Deep brain stimulation of the centromedian thalamic nucleus for the treatment of generalized and frontal epilepsies

Authors


Summary

Purpose

Deep brain stimulation (DBS) of the thalamus is an emerging surgical option for people with medically refractory epilepsy that is not suitable for resective surgery, or in whom surgery has failed. Our main aim was to evaluate the efficacy of bilateral centromedian thalamic nucleus (CMN) DBS for seizure control in generalized epilepsy and frontal lobe epilepsy with a two-center, single-blind, controlled trial.

Methods

Participants were adults with refractory generalized or frontal lobe epilepsy. Seizure diaries were kept by patients/carers prospectively from enrollment. The baseline preimplantation period was followed by a control period consisting of a blind stimulation-OFF phase of at least 3 months, a 3-month blind stimulation-ON phase, and a 6-month unblinded stimulation-ON phase. The control period was followed by an unblinded long-term extension phase with stimulation-ON in those patients in whom stimulation was thought to be effective.

Key Findings

Eleven patients were recruited at King's College Hospital (London, United Kingdom United Kingdom) and at University Hospital La Princesa (Madrid, Spain). Among the five patients with frontal lobe epilepsy, only one patient had >50% improvement in seizure frequency during the blind period. In the long-term extension phase, two patients with frontal lobe epilepsy had >50% improvement in seizure frequency. All six patients with generalized epilepsy had >50% improvement in seizure frequency during the blind period. In the long-term extension phase, five of the six patients showed >50% improvement in the frequency of major seizures (one became seizure free, one had >99% improvement, and three had 60–95% reduction in seizure frequency). Among patients with generalized epilepsy, the DBS implantation itself appears to be effective, as two patients remained seizure free during 12 and 50 months with DBS OFF, and the remaining four had 50–91% improvement in the initial 3 months with DBS OFF.

Significance

DBS implantation and stimulation of the CMN appears to be a safe and efficacious treatment, particularly in patients with refractory generalized epilepsy. CMN stimulation was not as effective in frontal lobe epilepsy, which requires further studies. DBS of the CMN should be considered as a treatment option, particularly in patients with refractory generalized epilepsy syndromes.

Approximately 25% of patients with epilepsy are not controlled by antiepileptic medication (Hart & Shorvon, 1995; Lhatoo et al., 2003; Schuele & Luders, 2008; Kwan et al., 2011). Only around 20% of these patients have focal seizures not arising from eloquent cortex and can benefit from resective surgery. Furthermore, around 30–40% of the operated patients do not improve satisfactorily after resective surgical treatment (Tellez-Zenteno et al., 2005; de Tisi et al., 2011). Consequently, approximately 20% of patients with epilepsy do not respond to medical or surgical treatment, and consume substantial health resources, have significant social disadvantage, and are at risk of accidents during seizures and sudden death (Nashef & Ryvlin, 2009).

Electrical high frequency deep brain stimulation (DBS) of the thalamus is an emerging treatment for patients with medically refractory epilepsy who are not suitable for resective surgery. Several clinical studies have tested the efficacy of stimulation of different brain structures, including the centromedian thalamic nucleus (Fisher et al., 1992; Velasco et al., 2007b), the anterior thalamic nucleus (SANTE trial; Fisher et al., 2010), the subthalamic nucleus (Loddenkemper et al., 2001; Chabardes et al., 2002), medial temporal structures (Vonck et al., 2005; Tellez-Zenteno et al., 2006; Velasco et al., 2007a), and other cortical structures (Sun et al., 2008; Morrell, 2011).

Stimulation of the centromedian thalamic nucleus (CMN) has been used by Velasco's group for >20 years in >57 patients (Velasco et al., 1987, 1993, 1995, 2001, 2007b). Their studies suggest that patients with generalized epilepsy, in particular, appear to benefit from CMN stimulation, as demonstrated by reductions in the incidence of electroencephalography (EEG) spiking and in the frequency of generalized tonic–clonic seizures and atypical absences. More specifically, patients with temporal lobe epilepsy appear not to improve with CMN stimulation.

Whereas Velasco's studies benefit from a large number of patients with specific syndromes stimulated over many years at a single target, these studies were neither controlled nor blinded. A double-blind controlled crossover trial evaluated CMN stimulation on seven patients (Fisher et al., 1992). CMN stimulation was judged to induce little effect, with a seizure reduction of 30% during stimulation versus 8% when stimulation was discontinued. However, three patients that were recruited had temporal lobe epilepsy, which may explain the apparent lack of efficacy. Contradictory results have been reported by a few studies with small patient numbers (Chkhenkeli et al., 2004; Andrade et al., 2006; Cukiert et al., 2009).

In the present study, we carried out a controlled single-blind trial to revisit the efficacy of bilateral CMN stimulation on patients with generalized or frontal lobe epilepsies. We have evaluated the effect of CMN electrode implantation and stimulation on seizure frequency and on patient reported outcome measures (PROs) in order to assess their effects on the epilepsy and on broader aspects of the patients' lives.

Methods

Patients

Among a total of 11 patients (9 male, 2 female), 8 patients were recruited at King's College Hospital (London, United Kingdom) and 3 were recruited at University Hospital La Princesa (Madrid, Spain). Six patients had removal of the vagus nerve stimulation (VNS) device before DBS implantation. Age at surgery ranged from 18 to 51 years (mean 37). Average follow-up period was 2 years. Six patients had idiopathic or presumed symptomatic generalized epilepsy and five had frontal epilepsy. Seizure diaries were prospectively kept by patients/parents. Seven patients completed PROs, which included Quality of Life in Epilepsy-Patient-Weighted (QOLIE-31-P), Seizure Severity Scale and Hospital Anxiety Disorders before and after the procedures.

Before implantation, patients had been evaluated with clinical history, examination, neuroimaging, and scalp EEG-video telemetry recorded with electrodes applied according to the 10/20 or the Maudsley (Fernández Torre et al., 1999; Kissani et al., 2001) system. Patients were recruited according to the following inclusion criteria: (1) age >18 years; (2) clear diagnosis of epilepsy (confirmed by scalp or intracranial telemetry); (3) patients were unsuitable for resective surgery; (4) seizure frequency ≥10/month; (5) patients were able to give informed consent; (6) patients or carers were able to keep seizure diaries; and (7) patients were on stable dose of AEDs. Exclusion criteria were the following: (1) major neurologic or psychiatric disorders; (2) history of poor compliance with medication; (3) temporal lobe epilepsy; and (4) previous intracranial surgery. All patients and/or their carers were informed and signed written consent.

Recruited patients were categorized as having either (1) “generalized epilepsies,” including patients with either symptomatic or idiopathic generalized epilepsy; or (2) “presumed frontal epilepsy,” including patients with unilateral or bilateral frontal lobe epilepsy who were not suitable for resective surgery. Patients with presumed frontal lobe epilepsy had seizure semiology suggestive of frontal lobe seizures and had previously been assessed for resective surgery but no clear focus was identified on neuroimaging or on scalp EEG recordings. Noninvasive tests did not suggest a clear hypothesis for implantation of intracranial electrodes except in one patient, in whom intracranial video telemetry failed to demonstrate a clear seizure focus. To simplify the wording throughout the paper, the patients with presumed frontal lobe epilepsy will be simply referred to as “frontal lobe” patients.

The study has been approved by King's College Hospital Ethics Committee and the New Clinical Procedures Committee in 2005, and by the University Hospital La Princesa Ethics Committee in 2009.

Study design

The study consists of a two-center, single-blind, non-randomized, controlled trial. Essentially, in order to avoid the confounding effects of implantation versus stimulation, the patients were told that they were to be randomized, but they all had an OFF period first, which acts as a “washout period” for the effects of implantation. Seizure outcome in all patients was evaluated with daily seizure diaries filled in by the patients and/or carers. The primary outcome measure was the frequency of major seizures (the generalized tonic–clonic seizures, or the complex partial seizures with or without secondary generalization). The study was divided in five stages: baseline, electrode implantation, parameter optimization with system internalization, blind period, and open-label follow-up.

Baseline

All patients had a 3-month baseline period before implantation during which the following tests were obtained and assessed: medical history, general and neurologic examinations, electrocardiography (ECG), electroencephalography (EEG) telemetry, magnetic resonance imaging (MRI), neuropsychological scores, and PRO questionnaires. Patients were asked to keep a seizure diary 3 months prior to implantation and throughout the trial.

Electrode implantation

One four-contact electrode (K-3387 or K-3389, Medtronic) was implanted stereotactically in the CMN of each hemisphere under general anaesthesia. Target location for each electrode was determined using frame (Leksell Coordinate Frame G, Elekta, Stockholm, Sweden)-based atlas coordinates and calculated with STEALTH FRAMELINK 5 software (Medtronic, Inc., Minneapolis, MN, U.S.A.). Electrodes were implanted through bilateral frontal burr holes in a transparenchymal extraventricular trajectory under neurophysiologic monitoring (Fig. 1). The position of implanted electrodes was checked with intraoperative X-ray or computed tomography (CT) scan (Fig. 1).

Figure 1.

Electrode implantation: (A) longitudinal spin echo T1-weighted image showing the position of the right and left CMN DBS electrodes in patient 6. (B) coronal image in the same patient. (C) Neurophysiologic unit recordings during a typical track for DBS in the CMN for the same patient. Extracellular recordings at three different levels (dots) for the anterior (red) and central electrode (blue) 2 mm separated. Horizontal bar above the central-posterior recording shows the neural response to tactile stimulation (Vc nucleus) of the contralateral dorsal hand. The CMN response is characterized by sparse action potentials of low amplitude. (D) Lateral X-ray showing the CMN DBS electrodes and the two, four-contact subdural strips in patient 7. (E) Scalp and intracranial recordings in the same patient, with single-pulse electrical stimulation (SPES) through the medial contacts of the left CMN DBS electrode. Clear responses were elicited bilaterally, with maximal amplitude over Fz and the ipsilateral frontal electrodes. S, seconds; Vc, ventralis caudalis; RF, right frontal subdural electrode; LF, left frontal subdural electrode; CMN, centromedian thalamic nucleus; DBS, deep brain stimulation.

Intraoperative scalp EEG was recorded in all patients. In seven patients, additional intraoperative intracranial EEG was also recorded with lateral frontal four- or eight-contact subdural strip electrodes implanted bilaterally. In four patients, microelectrode recordings were performed at different distances from the MRI target (Fig. 1). Scalp EEG in all patients, with additional subdural frontal electrodes in seven, were used to record cortical responses induced by electrical stimulation at 0.2 and 6 Hz via the DBS electrodes at various distances from the target starting at 10 mm above target. In all cases, electrical stimulation confirmed that there were cortical connections between deep structures and cortex. The final electrode position was decided on the basis of atlas-derived coordinates, microelectrode intraoperative recordings, and the thalamic location where stimulation induced the largest cortical responses. General anesthesia was induced with propofol and maintained with isoflurane. In seven patients the subdural frontal strips were not removed after this procedure, and were used to record EEG responses during postoperative telemetry recorded during parameter optimization. The last four patients were not studied with subdural electrodes, since it was observed that EEG responses could be recorded on the scalp during parameter optimization (see section below).

Parameter optimization and system internalization

All 11 patients underwent 1-week scalp video telemetry after DBS electrode implantation in order to identify the optimal DBS stimulation parameters. In eight patients, DBS electrodes were externalized during 1 week and were used for recording and stimulation. Among these eight patients, seven had subdural electrodes implanted bilaterally over the frontal convexities. Recordings from subdural strips were not used to estimate medium/long-term seizure outcome. The optimal stimulation parameters were identified by the evaluation of seizure frequency and cortical responses to CMN stimulation, including recruitment responses at 6 Hz (Fig. 2) and responses to single-pulse electrical stimulation (Lacruz et al., 2007; Flanagan et al., 2009; Alarcon & Valentin, 2012; Alarcon et al., 2012; Kokkinos et al., 2013). After telemetry, the subdural electrodes were removed, the system was internalized (Kinetra, Medtronic, Inc., Minneapolis, MN, U.S.A.), and the stimulator turned OFF.

Figure 2.

(A) Lateral X-ray showing the CMN DBS electrodes in patient 4. (B) CT and longitudinal fast spin echo T1-weighted image fusion in the same patient (arrows points to thalamic contacts). (C) Scalp and thalamic recordings in the same patient during stimulation at 6 Hz through the deepest contacts of the left CMN DBS electrode. (D) Stimulation at 6 Hz through the deepest contacts of the right CMN DBS electrode. Both stimuli provoked clear responses over the cerebral cortex, where the fast Fourier transformation (FFT) from Cz showed a clear scalp 6 Hz recruiting rhythm. msec, milliseconds; μV, microvolt; Rt, right thalamic electrode; Lt, left thalamic electrode.

Blind period

The initial period after internalization was a single-blind period during which patients did not know whether the stimulator was ON or OFF. Patients were told that during this period that they would be randomized to 3 months of stimulation ON and 3 months of stimulation OFF, or vice versa. However, in all patients the first period consisted of stimulation OFF in order to washout the potential “micro-lesioning” effects of implantation and of the stimulation carried out during parameter optimization.

Stimulation-OFF phase (at least 3 months)

The initial stimulation-OFF period used for seizure frequency evaluation was the first 3 months after implantation in all patients. However, the stimulator was not turned on in the two patients who became seizure free after implantation/optimization, as it was considered unnecessary to turn ON the stimulator in seizure-free patients on clinical grounds. Patient 1 has remained seizure free and stimulation OFF to date. In patient 2, the stimulator was turned on 13 months after implantation, when seizures recurred.

Stimulation-ON phase (3 months)

In the nine patients with seizures during the stimulation-OFF phase, stimulation was turned on 3 months after implantation. In patient 2, who remained seizure free during the initial 3-month stimulation-OFF phase, stimulation was started 13 months after implantation. Patients remained blind for 3 months after starting stimulation. Patient 1 had remained seizure free and stimulation was never started (no stimulation-ON phase available).

The initial stimulation protocol of the study was continuous stimulation, frequency of 130 Hz with pulse width of 90 μs and up-to 5 V. Because the first two patients with this protocol did not show clear improvement in severity or frequency of seizures by comparing the stimulation-OFF and stimulation-ON phases, a new protocol was adopted: continuous stimulation, frequency at 60 Hz with pulse width of 90 μs, and up to 5 V.

Open-label follow-up

Unblind stimulation-ON period (6 months)

The single-blind period was followed by an open 6-month period with stimulation-ON.

Unblind long-term extension phase

The unblind stimulation-ON period was followed by an unlimited period with stimulation-ON in those patients who responded to stimulation.

Patient reported outcomes

Patient reported outcome (PRO) questionnaires (Quality of Life in Epilepsy-Patient-Weighted [QOLIE-31-P], Seizure Severity Scale, Hospital Anxiety Disorders) were completed before and 6 months after implantation in seven patients. Questionnaires were answered by patients and relatives within 30–45 min, allowing for the assessment of quality of life, anxiety and depression levels, number and severity of habitual seizures, and for the identification new seizure types that might have arisen in the course of the study.

Statistics

Fisher exact test was used to assess significant differences between frontal and generalized epilepsy groups with regard to the proportion of patients with good/poor outcomes (>50% vs. <50% change in seizure frequency compared to baseline) at 6 or 12 months after DBS implantation. Wilcoxon matched-pairs signed-rank test was used to compare PRO values before and after implantation.

Results

Patients

Table 1 shows the electroclinical characteristics of patients. The study included 11 patients, 2 female and 9 male, aged between 18 and 65 years. Four patients had idiopathic generalized epilepsy, two patients had presumed symptomatic generalized epilepsy, and five patients had frontal lobe epilepsy. All patients with idiopathic or presumed symptomatic generalized epilepsies had normal imaging. Among the frontal patients, two had normal imaging and two showed mild cortical atrophy. All patients showed generalized or bilateral interictal epileptiform discharges. Among the six patients with generalized epilepsy, all had convulsive generalized tonic–clonic seizures and five also had absence seizures. Among the five frontal patients, all had complex partial seizures (with or without secondary generalization) and one also had simple partial seizures. The frequency of major seizures (the convulsive generalized tonic–clonic seizures or the complex partial seizures) was the main variable used to evaluate treatment efficacy as shown in Figs 3 and 4. No localized seizure onset was demonstrated by ictal recordings. All six patients with generalized epilepsy were on polytherapy. Among the frontal patients, one was on monotherapy, one on two antiepileptic drugs, and three were on polytherapy. During the first 12 months after implantation, 10 patients had no medication changes, and zonisamide was stopped in the remaining patient after 3 months.

Table 1. Clinical characteristics
PatientAge (year)/ sexEpilepsy onset (year)IQAEDs/VNSEEG abnormalitiesMajor seizuresMinor seizuresMedication changes
  1. M, male; F, female; m, months; CPS, complex partial seizures; SPS, simple partial seizures; GTCS, general tonic–clonic seizures; GEDS, generalized epileptiform discharges; Oxc, oxcarbazepine; Lev, levetiracetam; CBZ, carbamazepine; Top, topiramate; PGB, pregabalin; Clon, clonazepam; Clob, clobazam; Prim, Primidone; Acet, acetazolamide; Lac, lacosamide; LTG, lamotrigine; Phe, phenytoin; VP, sodium valproate; Zon, zonisamide; VNS, vagus nerve stimulation; IQ, intelligence quotient; V, verbal; P, performance; F, full.

Idiopathic generalized epilepsy        
140/M3V = 80/P = 795 (VP, Lev, LTG, CBZ, Clob)GEDS, slow backgroundGTCSAbsencesReduction drugs 36 m/stop VP 60 m
245/M28V = 77/P = 744 (VP, LTG, Lev, Clon)GEDS, normal backgroundGTCSAbsencesReduction LTG from 36 m
326/M5V = 101/P = 1064 (VP, Lev, Phe, Zon)GEDS, frontal slowGTCSAbsencesStop Zon 3 m
447/M12V = 108/P = 1173 (Zon, Lev, Prim)GEDS, normal backgroundGTCSAbsencesNo change
Presumed symptomatic generalized epilepsy        
535/M7F < 604 (Lev, Oxc, Lac, Clob), VNSGEDS, slow backgroundGTCSAbsencesNo change
623/M10F < 604 (Lev, CBZ, LTG, Clob), VNSGEDS, slow backgroundGTCSNoneNo change
Frontal epilepsy        
746/M24V = 92/P = 1181 (CBZ)Frontal spikes (bilateral), normal backgroundCPSNoneNo change
851/M5V = 102/P = 925 (Lev, CBZ, Top, PGB, Clob), VNSFrontal spikes (bilateral), normal backgroundCPSNoneRetigabine started 30 m
937/M11F = 77 (V > P)2 (Prim, LTG), VNSGEDS, left frontal spikes, normal backgroundCPSNoneNo change
1018/F4V = 111/P = 854 (Lev, LTG, Acet, Clon), VNSGEDS, focal spikes, slow backgroundCPSSPSLacosamide started 12 m
1137/F6F < 603 (PGB, CBZ, Clon), VNSGEDS, multifocal spikes, slow backgroundCPSNoneNo change
Figure 3.

Mean change in the frequency of major seizures compared to mean seizure frequency during baseline. Mean change is shown during baseline (months −3 to 0) and during the first 12 months after implantation (months 0–12) in all patients [green], in patients with generalized epilepsy [red], and in patients with frontal lobe epilepsy [blue]). Two patients with generalized epilepsy became seizure free after implantation and consequently remained with the stimulator OFF for the first 12 months following implantation. In the remaining patients, the stimulator was turned ON 3 months after implantation.

Figure 4.

Mean change in the frequency of major seizures compared to mean seizure frequency during baseline. Mean change is shown during baseline and during the longest follow-up available for patients with generalized epilepsy (A) and for patients with frontal lobe epilepsy (B).

Safety and side effects

The surgical procedure and electrical stimulation were well tolerated by all patients. One patient had the device removed 6 months after implantation due to infection. No patient showed postsurgical hemorrhage or edema in the post-insertion MRI or CT. One patient reported a transitory agraphia during the first 4 days after implantation, which completely resolved. Most patients were initially aware of device activation, reporting a tingling sensation in the contralateral arm, which disappeared within several minutes. After losing the tingling sensations, the patients were reexamined to preserve the blinding, and the stimulator was maintained ON or switched OFF according to the study stage.

Confirmation of electrode placement with recruiting responses

MRI or CT was performed after DBS implantation to confirm the DBS electrode position. Cortical EEG responses associated with unilateral electrical stimulation of the CMN could be recorded on the scalp intraoperatively under general anaesthesia and during postimplantation telemetry in all patients (Figs 1 and 2). Responses were induced when stimulating CMN on either side in 10 patients (Fig. 2), and only when stimulating the left CMN in the remaining patient. Cortical responses consisted of recruitment responses which waxed and waned when stimulating at 6 Hz, and early responses when stimulating with single-pulse electrical stimulation (Lacruz et al., 2007; Valentin et al., 2005a,b, 2002; Flanagan et al., 2009). On the scalp, responses showed a widespread bilateral distribution with maximal amplitude over the midfrontal and ipsilateral frontal regions (Fig. 2). Responses to single-pulse electrical stimulation usually consisted of biphasic positive/negative potentials with peak latency between 16 and 60 msec (Fig. 1). No patient required repositioning of DBS electrodes.

Seizure outcome

Symptomatic or idiopathic generalized epilepsy

The follow-up period in six patients with symptomatic or idiopathic generalized epilepsy ranged between 12 and 66 months (Table 2). During the blind period, all patients had >50% improvement in seizure frequency. In the long-term extension phase, five of the six patients showed >50% improvement in the frequency of major seizure. Among these five patients, three became seizure free and two had 67–80% reduction in seizure frequency. All patients had ≥50% improvement in the frequency of minor seizures.

Table 2. Seizure frequency
PatientsBaselineStimulation OFF (blind period 3 m)Stimulation ON (blind period 3 m)Stimulation ON (unblind period 6 m)Last 6 months of long-term extension phaseComments
GTCS(sz/m)Abs (sz/day)GTCS % changeAbs % changeGTCS % changeAbs % changeGTCS % changeAbs % changeGTCS % changeAbs % changeFollow up (months)
Idiopathic generalized epilepsy            
10.31,000−100−100n/an/an/an/an/an/a72DBS OFF since implantation
23030−100−100−100−100−100−100−99−1006613 m between DBS OFF-ON
3812−50−50−50−50−50−50−25−5024Minor improvement
42100−85−100−70−100−75−100−87−100202 szs in the last 18 m/no minor sz
Presumed symptomatic generalized epilepsy            
5120300−91−97−91−97−87−93−93−9523Reported improvement in QoL
630np−60np−70np−57np−60np22Most szs in 2 days per month
PatientsBaselineStimulation OFF (blind period 3 m)Stimulation ON (blind period 3 m)Stimulation ON (unblind period 6 m)Last 6 months of long-term extension phaseComments
CPS (sz/m)SPS (sz/day)CPS % changeSPS % changeCPS % changeSPS % changeCPS % changeSPS % changeCPS % changeSPS % changeFollow up (months)
  1. GTCS, general tonic–clonic seizures; Abs, absences; CPS, complex partial seizures; SPS, simple partial seizures; n/a, not applicable; np, not present; m, month; sz, seizures; QoL, quality of life.

Frontal epilepsy
712np−75np−58npn/anpn/anp6Explanted (infection), worse after
810np−20np−40np−80np−80np48Improved with new DBS parameters
935np0np14np0np0np42Battery not replaced
10610000−33−95−33−95−50−9536Improved with new DBS parameters
1135np0np0np−5np−14np22No clear improvement

Of interest, two patients became seizure free immediately after implantation. Patient 1 has remained seizure free to date with stimulator OFF (follow-up of 60 months). Patient 2 remained seizure free for 12 months and then had five seizures in 1 month. Stimulation was then turned on (60 Hz) and the patient was almost seizure free for 45 months (only one seizure in this period). The stimulator then stopped working for technical reasons and the patient had three seizures in a month. After stimulation was resumed, he became seizure free to date (3 months).

Frontal epilepsy

The five patients with frontal lobe epilepsy had follow-up periods between 6 and 36 months (Table 2). During the blind period, only one patient had ≥50% improvement in seizure frequency. In the long-term extension phase, one patient had 90% improvement in seizure frequency, one patient had a 60% improvement, one had a 50% improvement, and two patients did not show any clear improvement in frequency or severity.

Comparison between frontal and generalized epilepsies

Patients in the generalized group were compared with the frontal group in the first 12 months after DBS implantation (Fig. 3).

Three months after implantation (DBS OFF), only one of the five patients with frontal epilepsy had an improvement ≥50% in seizure frequency, whereas all six patients with generalized epilepsy had an improvement ≥50 (Fisher exact test, p = 0.015). The average reduction in seizure frequency was 77% among patients with generalized epilepsy and 23% among those with frontal epilepsy.

Similar findings were seen 6 months after implantation. Only one of the five patients with frontal epilepsy had an improvement ≥50% in seizure frequency, whereas all six patients with generalized epilepsy had an improvement ≥50% (Fisher exact test, p = 0.015). The average reduction in seizure frequency was 77% among patients with generalized epilepsy and 18% among those with frontal epilepsy.

Twelve months after DBS implantation, none of the four patients with frontal epilepsy who completed this period had an improvement of ≥50% in seizure frequency, whereas five of the six patients with generalized epilepsy enjoyed an improvement ≥50% (Fisher exact test, p = 0.048). The average reduction in seizure frequency was 81% among patients with generalized epilepsy and 10% among those with frontal epilepsy.

The three patients with generalized epilepsy maintained the seizure frequency observed at 12 months (Fig. 4A). Among the three patients with frontal epilepsy, two patients improved their seizure frequency (Fig. 4B) after changing the stimulation parameters.

Patient-related outcome

Table 3 shows the PRO results from the seven patients who completed the questionnaires before and 6 months after DBS implantation. All patients reported an increment in quality of life that was statistically significant. Among the five patients who were not seizure free after implantation, there was a perception of increased seizure severity. The distress scale of the QOLIE-31-P and the Hospital Anxiety Disorders scale did not show significant differences in anxiety or depression when scales obtained before and after implantation were compared.

Table 3. Patient-related outcome (PROs)
 PatientsPre-DBS6 Months post-DBSStatistics
MedianRangeMedianRange
  1. Statistically significant (nonparametric related samples Wilcoxon signed rank test).

  2. QOLIE-31-P, quality of life in epilepsy version 31P; HADS, Hospital Anxiety and Depression Scale; SSQ, seizure severity questionnaire; pt, patients; sz, seizure.

QOLIE-31-P total753.924.8–68.568.843.9–75.9*p = 0.018
QOLIE-31-P distress753.67.1–727539.3–92.9p = 0.075
HADS7170–1870–17p = 0.24
SSQ (2 pt sz free)53.120.9–17.13.461.1–18.1*p = 0.043

Discussion

Patients with refractory epilepsy present a particular therapeutic challenge. Our results suggest that bilateral DBS of the CMN is associated with a reduction in seizure frequency in patients with generalized epilepsies who do not respond to medical treatment and are not suitable for resective surgery. This pronounced therapeutic effect among the patients with generalized epilepsy is on the order of 80% reduction in seizure frequency. These findings support those of Velasco's group (Velasco et al., 2007b). Of interest, two patients with generalized epilepsy became seizure free after implantation with the stimulator OFF for at least 12 months, suggesting that the implantation procedure per se can have a significant effect. This unexpected long-lasting implantation effect happening only in patients with generalized epilepsy is unlikely to be a placebo effect, and should be considered in future research on DBS of the CMN.

In frontal lobe epilepsy, the beneficial effects after implantation were milder (around 20% reduction in seizures), and seizure frequency gradually remitted to baseline levels, possibly suggesting a placebo or transient effect. However, two patients with frontal lobe epilepsy showed an improvement following modification of stimulation parameters after the initial 12 months, suggesting that frontal lobe epilepsy may require more patient-specific stimulation parameters for efficacy.

The mechanisms whereby CMN is particularly effective in generalized epilepsies may be clarified by animal experiments. In primates, CMN projects mainly to the striatum and diffusely to the cerebral cortex, particularly to motor and premotor cortices (Smith et al., 2004; Parent & Parent, 2005). Consequently, CMN stimulation would be expected to be most effective in generalized and frontal epilepsies, and less so in temporal lobe epilepsy as initially suggested by Velasco (Velasco et al., 2007b).

An unexpected result was that DBS implantation itself appears to contribute to seizure control, particularly in the six patients with generalized epilepsies. Among these, two patients became seizure free for months or years after implantation with stimulation OFF and in the remaining four, seizure control improved by >50% during the initial stimulation-OFF period after implantation. The decision not to turn the stimulator ON in seizure-free patients until seizures reappeared was difficult but relied on ethical and practical grounds. It would not be ethically justified to start stimulation to treat seizures in patients who are seizure free. In addition, in the absence of seizures during the stimulation-OFF period, it would have been impossible to demonstrate further treatment efficacy by stimulating. It is difficult to explain the cause of such long periods of seizure freedom following implantation. Depth electrode studies for presurgical assessment of epilepsy have been associated with long-lasting seizure freedom in a small proportion of patients (Spencer et al., 1993), presumably due to the small lesions generated. A few publications have reported a microlesionectomy effect with a reduction in seizures immediately after the implantation of DBS electrodes in thalamic structures for the treatment of epilepsy (Hodaie et al., 2002; Velasco et al., 2006; Cukiert et al., 2011). Similar microlesionectomy effects have also been described in surgical treatment of movement disorders (Kondziolka & Lee, 2004; Koop et al., 2006; Maltête et al., 2008; Mann et al., 2009; Morishita et al., 2010). Nevertheless, CMN lesion is clearly not the only mechanism. After the initial seizure freedom after implantation with stimulation OFF, seizures recurred in patient 2 approximately 12 months after implantation. Stimulation was then started and he recovered seizure freedom for 40 months. The stimulator then went OFF for unclear technical reasons and in a month he had three seizures, which stopped when the stimulator was switched ON again. The findings imply that CMN lesion and stimulation may have additive effects, and suggest that the effects of electrical stimulation may derive from inactivating the CMN.

Our study incorporates a number of modifications over previous reports on CMN stimulation that may have contributed to showing efficacy. Our study design was equipped with sufficient flexibility to avoid removing from the trial precisely those patients in whom the procedure was most effective, due to either implantation or stimulation. Following Velasco's findings (Velasco et al., 2007b), recruitment was restricted to patients with generalized or frontal epilepsies. Patients with temporal lobe epilepsy were excluded in contrast to the study by Fisher et al. (1992), where effects might have been diluted by the inclusion of temporal epilepsy. To accommodate the effects of the implantation procedure, our design is a single-blind study with no time limit between the stimulation-OFF and stimulation-ON phases (i.e., DBS was maintained OFF if patients remained seizure free during the first 3 months after implantation). We have used EEG responses to optimize implantation and the contacts chosen for stimulation. The initial parameters (130 Hz and 90 μs) were chosen following typical stimulation settings for DBS for epilepsy (Therodore & Fisher, 2004). After unclear effects with stimulation frequency at 130 Hz, we decided to follow Velasco's and Fisher's protocols at 60 Hz for the CMN (Fisher et al., 1992; Velasco et al., 2001), which indeed appeared to be more effective in our patients. Based on our experience in >80 children with movement disorders, we chose continuous stimulation. Finally, we have excluded patients with focal resections, hemispherectomy, or callosotomy in order to avoid potential confounding factors.

It is interesting to note that although all seven patients who completed the PRO questionnaires showed an improvement in quality of life, the five patients who were not seizure free reported a mild increase in the severity of seizures. Consequently, it appears that the reduction observed in seizure frequency is associated with an increment in quality of life, despite a mild increment in seizure severity. More importantly, the procedure was not associated with anxiety or depression.

Conclusion

DBS implantation and stimulation of CMN appears to be a safe and efficacious treatment, particularly in patients with generalized epilepsy for which medical treatment has failed. CMN stimulation was less effective in frontal lobe epilepsy, which requires further studies. DBS of CMN should be considered as a treatment option in patients with refractory generalized epilepsy syndromes.

Acknowledgments

This work has been partially funded by a grant from Fund for Epilepsy, and the Plan Nacional de Investigación Científica, Desarrollo e Innovación Tecnológica (I + D + I), Instituto de Salud Carlos III, PI12/02839.

Disclosure

We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines. A Valentin has received funding for travel, expert advice, and speaker honoraria from Medtronic. R Selway and R Chelvarajah have received speaker honoraria from Cyberonics. The remaining authors report no disclosures or potential conflict of interest.

Ancillary