increase inseizure frequency, reinstallation of AEDs at the baselinedosage and/or administration of escape medication wasplanned. To determine the output voltage level for initiat-ing DBS, we connected the hippocampal electrode to theEEG recording system, whereas the amygdalar electrodewas connected to an external generator. We then graduallyincreased output voltage until a stimulation artifact wasobserved on the hippocampal electrode. We performedthis action to confirm that stimulation was actually per-formed when the electrodes were connected to the gener-ator. When a stimulation artifact occurred on the adjacentelectrode, the output voltage was decreased with 0.1 to0.2 V, which eliminated the stimulation artifact and wasdetermined to be subthreshold. Once amygdalar stimu-lation output was determined, the hippocampal electrodecontacts were stimulated by using the same output. Thefrequency for both the amygdalar and hippocampal elec-trodes was set to 130 Hz, and pulse width, to 450 µs, basedon earlier experience with DBS in the medial temporallobe by Velasco et al. (2001). Pairs of adjacent electrodecontacts on both DBS electrodes were continuously stim-ulated in a bipolar way with the most anterior electrodecontact and the third electrode contact serving as cathodes.Individual pulses consisted of biphasic square-wave Lillypulses.

For patients with unilateral MTL seizure onset, this ac-tion was performed ipsilateral to the side of seizure onset.In patients with bilateral MTL seizure onset, this actionwas performed bilaterally only for the hippocampal DBSelectrodes.

To obtain an objective and comparable efficacy param-eter during the acute stimulation condition, interictal spikeactivity in the stimulated area was evaluated. The criterionfor implantation of a pulse generator (Kinetra, Medtronic)and entering the chronic stimulation condition was thefinding of a reduction of interictal spikes in the stimulatedarea of >50% during 7 consecutive days in the acute stim-ulation condition compared with the AED-tapering condi-tion. The rationale for comparing spike counts during theAED-tapering condition and the acute stimulation con-dition was that the initiation of stimulation was the onlyintervention that differentiated these two conditions fromeach other. DBS was interrupted every morning at aboutthe same time for a 1-h period during quiet wakefulness,typically from 10 am to 11 am. The DBS electrodes

were reconnected with the video-EEG monitoring equip-ment for recording of EEG activity in the stimulated areaduring the remaining 23 h of the day. Epileptiform dis-charges were visually identified and manually countedduring four consecutive intervals of 15 min/day. The pe-riod of spike counting was ≥6 hours remote from the oc-currence of seizures. To assess acute side effects, clinicalneurologic examination and bedside neuropsychologicaltesting (including reading, naming, and memory testing)were performed daily during the acute stimulation condi-tion. Formal neuropsychological assessment was plannedafter 12 months in all patients. A detailed account of neu-ropsychological outcome data is the object of a separatestudy.

When a >50% reduction of interictal spikes was notachieved within 21 days, the study protocol allowed aprolonged acute-stimulation condition. This consisted ofa trial with an adjusted stimulation frequency of 200 Hz.When, after a prolonged acute-stimulation phase of an-other 3 weeks, a >50% reduction of interictal spikes wasnot achieved during 7 consecutive days, patients were of-fered resective surgery when indicated or a continuationof best medical therapy.

When patients entered the prolonged-stimulation con-dition, the implantable pulse generator was placed inan abdominal subcutaneous pouch. The Kinetra devicewas connected to two DBS electrodes through two ex-tension wires. This required a short surgical proce-dure under general anesthesia. After this surgical pro-cedure, patients were followed up in the epilepsy clinicat regular 2-week intervals. Similar to the short-termstimulation condition, the aim was to keep patients onthe tapered-AED regimen throughout the prolonged-stimulation condition. In case of an acute increase inseizure frequency, reinstallation of AEDs at the baselinedosage and/or administration of escape medication wasallowed.

After 12 months of long-term DBS, the AED regi-men could be changed according to best medical practice.Gradual increase of stimulation output current in patientswho were not seizure free was allowed.

Data analysisThe present study is a comparative pre–post test,

prospective, open pilot trial of efficacy and safety of unilat-eral DBS in the MTL. During the entire study period fromthe prospective 6 month preintervention baseline periodover the prospective AED tapering condition and short-term stimulation condition to the prolonged-stimulationcondition, seizure frequency, adverse events, and con-comitant AED use were carefully monitored by using aseizure diary. Mean spike counts per hour during the short-term stimulation condition were compared with meanspike counts per hour during the entire AED-taperingcondition. Frequency of CPS ± secondary generalization

Epilepsia, Vol. 48, No. 8, 2007

1554 P. BOON ET AL.

(SG) during the prolonged-stimulation condition wascompared with the mean monthly frequency of suchseizures during 6 months before DBS. Postinterventionseizure outcome was assessed during the last 6 monthsof follow-up and categorized into the following groups:seizure-free; >90% seizure reduction; ≥50% seizure re-duction; 30–49% seizure reduction; and <30% seizurereduction (nonresponder).

RESULTS

All patients included in this study were candidates forinvasive EEG recording because of the absence of anystructural abnormalities and/or incongruent results in thenoninvasive presurgical evaluation. Table 1 shows the pa-tients’ individual neuroimaging results. In eight of 12 pa-tients, optimal MRI showed no structural abnormalities;in two patients, unilateral MTL signal abnormalities andsome degree of atrophy were found; in one patient, MTLatrophy was associated with anterior temporal and neocor-tical temporal gliosis; and in one patient, bilateral subcor-tical white matter lesions in the parietal lobe were found.

The stereotactic implantation procedure of DBS elec-trodes was uneventful in 11 of 12 patients. In pa-tient 11, asymptomatic hemorrhages occurred along thetrajectory of the depth electrodes. This bleeding wasdiscovered when a routine MRI was performed forpostoperative localization of the invasive electrodes.Postoperative MRI imaging for electrode localizationconfirmed localization of one DBS electrode in theamygdala and one DBS electrode in the anterior hip-pocampus in each hemisphere in all patients (Fig. 2).

TABLE 1. Results of neuroimaging, invasive video-EEG monitoring, and consequenttherapeutic intervention

Ictal onset Short-term Long-termPatient MRI (invasive video- stimulation stimulation Resectivenumber results EEG recording) phase phase surgery

1 Normal L focal T L AH L AH —2 Normal R regional T R AH R AH —3 Normal R regional T RAH R AH —4 Normal L regional T with

early right-sidedinvolvement

L AH L AH —

5 L Hippocampal andanterior andneocortical T gliosis

L regional T L AH L AH —

6 Normal L focal T L AH L AH —7 Normal L focal T L AH — L SAH8 L HS L focal T L AH L AH —9 Normal B T B H B H —

10 Normal R regional T R AH R AH —11 B P WML L focal T L AH L AH —12 R HS R focal T — — R SAH

MRI, magnetic resonance imaging; L, left; T, temporal; AH, amygdalohippocampal; R, right; SAH, selectiveamygdaohippocampectomy; TL, temporal lobectomy; HS, hippocampal sclerosis; B, bilateral; P, parietal; WML,white matter lesions.

Invasive video-EEG monitoring yielded long-term inter-ictal EEG recordings, habitual seizures, and ictal EEGrecordings in all patients within a time period of 5 to12 days.

Table 1 describes invasive video-EEG monitoring re-sults. In all patients, unilateral or bilateral interictal spikeswere recorded from the MTL electrodes. Focal EEG ic-tal onset, involving one or more electrode contacts on asingle DBS electrode, was identified in six of 12 patients,typically consisting of a low-voltage, high-frequency dis-charge in the hippocampal electrode contacts on one side,occurring seconds before clinical seizure onset and fol-lowed by spread to ipsilateral neocortical areas and thecontralateral MTL structures. In five of 12 patients, in-tracranial ictal EEG onset was “regional,” referring to amore widespread distribution of early changes involvingmore electrode contacts on different electrodes. The re-gional ictal onset was unilateral in all patients, but in pa-tient 4, spreading to the contralateral medial structuresoccurred within a few seconds. In one patient, bilateralMTL onset was found, with five seizures originating fromthe left temporal lobe and one from the right temporallobe.

Subsequently, 11 of 12 patients entered the acute-stimulation condition. One patient with unilateral rightfocal ictal onset chose to undergo resective surgery imme-diately. During the acute stimulation condition, continu-ous bipolar 130-Hz stimulation ipsilateral to the ictal onsetwas delivered to the amygdalar and hippocampal electrodecontacts in 10 of 11 patients. The patient with bilateral in-dependent seizure onset was stimulated bilaterally with thesame stimulation parameters in the hippocampus. Ten of

Epilepsia, Vol. 48, No. 8, 2007

NEUROSTIMULATION FOR EPILEPSY 1555

FIG. 1. Overview of different phases of the study and the inclusion criteria at different stages of the study.

11 patients showed a >50% reduction in interictal spikesin 1-h recording sessions on 7 consecutive days. In patient7, interictal spike frequency remained unchanged after ashort-term stimulation condition of 6 weeks and despite anincrease of stimulation frequency from 130 to 200 Hz af-ter 3 weeks. This patient underwent temporal lobectomy.Table 1 provides an overview of therapeutic interventionsin each patient.

The other 10 patients entered the long-term stimula-tion condition and had a generator implanted. At maxi-mal follow-up, mean stimulation output current was 2.3V (range, 2–3 V); stimulation frequency was 130 Hz inall but one patient, who was stimulated at 200 Hz. Pulsewidth remained unchanged to 450 µs in all patients. Meanfollow-up was 31 months (range, 12–52 months). Whenmean monthly seizure frequency during the last 6 months

TABLE 2. Follow-up duration, results of changes in seizure frequency, and side effects per patient

Patient Mean monthly Mean monthly Seizure-number seizure frequency seizure frequency frequency(type of FU before DBS or RS after DBS or RS reduction Sidetherapy) (mo) CPSs/GTCs/SPSs CPSs/GTCs/SPSs (%) effects

1 (DBS) 52 30/4/20 1 per yr/0/0.5 >90 None2 (DBS) 47 20/0/0 15/0/0 30–49 None3 (DBS) 44 4/0/0 4/0/0 <30 None4 (DBS) 40 8/2/30 4/1/15 ≥50 None5 (DBS) 37 12/0/0 5/0/0 ≥50 None6 (DBS) 33 30/0/0 5 (only nightly)/0/0 ≥50 None7 (RS) 26 4/0/0 0/0/0 100 None8 (DBS) 26 8/0/0 0/0/0 100 None9 (DBS) 20 10/0/0 7/0/0 30–49 None

10 (DBS) 19 2/0/0 0.5/0/0 ≥50 None11 (DBS) 15 2/2/0 0.5/0.5/0 ≥50 Asymptomatic hemorrhages along

trajectory of depth electrodes12 (RS) 12 6/0/0 0/0/0 100 None

FU, follow-up; DBS, deep-brain stimulation; RS, resective surgery; CPSs, complex partial seizures; GTCs,generalized tonic–clonic seizures; SPSs, simple partial seizures.

of follow-up was compared with a preintervention base-line period, one of 10 patients was seizure free (for ≥12months); one in 10 patients had >90% reduction of seizurefrequency; five of 10 patients had a reduction of seizurefrequency of ≥50%; two of 10 patients had a reduction ofseizure frequency of 30–49%; and one of 10 patients hada <30% seizure frequency reduction and was considereda nonresponder. Both patients who underwent resectivesurgery have been seizure free for ≥12 months. Table 2provides an overview of seizure-frequency reduction perpatient.

Changes in AEDs and increase in output current wereallowed after 12 months of DBS. In none of the patientsdid this further affect seizure frequency. Table 3 showsseizure frequency at maximal follow-up and a summaryof changes in AEDs and output current.

Epilepsia, Vol. 48, No. 8, 2007

1556 P. BOON ET AL.

TABLE 3. Overview of changes in AED treatment and changes in output current after 1 year of DBS

Output Outputcurrent current

Patient AED treatment AED during first 12 mo during atnumber before DBS or RS of DBS or RS first max(type of Changes after 12 mo 12 mo follow-therapy) Number AED Number AED of DBS or RS of DBS up

1 (DBS) 4 PHT, 300; CBZ, 1,000;LTG, 100; CZP, 3

2 PHT, 300; CZP, 3 +LEV, 4,000, no effect 1.5 3

2 (DBS) 3 VPA, 1,000; LEV,2,000; CLB, 10

2 VPA, 500; LEV, 2,000 None 1 2

3 (DBS) 3 PHT, 300; CBZ, 1,600;PRM, 375

2 PHT, 300; CBZ, 1,600 PB, 100, no effect 1 2

4 (DBS) 3 CBZ, 1,200; GBP, 900;PHT, 300

2 CBZ, 1,200; PHT, 300 +PGB, 450, no effect 1.5 3

5 (DBS) 3 CBZ, 1,000; LTG, 200;VGB, 2,000

1 LTG, 450 +LEV, 2,000, no effect 1 2

6 (DBS) 3 VPA, 1,600; GBP,2,800; TGB, 10

1 VPA, 1,000 +LEV, 2,000, + PB, 50,no effect

1.5 2

7 (RS) 2 TPM, 600; OXC, 1,200 2 TPM, 350; OXC,1,200

— — —

8 (DBS) 3 VPA, 1,500; LTG, 400;LEV, 3,000

2 LTG, 400; LEV, 3,000 — 2.5 3

9 (DBS) 3 CBZ, 800; VPA, 1,000;CZP, 1

1 CBZ, 800 +LTG, 150, no effect 1 2

10 (DBS) 2 LTG, 400; TPM, 400 1 LTG, 300 +LEV, 2,000 +CZP, 1,no effect

1.5 3

11 (DBS) 4 CBZ, 900; GBP, 1,200;PHT, 300; LEV,1,000

3 CBZ, 600; PHT, 450 CZP, 2, no effect 2 3

12 (RS) 2 CBZ, 800; CZP, 1.5 2 CBZ, 800; CZP, 1.5 — — —

AED, antiepileptic drug; DBS, deep-brain stimulation; RS, resective surgery; PHT, phenytoin; CBZ, carbamazepine; LTG, lamotrigine; CZP,clonazepam; LEV, levetiracetam; VPA, valproic acid; CLB, clobazam; PRM, primidone; PB, phenobarbital; GBP, gabapentin; PGB, pregabalin; VGB,vigabatrin; TGB, tiagabine; TPM, topiramate; OXC, oxcarbazepine.

During the short- and long-term stimulation condition,none of the patients reported side effects. Surgical im-plantation of the generator and perioperative course wereuneventful in all patients. None of the patients showedchanges in bedside neurologic and neuropsychologicaltesting.

DISCUSSION

The development of neurostimulation for neurologicalindications is driven by two major concerns related to stan-dard available treatments. First, a general tendency is tofind treatments that are minimally invasive and minimallyharmful to the patient. Second, the refractoriness of someneurologic diseases and the inability to treat them with thecurrently available means provides an impetus to searchfor novel treatments.

MTL epilepsy is the most prevalent type of refractorypartial epilepsy. In patients with MTL epilepsy, amyg-dalohippocampectomy or modified temporal lobectomyand hippocampectomy are standard procedures of choice,with postoperative long-term seizure-freedom rates of 60–75%. However, several postoperative neuropsychologicalstudies have shown verbal memory decline after resec-tion, mainly in left-sided operation patients, despite suffi-cient memory scores during the Wada test (Helmstaedter

et al., 2003; Gleissner et al., 2004). Studies have shownthat TLE patients without hippocampal sclerosis have agreater risk for postoperative neuropsychological deficit.The removal of nonatrophic hippocampus is associatedwith verbal memory decline in patients with left TLE. Inright temporal lobe patients, a decline in visual-spatiallearning was observed postoperatively (Trenerry et al.,1993; Stroup et al., 2003).

Patients with bilateral ictal onset are unsuitable candi-dates for epilepsy surgery. Patients showing widespreadictal onset have a less successful outcome after tempo-ral lobectomy. The recurrence rate of seizures after long-term seizure freedom after temporal lobectomy is ∼15%(Keleman et al., 2006). Hence, epilepsy surgery is not anoption for all patients with refractory epilepsy; it may notcure patients in the long term; it is an irreversible proce-dure, less successful for patients who have regional ictalonset or in patients with normal MRI; and a risk of post-operative neurologic and/or neuropsychological deficitsexists. This provides an impetus for further developing al-ternative treatment modalities such as neurostimulation.

Strategy for a medial temporal lobe DBS pilot trialThe choice of targeting the MTL region for a pilot trial

in humans at Ghent University Hospital was based on sev-eral considerations. The MTL is a region that often shows

Epilepsia, Vol. 48, No. 8, 2007

NEUROSTIMULATION FOR EPILEPSY 1557

FIG. 2. MRI image showing susceptibility artifacts of (A) left amyg-dalar electrode in a sagittal plane and (B) left amygdalar and hip-pocampal electrode and right amygdalar electrode in a coronalplane in patient 2.

specific initial EEG epileptiform discharges as a reflec-tion of seizure onset in human MTL epilepsy (Spenceret al., 1992). Basic research involving evoked potentialexcitability studies in humans and anatomic studies withtracer injections and single-unit recordings with histologicstudies in animals have also confirmed the involvement ofthe amygdala and the hippocampus in the epileptogenicnetwork (Wilson and Engel, 1993; Bragin et al., 2000;Kemppainen et al., 2002). Some studies have applied elec-trical fields to in vitro hippocampal slices with positive ef-fects on epileptic activity (Lian et al., 2003). Bragin et al.(2002) described repeated stimulation of the hippocampalperforant path in rats showing spontaneous seizures 4–8months after kainate injection in the hippocampus. Dur-ing perforant-path stimulation, spontaneous seizures weresignificantly reduced. In humans, preliminary studies onstimulation of hippocampal structures showed promisingresults on interictal epileptiform activity and seizure fre-quency (Velasco et al., 2000).

Very limited data are available on the results of stim-ulating the ictal-onset zone itself, which is believed toconsist of hyperexcitable cortex. To date, several studieshave shown that DBS in MTL epilepsy has positive effectson interictal epileptiform discharges and seizures (Velascoet al., 2000; Wiebe et al., 2006). Our previously publishedpreliminary study in three patients showed that MTL DBSin nonlesional patients leads to a significant decrease ofthe number of seizures and interictal epileptiform abnor-malities and is a safe treatment (Vonck et al., 2002).

The present study describes a larger patient series, alsoincluding patients with structural abnormalities, who wereselected for invasive video-EEG monitoring because ofincongruent findings in the course of noninvasive presur-gical evaluation.

Stimulation parametersThe stimulation parameters that were used in the present

study were based on the early experience reported by Ve-lasco et al. and by our group. The rationale for using high-frequency stimulation was that low-frequency stimulationinduces EEG synchronization, whereas high frequency isassociated with desynchronization, the latter condition be-ing more likely to have a therapeutic effect in epilepsy. Therationale for increasing the stimulation frequency from130 Hz to 200 Hz was that Osorio and colleagues sug-gested better seizure control in some patients when stimu-lation frequency was increased (Osorio et al., 2001). Care-fully increasing the output current was thought to affecta larger volume of epileptic tissue, which could be usefulin patients who were still having seizures after 1 year ofcontinuous DBS. Also because of safety concerns, out-put currents were kept below the values used in Parkin-son’s disease. Pulse width was set at 450 µs and keptunchanged throughout the study. A decrease to lower val-ues could be feasible and based on particular stimulationparadigms used by Osorio and colleagues and lower pulse-width values applied in Parkinson’s disease. In the presenttrial, continuous stimulation was used. Intermittent or on-demand stimulation may become an option when DBS canbe linked to seizure anticipation and detection algorithmsin a closed-loop system (Le Van Quyen et al., 2001). Itmust be stressed that our basic assumptions with regard tostimulation parameters remain speculative, as only limitedanimal or human data are available on the effect of differ-ent stimulation parameters applied to temporal lobe tissue.Systematic evaluation of different stimulation parametersin experimental animal studies is urgently needed. Whenmore information becomes available on effective stimu-lation parameters, the implanted pulse generator in ourpatients allows further adjustments in the future.

Efficacy of DBS in MTL regionThe efficacy results of the present study further cor-

roborate our previous findings that long-term DBS in theictal-onset zone, as defined by invasive EEG recording,

Epilepsia, Vol. 48, No. 8, 2007

1558 P. BOON ET AL.

significantly reduces seizures. Because of the open de-sign of our study, additional antiseizure effects of simplyimplanting invasive electrodes cannot completely be ruledout (Boon and Vandekerckhove, 1996; Katariwala et al.,2001). Some reports in the literature support the hypoth-esis that actual stimulation is not necessary to achieveefficacy and claim that efficacy is based on the lesionprovoked by the insertion of the electrode (Hodaie et al.,2002). This observation was described for electrode inser-tion in the anterior nucleus of the thalamus and referredto as “microthalamotomy.” Because in some epilepsycenters, an interval occurs between the invasive record-ings and consequent resective procedures or because afterthe results of invasive recordings, a resective procedurewas not feasible, seizure frequency after the removal ofthe implanted electrodes was evaluated without immedi-ate further interventions (Katariwala et al., 2001). Whenthe anterior nucleus is targeted, provoked lesions may bemore easily effective in such a confined area comparedwith targets such as the MTL region. It is not impossiblethat in an individual patient, essential pathways could beaccidentally or coincidently lesioned during MTL target-ing. However, this seems an unlikely hypothesis in a seriesof treated patients.

In a SPECT study by Velasco et al. (2001) in six patientsthat underwent 3 weeks of hippocampal stimulation withsubdural strips, the postimplantation SPECT (before stim-ulation) shows similar findings to the preimplantation im-ages, whereas the images after 3 weeks of DBS show clearhippocampal hypoperfusion comparable to images takenafter anterior temporal lobectomy. Implantation causing alesion might be expected to provoke earlier signs of hy-poperfusion in the lesioned region. Moreover, in contrastto the findings by Hodaie et al., seizures did not decreaseuntil several days after initiation of stimulation.

It may be that the lesioning hypothesis holds true forsome targets but is not applicable for others. Blindedrandomization of patients to “on” and “off” stimulationparadigms after implantation for substantial periods (e.g.,6 months) might clarify this and might also simultane-ously clarify the potential effect of sham stimulation ofan implanted device. That patients are unaware of stimu-lation in MTL structures represents a favorable factor fordesigning blinded controlled studies in the future.

Side effects of DBS in the MTL regionNumerous reports in the literature describe the ef-

fects induced by stimulation of central nervous systemstructures by using implanted electrodes for diagnos-tic and/or therapeutic purposes in animals as well as inhumans. From these studies, it has become clear thatstimulation of the hippocampus predominantly affectsmemory functions, whereas amygdalar stimulation pre-dominantly provokes emotional or affective–autonomicchanges (Bancaud et al., 1966; Shandurina and Kalyagina,

1979). In 1969, Stevens et al. described a variety of bothnegative and positive emotional changes after stimulationof the amygdala and, to a lesser degree, the hippocam-pus in three patients who were bilaterally implanted withamygdalar and hippocampal depth electrodes for a pe-riod of ≤7 months (Stevens et al., 1969). In our study,in none of the patients were stimulation-related objectiveor subjective side effects found. Besides MRI, our safetydata rely mainly on self-reported side effects and bedsideneurologic and neuropsychological testing. Formal neu-ropsychological testing to assess cognitive function afterlong-term stimulation has been performed, and results arethe topic of a separate study (Vonck et al., 2004). The lackof side effects in MTL DBS patients is most likely dueto the use of different stimulation parameters. In previousstudies, stimulation was often performed with the purposeof eliciting observable effects, performing functional map-ping of human brain, or eliciting afterdischarges or seizureonset in the course of a presurgical evaluation. In the lattercases, stimulation frequencies usually ranged between 1and 60 Hz, and pulse width was often 1 ms, with relativelyhigh output currents. In our study, relatively low outputcurrents with high frequency (130–200 Hz) and mediumpulse widths of 0.45 ms were used. In none of the patientswere increased seizure frequency or afterdischarges ob-served. This is in agreement with the preliminary findingsin the short-term study from Velasco et al. (2001). Uni-lateral MTL DBS using similar stimulation parametersduring 2–3 weeks in patients with MTL epilepsy did notproduce any subjective or objective behavioral responses.

One occurrence of intracranial bleeding in the area sur-rounding the trajectory of the depth electrode was reportedas a result of our systematic MRI screening of the correctpositioning of the intracranial electrodes. Large groups ofpatients have been investigated with regard to the safetyof conventional depth electrodes implanted for diagnosticpurposes (Espinosa et al., 1994; Fernandez et al., 1997).Reported morbidity rates are no higher than 4% and mostoften include surgical complications such as infection and,less frequently, neurologic complications secondary tosmall infarcts or bleeding. Permanent sequelae occur in<1% of cases. Especially because of their size, conven-tional electrodes are not suitable for long-term implanta-tion for the purpose of prolonged stimulation. From theextensive experience of DBS for movement disorders, itis known that specifically designed DBS electrodes aresuitable for long-term implantation and stimulation (Rise,2000).

Although postoperative MRI could be safely performedbefore the implantation of the generator, no MRIs wereperformed when the generator was implanted. The pres-ence of an implanted generator limits further diagnos-tic workup in these patients, which has to be weighedagainst the potential therapeutic benefit of such a device.Future safety studies should address optimal conditions

Epilepsia, Vol. 48, No. 8, 2007

NEUROSTIMULATION FOR EPILEPSY 1559

for generator placement and orientation and specificMRI modalities that can be safely applied in DBSpatients.

Mechanism of action of DBS for epilepsyAlthough the precise mechanism of action of DBS re-

mains to be elucidated, local inhibition induced by ap-plied current to a certain structure is a likely factor. Thisis the hypothesis of the so-called “reversible functionallesion” in which, in the case of targeting crucial structuresin a network, nuclei that are involved in propagation, sus-taining, or triggering of epileptic activity are inhibited.In the case of targeting an ictal focus, similar reasoningmay be applied, suggesting that applied current inhibitsoverexcitable tissue. Apart from this “local” inhibition,the mechanism of action of DBS may be based on theeffect on projections leaving from the area of stimula-tion to other central nervous structures. This may be themost likely hypothesis when crucial structures in epilep-togenic networks are involved. However, considering thatthe MTL structures are also potentially involved in thesenetworks, it may be that targeting the ictal focus may alsoaffect the epileptogenic network. When projections fromone structure to another are involved, this may be throughthe activation of inhibitory projections or through the in-hibition of (over)excitatory projections.

CONCLUSIONS

This prospective, open, long-term follow-up studydemonstrates efficacy of DBS in MTL structures in bothlesional and nonlesional patients. Currently for lesionalMTL epilepsy, resective surgery remains the preferredtreatment. Beyond the issue of weighing risks and benefitsfor treating nonlesional MTL epilepsy, DBS clearly mayhave a role in nonresectable cases such as patients with bi-lateral MTL epilepsy. Assessment of side effects suggeststhat long-term DBS in the MTL region is a safe treatment.Multicenter randomized trials with a larger number of pa-tients should now be conducted to confirm the results ofthe open studies. The final aim is to investigate whetherDBS can be an alternative treatment for MTL epilepsy thatis less invasive, reversible, and adjustable to the individualpatient.

Acknowledgment: Professor Boon is a Senior Clinical In-vestigator of the Fund for Scientific Research-Flanders and issupported by grants from the Fund for Scientific Research–Flanders (FWO); grants from Ghent University Research Fund,and by the Clinical Epilepsy Grant from Ghent University Hos-pital 2004-2008. Dr. K. Vonck, Dr. V. De Herdt, and Dr. A. VanDycke are supported by junior researcher (“Aspirant”) grantsfrom the Fund for Scientific Research–Flanders. T. De Smedtis supported by grants from the Institute for the Promotion ofInnovation by Science and Technology (IWT) in Flanders andGhent University Research Fund. The implanted devices usedin this study were provided by Medtronic Europe, Maastricht,The Netherlands.

REFERENCES

Bancaud J, Talairach J, Morel P, Bresson M. (1966) La corne d’Ammonet le noyau amygdalien: effets clinique et electrique de leur stimula-tion chez l’homme. Revue Neurologique 115:329–352.

Ben-Menachem E. (2002) Vagus-nerve stimulation for the treatment ofepilepsy. Lancet Neurology 8:477–482.

Boon P, Vandekerckhove T. (1996) Relief from seizures after intracra-nial EEG recording in non-operated patients with refractory partialepilepsy. Epilepsia 37:146.

Bragin A, Wilson CL, Engel J Jr. (2002) Increased afterdischarge thresh-old during kindling in epileptic rats. Experimental Brain Research144:30–37.

Bragin A, Wislon CL, Engel J Jr. (2000) Chronic epileptogenesis requiresdevelopment of a network of pathologically interconnected neuronclusters: a hypothesis. Epilepsia 1:144–152.

Brodie M, Dichter M. (1996) Antiepileptic drugs. New England Journalof Medicine 334:168–175.

Chabardes S, Kahane P, Minotti L, Koudsie A, Hirsch E, Benabid AL.(2002) Deep brain stimulation in epilepsy with particular referenceto the subthalamic nucleus. Epileptic Disorders 4(suppl 3):S83–S93.

Chkhenkeli SA, Chkhenkeli IS. (1992) Effects of therapeutic stimulationof nucleus caudatus on epileptic electrical activity of brain in patientswith intractable seizures. Epilepsia 36:63–71.

Cooper IS. (1978) Cerebellar stimulation in man. Raven Press, NewYork.

Engel J Jr, Wiebe S, French J, Sperling M, Williamson P, Spencer D,Gumnit R, Zahn C, Westbrook E, Enos B. (2003) Practice parame-ter: temporal lobe and localized neocortical resection for epilepsy.Epilepsia 44:741–751.

Espinosa J, Oliver A, Andermann F, Quesney F, Dubeau F, Savard G.(1994) Morbidity of chronic recording with intracranial depth elec-trodes in 170 patients. Stereotactical and Functional Neurosurgery63:63–65.

European Federation of Neurological Societies Task Force. (2000)Presurgical evaluation for epilepsy surgery: European standards. Eu-ropean Journal of Neurology 7:119–122.

Feinstein B, Gleason CA, Libet B. (1989) Stimulation of locus coeruleus:preliminary trials for spasticity and epilepsy. Stereotactical andFunctional Neurosurgery 43:906.

Fernandez G, Hufnagel A, Van Roost D, Helmstaedter C, Wolf HK,Zetner J, Schramm J, Elger C. (1997) Safety of intrahippocam-pal depth electrodes for presurgical evaluation of patients with in-tractable epilepsy. Epilepsia 38:922–929.

Fisher RS, Uematsu S, Krauss GL Cysyk BJ, McPherson R, LesserRP, Gordon B, Schwerdt P, Rise M. (1992) Placebo-controlled pi-lot study of centromedian thalamic stimulation in treatment of in-tractable seizures. Epilepsia 33:841–851.

Fisher RS. (1993) Emerging antiepileptic drugs. Neurology43(suppl):12–20.

Gleissner U, Helmstaedter C, Schramm J, Elger CE. (2004) Mem-ory outcome after selective amygdalohippocampectomy in patientswith temporal lobe epilepsy: one-year follow-up. Epilepsia 45:960–962.

Hauser WA, Annegers JF, Kurland LT. (1993) The incidence ofepilepsy and unprovoked seizures in Rochester, Minnesota, 1935-1984. Epilepsia 34:453–468.

Helmstaedter C, Kurthen M, Lux S, Reuber M, Elger CE. (2003)Chronic epilepsy and cognition: a longitudinal study in temporallobe epilepsy. Annals of Neurology 54:425–432.

Hodaie M, Wennberg R, Dostrovsky JO, Lozano A. (2002) Chronic an-terior nucleus stimulation for intractable epilepsy. Epilepsia 43:603–608.

Iadarole MJ, Gale K. (1982) Substantia nigra: site of anticonvulsantactivity mediated by gamma-aminobutyric acid. Science 218:1237–1240.

Katariwala NM, Bakay RA, Pennel PB, Olson LD, Henry TR, EpsteinCM. (2001) Remission of intractable partial epilepsy following im-plantation of electrodes. Neurology 57:1505–1507.

Kelemen A, Barsi P, Vajda J, Czirjak S, Borbely C, Rasonyi G, Halasz P.(2006) Long-term outcome after temporal lobe surgery: predictionof late worsening of seizure control. Seizure 15:49–55.

Epilepsia, Vol. 48, No. 8, 2007

1560 P. BOON ET AL.

Kemppainen S, Jolkkonene E, Pitkanen A. (2002) Projections from theposterior cortical nucleus of the amygdala to the hippocampal regionin rat. Hippocampus 12:735–755.

King D, Spencer SS. (1995) Invasive EEG in mesial temporal lobeepilepsy. Journal of Clinical Neurophysiology 12:32–45.

Kwan P, Brodie MJ. (2000) Early identification of refractory epilepsy.New England Journal of Medicine 342:314–319.

Le Van Quyen M, Martinerie J, Navarro V, Boon P, D’Have M, AdamC, Renault B, Varela F, Baulac M. (2001) Anticipation of epilepticseizures from standard EEG recordings. Lancet 20:183–188.

Lee SK, Lee SY, Kim KK. (2005) Surgical outcome and prognosticfactors of cryptogenic neocortical epilepsy. Annals of Neurology58:525–532.

Leone M, Franzini A, Broggi G, Bussone G. (2003) Hypothalamic deepbrain stimulation for intractable chronic cluster headache: a 3-yearfollow-up. Neurological Sciences 24(suppl 2):143–145.

Leone M, Franzini A, D’Andrea G, Broggi G, Casucci G, Bussone G.(2005) Deep brain stimulation to relieve drug-resistant SUNCT. An-nals of Neurology 57:924–927.

Lian J, Bikson M, Sciortino C, Stacy WC, Durand DM. (2003) Localsuppression of epileptiform activity by electrical stimulation in rathippocampus in vitro. Journal of Physiology 547:427–434.

Ngyuen JP, Lefaucher JP, Le Guerinel C, Eizenbaum JF, Nakano N,Carpentier A, Brugieres P, Pollin B, Rostaing S, Keravel Y. (2000)Motor cortex stimulation in treatment of central and neuropathicpain. Archives of Medical Research 31:263–265.

Nuttin B, Cosyns P, Demeulemeester H, Gybels J, Meyerson B. (1999)Electrical stimulation in anterior limbs of internal capsules in patientswith obsessive-compulsive disorder. Lancet 30:1526.

Osorio I, Frei MG, Manly BF, Sunderam S, Bhavaraju NC, WilkinsonSB. (2001) An introduction to contigent (closed-loop) brain electricalstimulation for seizure blockage, to ultra-short-term clinical trials,and to multidimensional statistical analysis for therapeutic efficacy.Journal of Clinical Neurophysiology 18:533–544.

Pollak P, Fraix V, Krack P, Moro E, Mendes A, Chabardes S, Kouds A,Benabid AL. (2002) Treatment results: Parkinson’s disease. Move-ment Disorders 17(suppl 3):S75–S83.

Rise MT. (2000) Instrumentation for neuromodulation. Archives of Med-ical Research 31:237–247.

Shandurina AN, Kalyagina GB. (1979) Dynamics of mental functionsin epileptics during electrical stimulation of deep brain structures.Fiziologiya Chelova 6:1020–1030.

Spencer SS, Guimaraes P, Katz A, Kim J, Spencer D. (1992) Morpholog-ical patterns of seizures recorded intracranially. Epilepsia 33:537–545.

Stevens JR, Vernon HM, Erwin F, Pacheco P, Suematsu K. (1969) Deep

temporal stimulation in man, long latency, long lasting psychologicalchanges. Archives of Neurology 21:157–169.

Stroup E, Langfitt J, Berg M, McDermott M, Pilcher W, Como P. (2003)Predicting verbal memory decline following anterior temporal lobec-tomy. Neurology 60:1266–1273.

Trenerry MR, Jack CR, Ivnik RJ, Sharbrough FW, Cascino GD,Hirschorn KA, Marsch WR, Kelly PJ, Meyer FB. (1993) MRI hip-pocampal volumes and memory function before and after temporallobectomy. Neurology 43:1800–1805.

Upton AR, Cooper IS, Springman M, Amin I. (1985) Suppression ofseizures and psychosis of limbic system origin by chronic stimu-lation of anterior nucleus of the thalamus. International Journal ofNeurology 19:223–230.

Velasco F, Velasco M, Velasco AL, Jimenez F, Marquez I, Rise M.(1995) Electrical stimulation of the centromedian thalamic nucleusin control of seizures: long-term studies. Epilepsia 36:63–71.

Velasco M, Velasco F, Velasco AL, Boleaga B, Jimenez F, Brito F, Mar-quez I. (2000) Subacute electrical stimulation of the hippocampusblocks intractable temporal lobe seizures and paroxysmal EEG ac-tivities. Epilepsia 41:158–151.

Velasco M, Velasco F, Velasco AL. (2001) Centromedian-thalamic andhippocampal electrical stimulation for the control of intractableepileptic seizures. Journal of Clinical Neurophysiology 18:495–513.

Velasco F, Carrillo-Ruiz JD, Brito F Velasco M, Velasco AL, MarquezI, Davis R. (2005) Double-blind, randomized controlled pilot studyof bilateral cerebellar stimulation for treatment of intractable motorseizures. Epilepsia 2005;46:1071–1081.

Volkmann J, Allert N, Voges J, Sturm V, Schnitzler A, Freund HJ. (2004)Long-term results of bilateral pallidal stimulation in Parkinson’s dis-ease. Annals of Neurology 55:871–875.

Vonck K, Boon P, Achten E, De Reuck J, Caemaert J. (2002) Long-term amygdalohippocampal stimulation for refractory temporal lobeepilepsy. Annals of Neurology 19:461–468.

Vonck K, Miatton M, Van Dycke A, Boon P. (2004) Neuropsycho-logical outcome of amygdalohippocampal stimulation for refractoryepilepsy. Epilepsia 45(suppl 7):150.

Tellez-Zenteno JF, McLachlan RS, Parrent A, Kubu CS, Wiebe S. (2006)Hippocampal electrical stimulation in mesial temporal lobe epilepsy.Neurology 23:1490–1494.

Wilson CL, Engel J Jr. (1993) Electrical stimulation of the human limbiccortex. In Devinsky O, Berci A, Dogali M (Eds) Electric and mag-netic stimulation of the brain and spinal cord. Raven Press, NewYork, pp. 103–113.

Wright GD, McLellan DL, Brice JG. (1984) A double-blind trial ofchronic cerebellar stimulation in twelve patients with severe epilepsy.Journal of Neurology Neurosurgery and Psychiatry 47:769–774.

Epilepsia, Vol. 48, No. 8, 2007